<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: HMwork</title>
    <description>The latest articles on DEV Community by HMwork (@hmwork_2c698b7158d3).</description>
    <link>https://dev.to/hmwork_2c698b7158d3</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4102105%2F42afa6af-4e7c-4604-8f43-cc5d89f31c77.png</url>
      <title>DEV Community: HMwork</title>
      <link>https://dev.to/hmwork_2c698b7158d3</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/hmwork_2c698b7158d3"/>
    <language>en</language>
    <item>
      <title>Three Prices in One Second: Why the Chart Number Is Not the Number That Closes Your Position</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Fri, 20 Nov 2026 02:04:18 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/three-prices-in-one-second-why-the-chart-number-is-not-the-number-that-closes-your-position-36cf</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/three-prices-in-one-second-why-the-chart-number-is-not-the-number-that-closes-your-position-36cf</guid>
      <description>&lt;h1&gt;
  
  
  Three Prices in One Second: Why the Chart Number Is Not the Number That Closes Your Position
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;Disclosure: This article was produced with AI assistance under a human-supervised research process. All figures below come straight from public exchange API endpoints and are timestamped so you can re-run the measurements yourself. This is not financial advice.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;#ad — Affiliate disclosure: this publication may carry affiliate links in some articles; when present, they are marked. This particular article contains no affiliate or referral links at the time of publication.&lt;/strong&gt;&lt;/p&gt;




&lt;p&gt;Ask a trader what "the current price of Bitcoin" is, and they will point at the number on the chart — the price of the most recent trade. Ask an exchange's API the same question, and you often get three different numbers back in a single row. They sit side by side in the same response, at the same instant, and they are not equal.&lt;/p&gt;

&lt;p&gt;We measured this live on two independent exchanges on 2026-11-20, using nothing but free, key-less public endpoints.&lt;/p&gt;

&lt;h2&gt;
  
  
  The measurement
&lt;/h2&gt;

&lt;p&gt;At &lt;strong&gt;02:00:43.774 UTC&lt;/strong&gt;, Bybit's public v5 tickers endpoint (&lt;code&gt;/v5/market/tickers?category=linear&amp;amp;symbol=BTCUSDT&lt;/code&gt;) returned for the BTC perpetual:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;lastPrice&lt;/code&gt; — &lt;strong&gt;61,551.20&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;markPrice&lt;/code&gt; — &lt;strong&gt;61,560.90&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;indexPrice&lt;/code&gt; — &lt;strong&gt;61,591.22&lt;/strong&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Three numbers, one second apart on the same row. The chart price (&lt;code&gt;lastPrice&lt;/code&gt;) and the index price differ by &lt;strong&gt;40.02 USD&lt;/strong&gt;, roughly 6.5 basis points — a gap of the same order of magnitude as the maker/taker fee many traders spend hours comparing.&lt;/p&gt;

&lt;p&gt;Seconds later, at &lt;strong&gt;02:00:44–45 UTC&lt;/strong&gt;, OKX's public endpoints returned the same picture for &lt;code&gt;BTC-USDT-SWAP&lt;/code&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;markPx&lt;/code&gt; (from &lt;code&gt;/api/v5/public/mark-price&lt;/code&gt;) — &lt;strong&gt;61,560.9&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;last&lt;/code&gt; (from &lt;code&gt;/api/v5/market/ticker&lt;/code&gt;) — &lt;strong&gt;61,573.9&lt;/strong&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Here is the part that made us run the measurement twice: &lt;strong&gt;OKX's mark price (61,560.9) matched Bybit's mark price (61,560.90) to within a single tick, even though the two venues' last trade prices differed by 22.7 USD.&lt;/strong&gt; Each exchange's own tape tells a slightly different story about where Bitcoin "just traded." Their mark prices tell nearly the same one. That is consistent with what a mark price is supposed to be: a venue-computed reference anchored to a broad cross-market index, not a record of what happened to print last on that specific order book.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the documentation actually says
&lt;/h2&gt;

&lt;p&gt;Both venues list these fields in their public API documentation. Bybit's market tickers documentation (&lt;code&gt;bybit-exchange.github.io/docs/v5/market/tickers&lt;/code&gt;, fetched 2026-11-20, HTTP 200) enumerates &lt;code&gt;lastPrice&lt;/code&gt;, &lt;code&gt;indexPrice&lt;/code&gt; and &lt;code&gt;markPrice&lt;/code&gt; as distinct response fields. OKX's API v5 documentation (&lt;code&gt;okx.com/docs-v5&lt;/code&gt;, fetched 2026-11-20, HTTP 200) goes a step further and states, in the definition of an account-level PnL field, in plain words:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;"Calculated using mark price (not last trade price)"&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;Read that again. The exchange's own documentation says the profit-and-loss figure the platform computes for you is based on the mark price — explicitly &lt;em&gt;not&lt;/em&gt; the last traded price you see on the chart. A free public endpoint (&lt;code&gt;/api/v5/public/mark-price&lt;/code&gt;) exists solely to expose this number.&lt;/p&gt;

&lt;p&gt;Worth being honest about the limits: neither documentation page publishes the full internal formula for how each venue composes its index or adjusts its mark. We are not going to guess at formula internals. What is verifiable is narrower and still useful — (1) each venue returns multiple distinct price fields, (2) the venue's own docs say its calculations use the mark price rather than the last trade price, and (3) the mark prices of two independent venues sit close together while their trade tapes diverge.&lt;/p&gt;

&lt;h2&gt;
  
  
  The gap is not constant
&lt;/h2&gt;

&lt;p&gt;We re-ran the exact same calls about half a minute later (02:01:15–19 UTC). The market had dropped roughly 85 USD in the meantime, and the relationships had shifted:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Field&lt;/th&gt;
&lt;th&gt;First fetch&lt;/th&gt;
&lt;th&gt;Second fetch&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Bybit last&lt;/td&gt;
&lt;td&gt;61,551.20&lt;/td&gt;
&lt;td&gt;61,466.10&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Bybit mark&lt;/td&gt;
&lt;td&gt;61,560.90&lt;/td&gt;
&lt;td&gt;61,466.62&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Bybit index&lt;/td&gt;
&lt;td&gt;61,591.22&lt;/td&gt;
&lt;td&gt;61,494.43&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;OKX mark&lt;/td&gt;
&lt;td&gt;61,560.9&lt;/td&gt;
&lt;td&gt;61,460.9&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;OKX last&lt;/td&gt;
&lt;td&gt;61,573.9&lt;/td&gt;
&lt;td&gt;61,464.1&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;In the first snapshot, Bybit's mark sat 9.70 above its last price. In the second, it was 0.52 above. The index stayed about 28–30 USD above the mark both times (a basis effect specific to that moment in the perpetual market). The takeaway is uncomfortable but important: &lt;strong&gt;the distance between "the chart number" and "the settlement number" is not a fixed offset you can learn and forget.&lt;/strong&gt; It breathes with the market, second by second.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why this is the mechanism behind "my price never got there"
&lt;/h2&gt;

&lt;p&gt;A familiar complaint reads: "My stop or liquidation level was at X, the chart never touched X, yet my position closed." The three-price structure above is part of why that complaint can be literally true and still be explained: derivative venues publish their own rules for which price their engines watch, and the documentation quoted above shows at least one major venue explicitly calculating positions on the mark price rather than the chart price. A brief spike on one venue's own tape may move &lt;code&gt;last&lt;/code&gt; without moving &lt;code&gt;mark&lt;/code&gt;; a fast cross-market move may pull &lt;code&gt;mark&lt;/code&gt; through your level while your chart price lags behind it.&lt;/p&gt;

&lt;p&gt;We are not claiming any specific venue liquidates on any specific field beyond what its published documentation says. The honest, general advice: every venue's contract specifications and help pages define which price its funding calculations and forced-liquidation engine use. Read that definition once, before you ever need it — because the number you watched all day may not be the number that decided your outcome.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verify it yourself
&lt;/h2&gt;

&lt;p&gt;No account, no key, no SDK needed:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Bybit: &lt;code&gt;GET https://api.bybit.com/v5/market/tickers?category=linear&amp;amp;symbol=BTCUSDT&lt;/code&gt; — watch &lt;code&gt;lastPrice&lt;/code&gt;, &lt;code&gt;markPrice&lt;/code&gt;, &lt;code&gt;indexPrice&lt;/code&gt; in one row.&lt;/li&gt;
&lt;li&gt;OKX: &lt;code&gt;GET https://www.okx.com/api/v5/public/mark-price?instId=BTC-USDT-SWAP&lt;/code&gt; and &lt;code&gt;GET https://www.okx.com/api/v5/market/ticker?instId=BTC-USDT-SWAP&lt;/code&gt; — compare &lt;code&gt;markPx&lt;/code&gt; with &lt;code&gt;last&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Documentation: &lt;code&gt;https://bybit-exchange.github.io/docs/v5/market/tickers&lt;/code&gt; and &lt;code&gt;https://www.okx.com/docs-v5/&lt;/code&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Two small traps when you do: field names differ across venues (&lt;code&gt;lastPrice&lt;/code&gt; vs &lt;code&gt;last&lt;/code&gt;, &lt;code&gt;markPrice&lt;/code&gt; vs &lt;code&gt;markPx&lt;/code&gt;), and quote conventions differ — cross-venue number comparisons only mean something after you align symbol type, tick size and timestamp. A measurement is a snapshot; every figure in this article was captured inside a two-minute window on 2026-11-20 and the market has moved since.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;This post is for education about public market data only. It is not financial advice, not a trading recommendation, and not a judgment about which venue is better. Rates, prices and specifications change constantly — verify official sources before relying on any number.&lt;/strong&gt;&lt;/p&gt;




&lt;p&gt;&lt;em&gt;AI-assisted content notice: the text above was drafted with AI assistance from first-party API measurements taken by the publisher's automated research tooling, with the cited endpoints and documentation URLs listed for independent verification. Always verify official sources.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>api</category>
      <category>bitcoin</category>
      <category>crypto</category>
      <category>trading</category>
    </item>
    <item>
      <title>Same Order Book, Five Depth Caps — and One Venue That Quietly Answers 100</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Fri, 06 Nov 2026 02:15:40 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/same-order-book-five-depth-caps-and-one-venue-that-quietly-answers-100-20l7</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/same-order-book-five-depth-caps-and-one-venue-that-quietly-answers-100-20l7</guid>
      <description>&lt;h1&gt;
  
  
  Same Order Book, Five Depth Caps — and One Venue That Quietly Answers 100
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;How deep can you ask the L2 order book to go? We put the same question to five exchange APIs on 6 November 2026 and got five different answers — and one venue that ignores an out-of-range parameter without a word of warning.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Disclosure:&lt;/strong&gt; Produced with AI assistance from data pulled directly from official exchange APIs and documentation. No sponsored placements at the time of writing; if that changes, an #ad notice will appear where any affiliate link is used. Nothing here is financial advice — a technical write-up about API behavior.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why depth caps matter
&lt;/h2&gt;

&lt;p&gt;If you build anything that touches market data, you will call an order book endpoint. Almost every venue exposes one with a size parameter, and most developers assume it works the same way everywhere: ask for N levels, get N levels, get an error if N is too big.&lt;/p&gt;

&lt;p&gt;That assumption is wrong in an interesting way. We sent the same request — "give me the BTC order book, N levels deep" — to five venues and recorded both the documented rules and the actual responses. The caps differ by more than ten times, the defaults by one hundred times, and overshoot behavior splits into four failure modes.&lt;/p&gt;

&lt;h2&gt;
  
  
  The measurements (6 Nov 2026, 02:03–02:06 UTC)
&lt;/h2&gt;

&lt;p&gt;All requests hit the public, no-authentication market data endpoints directly. Counts below are array lengths in the JSON responses.&lt;/p&gt;

&lt;h3&gt;
  
  
  Binance — &lt;code&gt;GET /api/v3/depth&lt;/code&gt;
&lt;/h3&gt;

&lt;p&gt;Binance's official REST spec (&lt;code&gt;rest-api.md&lt;/code&gt; in the exchange's own GitHub docs repository) is explicit: &lt;code&gt;Default: 100; Maximum: 5000. If limit &amp;gt; 5000, only 5000 entries will be returned.&lt;/code&gt;&lt;/p&gt;

&lt;p&gt;Live check on &lt;code&gt;BTCUSDT&lt;/code&gt;: &lt;code&gt;limit=9999&lt;/code&gt; returned exactly 5,000 bids and 5,000 asks — the honest truncation the docs promise. Omitting the parameter returned 100 per side, matching the documented default. The response never echoes your parameter: to know what you received, count the array yourself.&lt;/p&gt;

&lt;h3&gt;
  
  
  Bybit — &lt;code&gt;GET /v5/market/orderbook&lt;/code&gt;
&lt;/h3&gt;

&lt;p&gt;Bybit's docs give a range per product line: spot &lt;code&gt;limit&lt;/code&gt; is &lt;code&gt;[1, 1000]&lt;/code&gt; with &lt;code&gt;Default: 1&lt;/code&gt;; linear and inverse contracts share that range but default to 25; options is &lt;code&gt;[1, 25]&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;Live check on spot &lt;code&gt;BTCUSDT&lt;/code&gt;: &lt;code&gt;limit=600&lt;/code&gt; returned 600 levels per side; &lt;code&gt;limit=2000&lt;/code&gt; was trimmed to 1,000 without an error; and — the sneaky one — &lt;strong&gt;omitting the parameter returned a single level per side&lt;/strong&gt;. A one-level book is a legitimate response, not an error: a script that forgot the parameter would compute depth from one bid and one ask without anything failing loudly.&lt;/p&gt;

&lt;h3&gt;
  
  
  OKX — &lt;code&gt;GET /api/v5/market/books&lt;/code&gt;
&lt;/h3&gt;

&lt;p&gt;OKX documents the size parameter &lt;code&gt;sz&lt;/code&gt; with &lt;code&gt;Maximum 400, e.g. 400 bids + 400 asks&lt;/code&gt;, and the default is 1 level.&lt;/p&gt;

&lt;p&gt;Live check on &lt;code&gt;BTC-USDT&lt;/code&gt;: &lt;code&gt;sz=400&lt;/code&gt; returned 400 per side; omitting &lt;code&gt;sz&lt;/code&gt; returned 1 per side, both as documented. Overshoot is the interesting part: &lt;code&gt;sz=401&lt;/code&gt; and &lt;code&gt;sz=500&lt;/code&gt; both came back with &lt;code&gt;code "51000"&lt;/code&gt; / &lt;code&gt;msg "Parameter sz error."&lt;/code&gt; — the one venue here that rejects an out-of-range parameter explicitly inside the returned body. HTTP status is still 200: the error lives inside the JSON, so a client that only checks status codes misses it.&lt;/p&gt;

&lt;h3&gt;
  
  
  Kraken — the double surprise
&lt;/h3&gt;

&lt;p&gt;&lt;strong&gt;Surprise one: the endpoint moved.&lt;/strong&gt; When we ran the first pass of this research in early September 2026, the method was &lt;code&gt;GET /0/public/GetOrderBook&lt;/code&gt;. On 6 November that call returns HTTP 404 with &lt;code&gt;{"error":["EGeneral:Unknown method"]}&lt;/code&gt;. The book now lives at &lt;code&gt;GET /0/public/Depth&lt;/code&gt; — confirmed both by the live response (full &lt;code&gt;asks&lt;/code&gt; and &lt;code&gt;bids&lt;/code&gt; arrays, now with a timestamp on every level) and by Kraken's current OpenAPI specification, where the &lt;code&gt;getOrderBook&lt;/code&gt; operation is defined as &lt;code&gt;get /public/Depth&lt;/code&gt;. Old code breaks loudly, which is the good kind of breakage.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Surprise two: the silent fallback.&lt;/strong&gt; The OpenAPI schema for &lt;code&gt;Depth&lt;/code&gt; defines &lt;code&gt;count&lt;/code&gt; as an integer with &lt;code&gt;minimum: 1&lt;/code&gt;, &lt;code&gt;maximum: 500&lt;/code&gt;, &lt;code&gt;default: 100&lt;/code&gt;. Live behavior on &lt;code&gt;XBTUSD&lt;/code&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;count=500&lt;/code&gt; → 500 bids, 500 asks (as documented)&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;count=501&lt;/code&gt; → &lt;strong&gt;100 bids, 100 asks&lt;/strong&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;count=1000&lt;/code&gt; → 100 bids, 100 asks&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;count=abc&lt;/code&gt; (not even a number) → 100 bids, 100 asks&lt;/li&gt;
&lt;li&gt;parameter omitted → 100 bids, 100 asks&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;An out-of-range value — even a non-numeric one — is neither clamped to 500 nor rejected: it falls back to the default of 100 with an empty error array. Your tool asked for 500 levels, received 20 percent of that, and nothing in the response says so. Of the four failure modes observed, this is the one most likely to poison a dataset for weeks before anyone notices.&lt;/p&gt;

&lt;h3&gt;
  
  
  Coinbase Exchange — &lt;code&gt;GET /products/BTC-USD/book?level=2&lt;/code&gt;
&lt;/h3&gt;

&lt;p&gt;The Coinbase Exchange full level-2 request at 02:03:19 UTC returned 21,951 asks and 17,395 bids in one response — a complete book, not a capped slice. No published depth cap exists for it, so we report this as an observation, not a comparison point: undocumented behavior deserves the same skepticism as documented behavior.&lt;/p&gt;

&lt;h2&gt;
  
  
  Four ways to fail the same request
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Honest truncation&lt;/strong&gt; (Binance): you overshoot, you get the cap, and the docs told you in advance.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Silent trim to cap&lt;/strong&gt; (Bybit): you overshoot, you get the cap, no complaint in the returned data.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Explicit rejection&lt;/strong&gt; (OKX): you overshoot, you get a named error code — but inside a 200 response.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Silent fallback to default&lt;/strong&gt; (Kraken): you overshoot, you get far less than the cap, looking completely normal.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;A fifth lesson comes from the renamed method: endpoints get renamed and removed without your code changing a character. A script written in September 2026 was broken by November — loudly here, but only because the failure was a named unknown-method error rather than a generic timeout.&lt;/p&gt;

&lt;h2&gt;
  
  
  What to do with this
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Count the arrays you receive.&lt;/strong&gt; Never assume length matches request: none of the venues above, even the well-behaved ones, echoes your parameter.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Treat the default as a hostile value.&lt;/strong&gt; One venue defaults to a single level. Assert that received depth meets what your logic requires, and alert when it does not.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Parse the body, not just the status code.&lt;/strong&gt; The only venue that rejects an out-of-range size does so with HTTP 200 and an error field in JSON.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Re-read the specs on a schedule.&lt;/strong&gt; Caps, defaults and method names are documentation, and documentation changes. The venue frozen in your notes from last quarter may have moved its door.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;None of this makes one venue better than another for trading. It does mean "the API" is not one API, and the cheapest way to corrupt a market-data pipeline is to assume consistency that was never promised.&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;Verify official sources.&lt;/strong&gt; Every number above was captured live from the venues' own public endpoints on 6 November 2026 (02:03–02:06 UTC) and can be reproduced: Binance &lt;code&gt;https://api.binance.com/api/v3/depth?symbol=BTCUSDT&lt;/code&gt; with rules in &lt;code&gt;https://raw.githubusercontent.com/binance/binance-spot-api-docs/master/rest-api.md&lt;/code&gt;; Bybit &lt;code&gt;https://api.bybit.com/v5/market/orderbook?category=spot&amp;amp;symbol=BTCUSDT&lt;/code&gt; with docs at &lt;code&gt;https://bybit-exchange.github.io/docs/v5/market/orderbook&lt;/code&gt;; OKX &lt;code&gt;https://www.okx.com/api/v5/market/books?instId=BTC-USDT&lt;/code&gt; with docs at &lt;code&gt;https://www.okx.com/docs-v5/en/&lt;/code&gt;; Kraken &lt;code&gt;https://api.kraken.com/0/public/Depth?pair=XBTUSD&lt;/code&gt; with the schema at &lt;code&gt;https://docs.kraken.com/api-reference/market-data/get-order-book&lt;/code&gt;; Coinbase Exchange &lt;code&gt;https://api.exchange.coinbase.com/products/BTC-USD/book?level=2&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Transparency notes: written with AI assistance, using only the official sources listed above. Affiliate disclosure (#ad): no affiliate or sponsored links appear in this article. This documents public API behavior only — not investment, trading or platform advice. Exchange specs change; confirm current rules on the official pages before relying on anything here.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>api</category>
      <category>crypto</category>
      <category>bitcoin</category>
      <category>beginners</category>
    </item>
    <item>
      <title>One JSON-RPC Call, Nine Endpoints, Four Chains: What Public Gas Prices Actually Say Right Now</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Tue, 03 Nov 2026 02:10:35 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/one-json-rpc-call-nine-endpoints-four-chains-what-public-gas-prices-actually-say-right-now-4ae2</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/one-json-rpc-call-nine-endpoints-four-chains-what-public-gas-prices-actually-say-right-now-4ae2</guid>
      <description>&lt;h1&gt;
  
  
  One JSON-RPC Call, Nine Endpoints, Four Chains: What Public Gas Prices Actually Say Right Now
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;Disclosure: this article was written with AI assistance from measurements taken by our own tooling, then reviewed for accuracy. Nothing here is financial advice — this is a measurement log, not a recommendation to use, avoid, or move funds to any chain.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;#ad — affiliate note: this article currently contains no affiliate or sponsored links. If that ever changes, affiliate links will be marked here and labeled clearly.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Every comparison of "which chain is cheaper" eventually rests on one number: the gas price. And every one of those numbers is quoted from &lt;em&gt;someone else's&lt;/em&gt; readout. We wanted to know what the raw, first-hand data looks like when you skip the dashboards and ask the chains directly — so we made one standard JSON-RPC call, &lt;code&gt;eth_gasPrice&lt;/code&gt;, to four EVM chains through nine public endpoints run by independent operators, twice, ninety seconds apart. What we found is that the prices agree remarkably well inside each chain, several sit on hard floors, and one common assumption did not survive contact with the data.&lt;/p&gt;

&lt;h2&gt;
  
  
  The method (repeatable in five minutes)
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;eth_gasPrice&lt;/code&gt; is the standard JSON-RPC method every EVM node exposes for fee suggestions. It needs no account and no API key. The exact request we sent to each endpoint, four chains, two rounds:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight http"&gt;&lt;code&gt;&lt;span class="err"&gt;POST &amp;lt;rpc-url&amp;gt;
Content-Type: application/json

{"jsonrpc":"2.0","id":1,"method":"eth_gasPrice","params":[]}
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The response is a hex quantity in wei. The endpoints, chosen so each chain has at least two operators with no shared infrastructure:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Ethereum mainnet&lt;/strong&gt;: &lt;code&gt;ethereum-rpc.publicnode.com&lt;/code&gt;, &lt;code&gt;eth.drpc.org&lt;/code&gt;, &lt;code&gt;1rpc.io/eth&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;BNB Smart Chain&lt;/strong&gt;: &lt;code&gt;bsc-dataseed1.binance.org&lt;/code&gt;, &lt;code&gt;bsc-mainnet.public.blastapi.io&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Arbitrum One&lt;/strong&gt;: &lt;code&gt;arb1.arbitrum.io/rpc&lt;/code&gt; (the chain's own official endpoint) and &lt;code&gt;arbitrum-one-rpc.publicnode.com&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Optimism (OP Mainnet)&lt;/strong&gt;: &lt;code&gt;mainnet.optimism.io&lt;/code&gt; (official) and &lt;code&gt;optimism-rpc.publicnode.com&lt;/code&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Round 1 ran at 02:03:42 UTC and round 2 at 02:05:24 UTC on 2026-11-03. Ten requests total; one (Arbitrum's publicnode endpoint, round 2) failed with a connection reset, so we report nine completed readings. All values below are snapshots that will be wrong by the time you read this — the &lt;em&gt;structure&lt;/em&gt; is the point, and you can re-run the probes yourself.&lt;/p&gt;

&lt;h2&gt;
  
  
  The readings
&lt;/h2&gt;

&lt;p&gt;Values converted from wei; gwei = one-billionth of an ETH/BNB/ARB token unit.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Chain&lt;/th&gt;
&lt;th&gt;Round 1 (gwei)&lt;/th&gt;
&lt;th&gt;Round 2 (gwei)&lt;/th&gt;
&lt;th&gt;Agreement across providers&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Ethereum&lt;/td&gt;
&lt;td&gt;0.051654 / 0.051654 / 0.051651&lt;/td&gt;
&lt;td&gt;0.040292 / 0.040292 / 0.040292&lt;/td&gt;
&lt;td&gt;Byte-identical hex, all three, both rounds&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;BSC&lt;/td&gt;
&lt;td&gt;0.050000 / 0.050000&lt;/td&gt;
&lt;td&gt;0.050000 / 0.050000&lt;/td&gt;
&lt;td&gt;Identical all four calls&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Arbitrum One&lt;/td&gt;
&lt;td&gt;0.020052 / 0.020048&lt;/td&gt;
&lt;td&gt;0.020000&lt;/td&gt;
&lt;td&gt;Within 0.3%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Optimism&lt;/td&gt;
&lt;td&gt;0.001000 / 0.001000&lt;/td&gt;
&lt;td&gt;0.001000 / 0.001000&lt;/td&gt;
&lt;td&gt;Byte-identical hex, all four calls&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The Ethereum agreement is the striking one. On round 1, publicnode and drpc returned exactly &lt;code&gt;0x3142e03&lt;/code&gt; — not "roughly the same number," the same 54,048,771 wei — while 1rpc was within 0.006%. On round 2, &lt;em&gt;all three&lt;/em&gt; independent operators returned the identical hex &lt;code&gt;0x266ccbe&lt;/code&gt;, and mainnet had meanwhile fallen 22% (from 0.0517 to 0.0403 gwei) inside 102 seconds. Independent providers converge because they read the same pending-transaction landscape and, at today's fee levels, everyone's suggestion logic lands on the same floor.&lt;/p&gt;

&lt;h2&gt;
  
  
  Three floors, and one assumption that broke
&lt;/h2&gt;

&lt;p&gt;Look at BSC: four calls to two independent providers, and every single one returned exactly &lt;code&gt;0x2faf080&lt;/code&gt; = 0.05 gwei. A number that pinned cannot be a market price — it is a minimum accepted gas price, a floor below which the chain will not accept transactions. Optimism shows the same signature at exactly 0.001 gwei (hex &lt;code&gt;0xf4364&lt;/code&gt;, one-tick variation between rounds). Arbitrum hovers at 0.02 gwei. These endpoints are telling you the chains' &lt;em&gt;rate limits&lt;/em&gt;, not their congestion.&lt;/p&gt;

&lt;p&gt;Which brings us to the assumption that broke. If you asked most people to rank "mainnet vs the BNB chain" on gas price, they would not put Ethereum first. Yet in this window, Ethereum mainnet's suggested gas price (0.0403 gwei on round 2) came in &lt;em&gt;below&lt;/em&gt; BSC's floor (0.05). Read that sentence precisely, because the temptation to over-interpret it is exactly the trap this article exists to describe.&lt;/p&gt;

&lt;h2&gt;
  
  
  What gas price is not
&lt;/h2&gt;

&lt;p&gt;A gas price is one factor of a three-factor invoice, and none of the readings above tells you what an operation costs:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Units differ per operation and per chain.&lt;/strong&gt; A plain ETH transfer burns a fixed 21,000 gas, so mainnet cost ≈ gas price × 21,000. Other operations — swaps, mints, deposits — multiply by tens or hundreds of times more gas, and the exact figure depends on the contract and state access, not the chain's headline rate.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;On the two Layer 2s above, &lt;code&gt;eth_gasPrice&lt;/code&gt; captures only the execution leg.&lt;/strong&gt; Arbitrum and Optimism additionally charge for posting your transaction data to Ethereum (Arbitrum's two-dimensional fee model; Optimism's L1 cost component). A low L2 gas price multiplied by a small execution unit still has to be added to an L1 data fee that this method does not report at all.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The token price is the third multiplier.&lt;/strong&gt; 0.04 gwei priced in ETH and 0.001 gwei priced in OP are only comparable after you multiply by two different market prices.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;So the honest ratios this window supports are &lt;em&gt;gas-price-term&lt;/em&gt; ratios only: Ethereum's suggested gas price sat about 40x above Optimism's, roughly 2x above Arbitrum's, and just under 1x of BSC's floor. Whether any given transfer actually costs less on Optimism than on BSC depends on units, the L1 fee, and token prices — numbers we deliberately did not convert here, because we measured the inputs, not your invoice.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why asking the chains directly is worth the five minutes
&lt;/h2&gt;

&lt;p&gt;Every number in this article came from the chain operators themselves — two of the nine endpoints are the chains' own official public RPCs — with no aggregator, no dashboard, and no account in between. When three operators that share nothing return the identical hex quantity, you have a fact about chain state, not a product's opinion. When a value freezes at the same hex across providers and minutes, you have found a floor. Both are things you can only see by asking directly.&lt;/p&gt;

&lt;p&gt;Re-run the probe any day you like: the request body is above, all endpoints are public, and no key is required. Note the UTC timestamp on your readings, because on Ethereum, gas prices we logged moved 22% inside under two minutes.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Not financial advice.&lt;/strong&gt; Fee data changes block by block; nothing here predicts future costs or recommends any chain, token, RPC provider, or exchange.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Verify official sources.&lt;/strong&gt; Read the fee mechanics yourself at the primary sources: the official Arbitrum and Optimism documentation on how L2 fees are composed (arbitrum.io and docs.optimism.io), each RPC operator's own site for its endpoint terms, and your wallet or node for your actual transaction's gas usage. The readings in this article were captured live on 2026-11-03 at 02:03–02:05 UTC directly from the listed endpoints and can be re-measured by anyone, without an account.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;FeeAtlas publishes measurement logs of exchange and network fees, written with AI assistance, from first-hand data only.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>crypto</category>
      <category>ethereum</category>
      <category>api</category>
      <category>education</category>
    </item>
    <item>
      <title>Not All Funding Settles Every 8 Hours: 417 Contracts on One Exchange Pay Every 4, and Four Pay Every Hour</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Fri, 30 Oct 2026 02:18:16 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/not-all-funding-settles-every-8-hours-417-contracts-on-one-exchange-pay-every-4-and-four-pay-2nn5</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/not-all-funding-settles-every-8-hours-417-contracts-on-one-exchange-pay-every-4-and-four-pay-2nn5</guid>
      <description>&lt;h1&gt;
  
  
  Not All Funding Settles Every 8 Hours: 417 Contracts on One Exchange Pay Every 4, and Four Pay Every Hour
&lt;/h1&gt;

&lt;p&gt;If you learned crypto derivatives from a textbook, the sentence is probably burned into your brain: &lt;em&gt;perpetual futures pay funding every 8 hours&lt;/em&gt;. It is a useful default, and it is also wrong about half the market. We pulled the public, no-key market endpoints of three exchanges on 2026-10-30 between 02:05 and 02:10 UTC, counted every live contract, and found that on one venue alone, nearly half of all perpetuals settle funding every 4 hours — and four contracts settle it every single hour.&lt;/p&gt;

&lt;p&gt;This matters for a very practical reason: the number quoted as a "funding rate" is meaningless without its interval. The same 0.01% quoted per 8 hours and per 1 hour differ by a factor of 24 once you annualize them.&lt;/p&gt;

&lt;h2&gt;
  
  
  What we measured, and how
&lt;/h2&gt;

&lt;p&gt;Everything below comes from three independent, official, publicly documented endpoints that require no API key:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Bybit: &lt;code&gt;GET /v5/market/tickers?category=linear&lt;/code&gt; (one-time snapshot, 928 entries) and &lt;code&gt;GET /v5/market/instruments-info?category=linear&lt;/code&gt; (used to filter to &lt;code&gt;contractType&lt;/code&gt; containing "Perpetual": 881 contracts, versus 40 &lt;code&gt;LinearFutures&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;Binance USDⓈ-M Futures: &lt;code&gt;GET /fapi/v1/fundingRate?symbol=BTCUSDT&lt;/code&gt; (settlement history)&lt;/li&gt;
&lt;li&gt;OKX: &lt;code&gt;GET /api/v5/public/funding-rate&lt;/code&gt; for &lt;code&gt;BTC-USDT-SWAP&lt;/code&gt; and &lt;code&gt;PEPE-USDT-SWAP&lt;/code&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;We ran the probes twice within a five-minute window. All figures repeated; contract counts are snapshots of a moving listing universe.&lt;/p&gt;

&lt;h2&gt;
  
  
  Bybit: the 8-hour grid is a minority of one
&lt;/h2&gt;

&lt;p&gt;The Bybit tickers response carries a &lt;code&gt;fundingIntervalHour&lt;/code&gt; field per contract. Filtering to the 881 contracts whose &lt;code&gt;contractType&lt;/code&gt; is &lt;code&gt;LinearPerpetual&lt;/code&gt;, the distribution at 02:09 UTC was:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Funding interval&lt;/th&gt;
&lt;th&gt;Contracts&lt;/th&gt;
&lt;th&gt;Share of 881 perpetuals&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;8 hours&lt;/td&gt;
&lt;td&gt;460&lt;/td&gt;
&lt;td&gt;52.2%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;4 hours&lt;/td&gt;
&lt;td&gt;417&lt;/td&gt;
&lt;td&gt;47.3%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;1 hour&lt;/td&gt;
&lt;td&gt;4&lt;/td&gt;
&lt;td&gt;0.5%&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;So the "standard" 8-hour rhythm covers barely half of the venue's own perpetual listings. The 4-hour camp is not a rounding error — it is 417 contracts, within a handful of the majority.&lt;/p&gt;

&lt;p&gt;The hourly outliers, listed by their official symbols, are &lt;code&gt;BLURUSDT&lt;/code&gt;, &lt;code&gt;BMTUSDT&lt;/code&gt;, &lt;code&gt;IOSTUSDT&lt;/code&gt; and &lt;code&gt;IRYSUSDT&lt;/code&gt;. We are stating a spec, not recommending or criticizing anything: on these four contracts, funding is charged or paid 24 times a day.&lt;/p&gt;

&lt;p&gt;Two cross-checks inside the same responses confirm the structure:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;BTCUSDT&lt;/code&gt; itself reports &lt;code&gt;fundingIntervalHour: "8"&lt;/code&gt;, &lt;code&gt;fundingCap: "0.00333"&lt;/code&gt; (i.e., ±0.333% per interval), with a current rate of 0.0001 (0.01%) at snapshot time.&lt;/li&gt;
&lt;li&gt;The remaining 40 contracts in the linear category returned &lt;code&gt;fundingIntervalHour: "0"&lt;/code&gt; — and &lt;code&gt;instruments-info&lt;/code&gt; shows exactly 40 contracts with &lt;code&gt;contractType: "LinearFutures"&lt;/code&gt;. Quarterly futures charge no funding at all; their carry hides in the basis instead. The zero in one endpoint and the count in another cross-confirm each other.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  OKX: same grid, different ceilings
&lt;/h2&gt;

&lt;p&gt;OKX exposes &lt;code&gt;fundingTime&lt;/code&gt; and &lt;code&gt;nextFundingTime&lt;/code&gt; per instrument. For both &lt;code&gt;BTC-USDT-SWAP&lt;/code&gt; and &lt;code&gt;PEPE-USDT-SWAP&lt;/code&gt;, the difference at both snapshot rounds was exactly 28,800,000 milliseconds — precisely 8 hours. So these two contracts follow the textbook cadence.&lt;/p&gt;

&lt;p&gt;What differs is the ceiling. &lt;code&gt;BTC-USDT-SWAP&lt;/code&gt; reports &lt;code&gt;maxFundingRate&lt;/code&gt; of 0.00375 (±0.375%), while &lt;code&gt;PEPE-USDT-SWAP&lt;/code&gt; allows ±1%. Same exchange, same interval, a cap 2.67x apart. The cap is a per-contract parameter, not a per-venue one.&lt;/p&gt;

&lt;h2&gt;
  
  
  Binance: proving the grid is real, to the millisecond
&lt;/h2&gt;

&lt;p&gt;Binance does not publish an "interval" field; it just settles. But the funding-history endpoint lets you measure the grid directly. The consecutive &lt;code&gt;fundingTime&lt;/code&gt; values for &lt;code&gt;BTCUSDT&lt;/code&gt; came back 28,799,997 / 28,800,000 / 28,800,003 milliseconds apart — the tiny jitter is server-side rounding, and the pattern is an exact 8-hour lattice at 00:00, 08:00, 16:00 UTC. The three most recent settled rates in our pull were 0.00005306, 0.00010000 and 0.00003619.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why the interval changes the whole arithmetic
&lt;/h2&gt;

&lt;p&gt;Funding is quoted per interval, and the annualization is just:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;annualized ≈ rate_per_interval × settlements_per_day × 365
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Plug in the same 0.01% rate at three different cadences:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Every 8 hours → ×3 per day → ≈ 10.95% per year&lt;/li&gt;
&lt;li&gt;Every 4 hours → ×6 per day → ≈ 21.9% per year&lt;/li&gt;
&lt;li&gt;Every 1 hour → ×24 per day → ≈ 87.6% per year&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;A backtest or a carry dashboard that hard-codes "×3×365" is silently wrong for the 47% of Bybit perpetuals that settle every 4 hours, and catastrophically wrong for the hourly ones. Before you compare the funding rate of two contracts, compare their intervals first — the same logic applies to the cap: Bybit's BTC ceiling is ±0.333% per 8 hours, OKX's is ±0.375% per 8 hours, and OKX's PEPE is ±1% per 8 hours. And if you look at venues like Deribit discussed elsewhere, some settle BTC funding hourly — the 8-hour "industry standard" is really a habit of a few large venues, not a law.&lt;/p&gt;

&lt;h2&gt;
  
  
  Caveats, stated plainly
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Every number here is a snapshot taken 2026-10-30 02:05–02:10 UTC. Contract counts drift as venues list and delist; caps, intervals and current rates can be edited by each exchange at any time.&lt;/li&gt;
&lt;li&gt;We re-ran the probes minutes later and the structure repeated, but only the two-round consistency is ours; the live values belong to the exchanges.&lt;/li&gt;
&lt;li&gt;We are describing published specifications and settled data. No ranking of exchanges, no suggestion to trade any listed symbol.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Disclosures
&lt;/h2&gt;

&lt;p&gt;&lt;em&gt;AI-assisted content: this article was researched and drafted with AI assistance under human editorial oversight, using official exchange endpoints as primary sources.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;#ad affiliate disclosure: this post contains no sponsored or affiliate links at the time of publication. If that ever changes, any affiliate link will be clearly marked with #ad immediately before the link.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Not financial advice: nothing here is investment, trading, or financial advice. Derivatives involve substantial risk of loss. Verify every figure against the official sources below before relying on it.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Verify official sources
&lt;/h2&gt;

&lt;p&gt;You can reproduce every claim without an account:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Bybit market tickers (field &lt;code&gt;fundingIntervalHour&lt;/code&gt;): &lt;code&gt;https://api.bybit.com/v5/market/tickers?category=linear&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;Bybit instrument specs (&lt;code&gt;contractType&lt;/code&gt;): &lt;code&gt;https://api.bybit.com/v5/market/instruments-info?category=linear&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;Bybit public API docs: &lt;code&gt;https://bybit-exchange.github.io/docs/v5/market/tickers&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;Binance funding history: &lt;code&gt;https://fapi.binance.com/fapi/v1/fundingRate?symbol=BTCUSDT&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;OKX funding rate (fields &lt;code&gt;fundingTime&lt;/code&gt;, &lt;code&gt;nextFundingTime&lt;/code&gt;, &lt;code&gt;maxFundingRate&lt;/code&gt;): &lt;code&gt;https://www.okx.com/api/v5/public/funding-rate?instId=BTC-USDT-SWAP&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;OKX public API docs: &lt;code&gt;https://www.okx.com/docs-v5/en/&lt;/code&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Snapshots: 2026-10-30 02:05–02:10 UTC. Live values will differ — that is the point of checking.&lt;/p&gt;

</description>
      <category>crypto</category>
      <category>bitcoin</category>
      <category>api</category>
      <category>education</category>
    </item>
    <item>
      <title>Your Maximum Leverage Is a Staircase, Not a Number: 35 Rungs on One Exchange, 99 on the Other — Both Straight From the API</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Thu, 29 Oct 2026 07:08:47 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/your-maximum-leverage-is-a-staircase-not-a-number-35-rungs-on-one-exchange-99-on-the-other--1k0j</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/your-maximum-leverage-is-a-staircase-not-a-number-35-rungs-on-one-exchange-99-on-the-other--1k0j</guid>
      <description>&lt;h1&gt;
  
  
  Your Maximum Leverage Is a Staircase, Not a Number: 35 Rungs on One Exchange, 99 on the Other — Both Straight From the API
&lt;/h1&gt;

&lt;p&gt;Every exchange landing page sells leverage as a single number. "Up to 150x" is easy to print and easy to remember. But the venue's own public API dissolves that number into a ladder: each rung pairs a position-size boundary with a lower leverage ceiling and a higher maintenance-margin rate. On 2026-10-29 (06:51 UTC) I fetched the complete ladders for the BTCUSD perpetual from two exchanges — no API key needed — and counted the steps: 35 rungs at one venue, 99 at the other. Same asset, same concept, two very different staircases.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Quick disclosures:&lt;/strong&gt; &lt;em&gt;AI-assisted content: this article was researched and drafted with AI assistance, and every number below was re-checked against the official endpoints and documentation cited at the end. #ad — affiliate disclosure: this post carries no affiliate or referral links; if future posts in this series do, they will be labelled right where they appear. This article is not financial advice — it is a measurement of published contract specifications.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  The advertised ceiling belongs to your smallest possible position
&lt;/h2&gt;

&lt;p&gt;Bybit's public risk-limit endpoint (&lt;code&gt;GET /v5/market/risk-limit?category=linear&amp;amp;symbol=BTCUSDT&lt;/code&gt; on api.bybit.com) returns exactly &lt;strong&gt;35 tiers&lt;/strong&gt; for the BTCUSDT linear perpetual. The very first rung reads: &lt;code&gt;riskLimitValue&lt;/code&gt; 300,000, &lt;code&gt;maxLeverage&lt;/code&gt; 150.00, &lt;code&gt;maintenanceMargin&lt;/code&gt; 0.0033. In plain words: the headline 150x is reserved for positions under &lt;strong&gt;$300,000&lt;/strong&gt; of notional. Here are selected rungs of that staircase, verbatim from the wire:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Rung&lt;/th&gt;
&lt;th&gt;Notional boundary&lt;/th&gt;
&lt;th&gt;Max leverage&lt;/th&gt;
&lt;th&gt;Maintenance margin&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;up to $300,000&lt;/td&gt;
&lt;td&gt;150x&lt;/td&gt;
&lt;td&gt;0.33%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;up to $2,000,000&lt;/td&gt;
&lt;td&gt;100x&lt;/td&gt;
&lt;td&gt;0.50%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;5&lt;/td&gt;
&lt;td&gt;up to $3,800,000&lt;/td&gt;
&lt;td&gt;75x&lt;/td&gt;
&lt;td&gt;0.67%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;10&lt;/td&gt;
&lt;td&gt;up to $10,000,000&lt;/td&gt;
&lt;td&gt;45x&lt;/td&gt;
&lt;td&gt;1.30%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;20&lt;/td&gt;
&lt;td&gt;up to $75,000,000&lt;/td&gt;
&lt;td&gt;15x&lt;/td&gt;
&lt;td&gt;4.00%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;25&lt;/td&gt;
&lt;td&gt;up to $100,000,000&lt;/td&gt;
&lt;td&gt;10x&lt;/td&gt;
&lt;td&gt;6.50%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;30&lt;/td&gt;
&lt;td&gt;up to $150,000,000&lt;/td&gt;
&lt;td&gt;5x&lt;/td&gt;
&lt;td&gt;12.00%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;35&lt;/td&gt;
&lt;td&gt;up to $1,200,000,000&lt;/td&gt;
&lt;td&gt;1x&lt;/td&gt;
&lt;td&gt;60.00%&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;A $10 million position does not get 150x — it gets 45x. The final rung is not leverage at all: 1x with a 60% maintenance-margin rate. From first rung to last, the ceiling falls by a factor of 150 while the margin rate rises by a factor of 182 (0.33% → 60%) — less leverage &lt;em&gt;and&lt;/em&gt; a thinner buffer to liquidation, tightening together at every step.&lt;/p&gt;

&lt;h2&gt;
  
  
  The other exchange counts its staircase in contracts — and it has 99 rungs
&lt;/h2&gt;

&lt;p&gt;OKX publishes the same idea at &lt;code&gt;GET /api/v5/public/position-tiers?uly=BTC-USDT&amp;amp;instType=SWAP&amp;amp;tdMode=cross&lt;/code&gt; on okx.com. One live observation worth flagging: on the capture date this endpoint rejected every request that omitted &lt;code&gt;instType&lt;/code&gt; with code &lt;code&gt;50014&lt;/code&gt; ("Parameter instType can not be empty") — the ladder itself is unchanged from earlier measurements, but required parameters on public endpoints can tighten without notice, so your own clients should be re-checked against the current spec.&lt;/p&gt;

&lt;p&gt;This ladder is expressed in &lt;strong&gt;contracts, not dollars&lt;/strong&gt;. The instruments endpoint (&lt;code&gt;GET /api/v5/public/instruments?instType=SWAP&amp;amp;instId=BTC-USDT-SWAP&lt;/code&gt;) gives &lt;code&gt;ctVal&lt;/code&gt; = 0.01, so one contract is 0.01 BTC. Selected rungs from the live response:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Rung&lt;/th&gt;
&lt;th&gt;Position size&lt;/th&gt;
&lt;th&gt;Max leverage&lt;/th&gt;
&lt;th&gt;MMR&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;up to 1,000 contracts (= 10 BTC)&lt;/td&gt;
&lt;td&gt;100x&lt;/td&gt;
&lt;td&gt;0.40%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;up to 5,000 contracts&lt;/td&gt;
&lt;td&gt;66.66x&lt;/td&gt;
&lt;td&gt;0.50%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;3&lt;/td&gt;
&lt;td&gt;up to 20,000 contracts&lt;/td&gt;
&lt;td&gt;50x&lt;/td&gt;
&lt;td&gt;0.75%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;4&lt;/td&gt;
&lt;td&gt;up to 40,000 contracts&lt;/td&gt;
&lt;td&gt;40x&lt;/td&gt;
&lt;td&gt;1.25%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;5&lt;/td&gt;
&lt;td&gt;up to 60,000 contracts&lt;/td&gt;
&lt;td&gt;33.33x&lt;/td&gt;
&lt;td&gt;1.75%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;6&lt;/td&gt;
&lt;td&gt;up to 80,000 contracts&lt;/td&gt;
&lt;td&gt;28.57x&lt;/td&gt;
&lt;td&gt;2.25%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;99&lt;/td&gt;
&lt;td&gt;up to 1,940,000 contracts (= 19,400 BTC)&lt;/td&gt;
&lt;td&gt;2x&lt;/td&gt;
&lt;td&gt;48.75%&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;To give the dollars some anchor: at the venue's own BTC-USDT-SWAP last price of &lt;strong&gt;79,899.4&lt;/strong&gt; (same minute, 06:51 UTC; Binance's BTCUSDT quoted 79,936.45 in parallel, agreeing within 5 bps), the first rung's 10 BTC is roughly &lt;strong&gt;$799,000&lt;/strong&gt;. So 100x on OKX lives below about $0.8 million of notional — the same physics Bybit encodes as "150x below $300,000".&lt;/p&gt;

&lt;h2&gt;
  
  
  Same physics, different bookkeeping
&lt;/h2&gt;

&lt;p&gt;Put the two ladders side by side and three neutral observations fall out:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;The units are not comparable at face value.&lt;/strong&gt; Bybit's boundaries are USD notional; OKX's are contract counts that need &lt;code&gt;ctVal&lt;/code&gt; and a live price before any number means anything. Comparing "300,000" against "1,000" is a category error.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The granularity is wildly different: 35 steps versus 99.&lt;/strong&gt; Bybit drops 50 leverage points in a single step (150 → 100). OKX near the bottom of its ladder steps down in increments only visible to two decimals: rungs 96–99 run 2.06x → 2.04x → 2.02x → 2x. One staircase descends in leaps, the other in millimetres.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The floors disagree.&lt;/strong&gt; Bybit's last rung is 1x at $1.2 billion with 60% MMR; OKX's is 2x at 19,400 BTC — roughly $1.55 billion at the same snapshot — with 48.75% MMR. Both say "at whale size, leverage stops being borrowable", but they stop at different rungs, and either table can change without announcement.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  What this changes for anyone who builds or measures
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Treat headline leverage as a floor-tier spec value.&lt;/strong&gt; The advertised number belongs to the smallest position on the book.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Compute margin from the tier you actually sit in.&lt;/strong&gt; A tool that divides notional by the first-tier rate while your size sits on the tenth rung can be off by more than 3x (1.30% vs 0.33%).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Liquidation distance is set by your rung's MMR, not the advertised one.&lt;/strong&gt; Any model that hard-codes 0.33% will systematically underestimate how close large positions run to the edge.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Refetch, do not cache forever.&lt;/strong&gt; Both ladders here were byte-stable across two fetches seconds apart on the capture date — a property of this week, not of the endpoint contract.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;None of this ranks one venue against the other. The point is smaller and more useful: "maximum leverage" is not one number, it is a table, and both exchanges hand you the whole table for free if you ask.&lt;/p&gt;

&lt;h2&gt;
  
  
  Numbers move — verify official sources
&lt;/h2&gt;

&lt;p&gt;Every figure here is a snapshot from 2026-10-29 06:51–06:52 UTC, captured twice within seconds of each other (the two Bybit fetches agreed on all 35 rungs; the two OKX fetches were byte-identical). Re-run the calls yourself: api.bybit.com &lt;code&gt;GET /v5/market/risk-limit?category=linear&amp;amp;symbol=BTCUSDT&lt;/code&gt;, okx.com &lt;code&gt;GET /api/v5/public/position-tiers?uly=BTC-USDT&amp;amp;instType=SWAP&amp;amp;tdMode=cross&lt;/code&gt; and &lt;code&gt;GET /api/v5/public/instruments?instType=SWAP&amp;amp;instId=BTC-USDT-SWAP&lt;/code&gt;. For the written field definitions: the risk-limit page on bybit-exchange.github.io/docs/v5/market and the position-tiers section of okx.com/docs-v5 — both reachable (HTTP 200) at capture time. Verify official sources before you hard-code any of this into your own code.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Not financial advice. This post reports published contract parameters from public APIs only; it recommends no exchange, no leverage level, no strategy, and no trade. Leverage magnifies losses as fast as gains.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;AI-assisted content disclosure: researched and drafted with AI assistance on 2026-10-29 against the cited official sources. #ad: no affiliate or referral links are present in this post.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>api</category>
      <category>bitcoin</category>
      <category>crypto</category>
      <category>trading</category>
    </item>
    <item>
      <title>One Call, Five Answers: How Many Candles Each Exchange API Actually Returns</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Tue, 20 Oct 2026 02:15:42 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/one-call-five-answers-how-many-candles-each-exchange-api-actually-returns-g6d</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/one-call-five-answers-how-many-candles-each-exchange-api-actually-returns-g6d</guid>
      <description>&lt;h1&gt;
  
  
  One Call, Five Answers: How Many Candles Each Exchange API Actually Returns
&lt;/h1&gt;

&lt;p&gt;Every trading-data script eventually grows the same loop: request candles, remember the last timestamp, ask for the next page. The loop breaks the first time it touches a second exchange, because "how many candles come back in one call" is not one number. On the morning of 2026-10-20 (02:00–02:10 UTC) I asked five exchange APIs the same question with the same kind of request — hourly candles on BTC — and got five different answers. One venue's documentation and its own endpoint even disagree by exactly one candle.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Quick disclosures:&lt;/strong&gt; &lt;em&gt;AI-assisted content: this article was researched and drafted with AI assistance, and every number below was re-checked against the official endpoints and documentation cited at the end. #ad — affiliate disclosure: this post carries no affiliate or referral links today; if future posts in this series do, they will be labelled right where they appear. This article is not financial advice — it is a documentation study of API behaviour.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  The five answers, up front
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Venue&lt;/th&gt;
&lt;th&gt;Docs say&lt;/th&gt;
&lt;th&gt;Wire says&lt;/th&gt;
&lt;th&gt;Docs' default size&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Binance&lt;/td&gt;
&lt;td&gt;Max 1000, default 500&lt;/td&gt;
&lt;td&gt;1000 bars (asked 1500); 500 by default&lt;/td&gt;
&lt;td&gt;500&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Bybit&lt;/td&gt;
&lt;td&gt;[1, 1000], default 200&lt;/td&gt;
&lt;td&gt;1000 bars (asked 1001); 200 by default&lt;/td&gt;
&lt;td&gt;200&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;OKX&lt;/td&gt;
&lt;td&gt;Max 300, default 100&lt;/td&gt;
&lt;td&gt;300 bars (asked 500); 100 by default&lt;/td&gt;
&lt;td&gt;100&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Kraken&lt;/td&gt;
&lt;td&gt;"Maximum 720 candles per call"&lt;/td&gt;
&lt;td&gt;
&lt;strong&gt;721 rows&lt;/strong&gt;, no limit parameter exists&lt;/td&gt;
&lt;td&gt;—&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;KuCoin&lt;/td&gt;
&lt;td&gt;no text documentation found&lt;/td&gt;
&lt;td&gt;100 bars&lt;/td&gt;
&lt;td&gt;—&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Four different caps (100 / 300 / 720 / 1000) and three different defaults (100 / 200 / 500) for the same everyday request. Same asset, same interval, same minute, five rulebooks.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I actually ran
&lt;/h2&gt;

&lt;p&gt;All five calls are public, key-free, and repeatable:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Binance&lt;/strong&gt; &lt;code&gt;GET /api/v3/klines&lt;/code&gt; on BTCUSDT, interval 1h, &lt;code&gt;limit=1500&lt;/code&gt; → exactly &lt;strong&gt;1,000&lt;/strong&gt; rows. Same call with no &lt;code&gt;limit&lt;/code&gt; → &lt;strong&gt;500&lt;/strong&gt; rows.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Bybit&lt;/strong&gt; &lt;code&gt;GET /v5/market/kline&lt;/code&gt; (category spot, BTCUSDT, interval 60), &lt;code&gt;limit=1001&lt;/code&gt; → exactly &lt;strong&gt;1,000&lt;/strong&gt; rows. No &lt;code&gt;limit&lt;/code&gt; → &lt;strong&gt;200&lt;/strong&gt; rows.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;OKX&lt;/strong&gt; &lt;code&gt;GET /api/v5/market/candles&lt;/code&gt; on BTC-USDT, bar 1H, &lt;code&gt;limit=500&lt;/code&gt; → exactly &lt;strong&gt;300&lt;/strong&gt; rows. No &lt;code&gt;limit&lt;/code&gt; → &lt;strong&gt;100&lt;/strong&gt; rows.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Kraken&lt;/strong&gt; &lt;code&gt;GET /0/public/OHLC&lt;/code&gt; on XBTUSD, interval 60 → &lt;strong&gt;721&lt;/strong&gt; rows. The endpoint has no limit parameter at all.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;KuCoin&lt;/strong&gt; &lt;code&gt;GET /api/v1/market/candles&lt;/code&gt; (BTC-USDT, type 1hour) → &lt;strong&gt;100&lt;/strong&gt; rows.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Over caps are not errors — Binance and Bybit did not complain about &lt;code&gt;limit=1500&lt;/code&gt; or &lt;code&gt;limit=1001&lt;/code&gt;. They silently returned their maximum. A client that asks for 5,000 and trusts the request will quietly believe it fetched 5,000.&lt;/p&gt;

&lt;h2&gt;
  
  
  The documentation layer, quoted
&lt;/h2&gt;

&lt;p&gt;A number on the wire is nice; a number the venue actually commits to is better. Same morning, official text surfaces:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Binance&lt;/strong&gt; spot REST spec, klines section: "limit | INT | NO | Default: 500; Maximum: 1000."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Bybit&lt;/strong&gt; kline page: "Limit for data size per page. [1, 1000]. Default: 200."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;OKX&lt;/strong&gt; REST docs, candles section: "Number of results per request. The maximum is 300. The default is 100."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Kraken&lt;/strong&gt; historical-data guide: "Maximum 720 candles per call."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;KuCoin&lt;/strong&gt;: the docs site is a JavaScript application; I found no text-rendered page stating the candle depth. Reported as observed — the 100-row figure above is measurement, not a quote.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;So four of five venues have their number in writing, and the live wire agrees with every written number. Almost.&lt;/p&gt;

&lt;h2&gt;
  
  
  The one-candle disagreement
&lt;/h2&gt;

&lt;p&gt;Kraken's guide says 720. The endpoint returned 721 rows.&lt;/p&gt;

&lt;p&gt;The arithmetic makes the disagreement visible: the first row opens at 2026-09-20 02:00 UTC and the last row is stamped 2026-10-20 02:00 UTC — exactly 2,592,000 seconds apart, which is precisely 720 hour-wide slots. Seven hundred twenty-one rows spanning seven hundred twenty slots. The most plausible reading is that the final row is the bar still forming at request time (I called shortly after 02:00 UTC, and a closed candle's close price is simply the last traded price on most venues, so a live bar looks identical to a finished one in the response). Kraken's own pagination hint — the &lt;code&gt;last&lt;/code&gt; field in the response — pointed at 01:00, i.e. the second-to-last row, as the resume marker, which quietly treats the final row as not-yet-final.&lt;/p&gt;

&lt;p&gt;Kraken's documentation does not adjudicate this, so I will not either. The point for builders: a documented cap and the row count are two different measurements, and one candle of drift per page is exactly the kind of off-by-one that turns into a silent gap — or a duplicated bar — after two hundred pages.&lt;/p&gt;

&lt;h2&gt;
  
  
  A quiet cross-check worth copying
&lt;/h2&gt;

&lt;p&gt;Binance and Bybit disagree on defaults (500 vs 200), but both cap at 1000 — and on the same minute of capture, both handed me a window whose oldest bar was 2026-09-08 11:00 UTC. Two independent venues, one thousand hours back from the same moment, agreeing to the hour. When you build a cross-venue pipeline, checks like this — where two sources should agree and do — cost nothing and catch whole classes of silent truncation.&lt;/p&gt;

&lt;h2&gt;
  
  
  What a portable fetcher must carry
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;A per-venue page-size constant.&lt;/strong&gt; Read each venue's own limit line; never port one venue's constant to another. 100-to-1000 across five mainstream venues.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Never omit &lt;code&gt;limit&lt;/code&gt;.&lt;/strong&gt; The three documented defaults differ 5× (100/200/500), and KuCoin offers no documented default at all.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Row order is also per-venue.&lt;/strong&gt; In my same-minute captures, Binance returned oldest bar first (row 0 was the oldest), while Bybit returned newest first (the oldest sat at the end of the array). Same pair, same interval, opposite arrays.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Decide what to do with live bars.&lt;/strong&gt; Kraken hands you the forming candle; whether your backtest wants it is a policy decision, not a default.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Treat over-cap requests as truncation signals.&lt;/strong&gt; Asking for 1500 and receiving 1000 is correct behaviour — but only if your code compares asked versus received instead of trusting the request.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;None of this is exotic. It is five ordinary rulebooks, each written by one venue, each silently binding the moment your script touches that venue.&lt;/p&gt;

&lt;h2&gt;
  
  
  Numbers move — verify official sources
&lt;/h2&gt;

&lt;p&gt;Every figure here is a snapshot from 2026-10-20 02:00–02:10 UTC, and any venue can change its limits without announcing it. Re-run the calls yourself: api.binance.com &lt;code&gt;GET /api/v3/klines&lt;/code&gt;; api.bybit.com &lt;code&gt;GET /v5/market/kline&lt;/code&gt;; okx.com &lt;code&gt;GET /api/v5/market/candles&lt;/code&gt;; api.kraken.com &lt;code&gt;GET /0/public/OHLC&lt;/code&gt;; api.kucoin.com &lt;code&gt;GET /api/v1/market/candles&lt;/code&gt;. For the written rules: the klines section of Binance's spot REST spec on GitHub, the kline page on bybit-exchange.github.io/docs/v5, the candles section of okx.com/docs-v5, and the historical-data guide on docs.kraken.com. Verify official sources before you hard-code any number into your own code.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Not financial advice. This post reports on API documentation and observed endpoint behaviour only; it recommends no exchange, no strategy, and no trade.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;AI-assisted content disclosure: researched and drafted with AI assistance on 2026-10-20 against the cited official sources. #ad: no affiliate or referral links are present in this post.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>api</category>
      <category>crypto</category>
      <category>bitcoin</category>
      <category>developers</category>
    </item>
    <item>
      <title>Funding Settles Every 8 Hours — but Four of These Five Exchanges Have No 8-Hour Candle</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Fri, 16 Oct 2026 02:18:52 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/funding-settles-every-8-hours-but-four-of-these-five-exchanges-have-no-8-hour-candle-1plc</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/funding-settles-every-8-hours-but-four-of-these-five-exchanges-have-no-8-hour-candle-1plc</guid>
      <description>&lt;h1&gt;
  
  
  Funding Settles Every 8 Hours — but Four of These Five Exchanges Have No 8-Hour Candle
&lt;/h1&gt;

&lt;p&gt;If you hold a BTC perpetual on Binance, Bybit, or OKX, your funding payment changes hands every 8 hours, on the 00:00 / 08:00 / 16:00 UTC grid. That number is written into the exchanges' own APIs: the interval parameter and the settlement timestamps line up on the same three daily points.&lt;/p&gt;

&lt;p&gt;So when you want to plot the cost you are actually paying on the same rhythm as the payments, the obvious move is to pull an 8-hour candle. And that is where the surprise is: most exchanges simply do not have one.&lt;/p&gt;

&lt;p&gt;I went looking for the 8-hour candle at five exchanges on 2026-10-16, between 02:05 and 02:25 UTC, using only free, key-free official endpoints and official documentation pages. Each claim below rests on two layers: what the docs say, and what the live API actually returns. One venue supports the interval. Four do not.&lt;/p&gt;

&lt;h2&gt;
  
  
  Binance: the interval exists — and its documentation says so
&lt;/h2&gt;

&lt;p&gt;Binance's spot REST documentation keeps a small table of valid kline intervals under the anchor &lt;code&gt;kline-intervals&lt;/code&gt;. The hours row reads, verbatim, as fetched from the official repository today:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;hours | &lt;code&gt;1h&lt;/code&gt;, &lt;code&gt;2h&lt;/code&gt;, &lt;code&gt;4h&lt;/code&gt;, &lt;code&gt;6h&lt;/code&gt;, &lt;code&gt;8h&lt;/code&gt;, &lt;code&gt;12h&lt;/code&gt;&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;And the live API agrees. &lt;code&gt;GET /api/v3/klines?symbol=BTCUSDT&amp;amp;interval=8h&lt;/code&gt; returns data. The futures side (&lt;code&gt;/fapi/v1/klines&lt;/code&gt;, interval &lt;code&gt;8h&lt;/code&gt;) also answered — and the difference between two adjacent returned timestamps was exactly 28,800,000 milliseconds, which is 8 hours to the millisecond, no calendar trickery.&lt;/p&gt;

&lt;p&gt;There is a second, quieter lesson hiding in the same test. Spot supports a &lt;code&gt;1s&lt;/code&gt; interval — my probe returned rows. The futures API rejected the exact same parameter with error code &lt;code&gt;-1120&lt;/code&gt; ("Invalid interval"). Same company, same API family, two different interval enums. Never assume an interval that works on one endpoint of a venue exists on its sibling endpoint.&lt;/p&gt;

&lt;h2&gt;
  
  
  Bybit: the ladder stops at 12 hours; 480 minutes is not on it
&lt;/h2&gt;

&lt;p&gt;Bybit's kline documentation lists the accepted values for the &lt;code&gt;interval&lt;/code&gt; parameter as 1, 3, 5, 15, 30, 60, 120, 240, 360, 720, D, W, M — minutes, with day/week/month letters at the end. Notice there is no 480 between 360 and 720. Eight hours is not a step on this ladder.&lt;/p&gt;

&lt;p&gt;The live endpoint behaves exactly as the enum implies:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;interval=360&lt;/code&gt; on spot BTCUSDT: accepted, and adjacent timestamps are 21,600,000 ms apart (6 hours, as advertised);&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;interval=480&lt;/code&gt;: rejected with &lt;code&gt;retCode 10001&lt;/code&gt;, message "Invalid period!";&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;interval=90&lt;/code&gt; (another gap in the ladder): also rejected with the same 10001.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;So the documentation enum and the rejection behavior confirm each other from two independent surfaces: no 8-hour candle, and the API will tell you so explicitly rather than silently rounding to something else.&lt;/p&gt;

&lt;h2&gt;
  
  
  OKX: two candle grids, neither of them 8 hours
&lt;/h2&gt;

&lt;p&gt;OKX's documentation (fetched live today, the same page I cross-checked against a 2026-08-06 Internet Archive snapshot) splits candles into three enumerated groups. The intraday list is:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;[1s/1m/3m/5m/15m/30m/1H/2H/4H]&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;and above 4 hours it jumps straight to two daily-grid families:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;UTC+8 opening price k-line: [6H/12H/1D/2D/3D/1W/1M/3M]&lt;br&gt;
UTC+0 opening price k-line: [6Hutc/12Hutc/1Dutc/2Dutc/3Dutc/1Wutc/1Mutc/3Mutc]&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;No 8H in any of the three lists. The live market endpoint matches the docs: &lt;code&gt;bar=6H&lt;/code&gt; returned candles 21,600,000 ms apart, while &lt;code&gt;bar=8H&lt;/code&gt; came back with error &lt;code&gt;51000&lt;/code&gt; ("Parameter bar error"), as did &lt;code&gt;bar=9H&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;One extra wrinkle worth knowing before you try to synthesize 8-hour bars from OKX data: its default 6H/12H/1D candles open on a UTC+8 midnight grid, and only the suffixed &lt;code&gt;*utc&lt;/code&gt; variants open at UTC midnight. If you aggregate 1-hour candles into 8-hour buckets anchored at UTC 00:00 while mixing in OKX's default bars, your bars will be silently shifted by 8 hours. The funding settlement grid is UTC-anchored; OKX's default candle grid is not.&lt;/p&gt;

&lt;h2&gt;
  
  
  Kraken: the enum is now printed in the docs — 8 hours is absent
&lt;/h2&gt;

&lt;p&gt;Kraken used to be a venue where the interval list could only be inferred from trial and error. That has changed: the machine-readable documentation dump at docs.kraken.com now carries the OHLC endpoint's parameter table explicitly:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;interval | 1, 5, 15, 30, 60, 240, 1440, 10080, 21600 (minutes) | default 1&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;No 480. No 360, either. And the API enforces it — querying XBTUSD with &lt;code&gt;interval=360&lt;/code&gt; or &lt;code&gt;interval=720&lt;/code&gt; returns &lt;code&gt;EGeneral:Invalid arguments&lt;/code&gt;, while 15, 240, and 1440 all return their usual ~720-row responses.&lt;/p&gt;

&lt;p&gt;Worth noting for tooling: this enum includes 1 and 5 minute intervals that many third-party tutorials omit, and its 10080/21600 entries mean "1 week / 2 weeks in minutes" — Kraken expresses weekly candles as minute counts where other venues use letters.&lt;/p&gt;

&lt;h2&gt;
  
  
  Coinbase: six granularities, and 28800 is not one of them
&lt;/h2&gt;

&lt;p&gt;Coinbase Exchange's candle endpoint takes a &lt;code&gt;granularity&lt;/code&gt; in seconds. Today's direct probes: 60, 300, 900, 3600, 21600 and 86400 all answered 200 with data; 7200 (2h), 14400 (4h) and — the point of this exercise — 28800 (8h) all answered HTTP 400 with &lt;code&gt;Unsupported granularity&lt;/code&gt;. Odd values like 3601 fail the same way, so the check is a strict allow-list, not a range.&lt;/p&gt;

&lt;h2&gt;
  
  
  What this means for your charts
&lt;/h2&gt;

&lt;p&gt;The funding cycle is a real 8-hour cash-flow event; the 8-hour candle is a UI convention that most venues never implemented. Three practical consequences:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Build the bar yourself, and say so.&lt;/strong&gt; Aggregating 1h or 4h candles into 00/08/16 UTC buckets is straightforward, but your derived bars are yours, not the exchange's — label them as derived, especially if you publish them.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Check the calendar anchor before aggregating.&lt;/strong&gt; Binance spot candles open at UTC midnight by default, OKX defaults to UTC+8 (with &lt;code&gt;*utc&lt;/code&gt; variants available), and the mismatch is invisible until two charts disagree by one bar.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Expect silent enum differences even within one venue.&lt;/strong&gt; Binance spot &lt;code&gt;1s&lt;/code&gt; works while futures reject it; Kraken encodes weeks as minutes; Coinbase answers in seconds. A cross-venue fetcher needs a per-venue interval map — there is no shared standard.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;None of this makes one exchange better than another; it is just five independent design choices that your software has to know about.&lt;/p&gt;

&lt;h2&gt;
  
  
  Reproduce it yourself
&lt;/h2&gt;

&lt;p&gt;Everything above is one request away: Binance &lt;code&gt;GET /api/v3/klines?interval=8h&lt;/code&gt; plus the kline-intervals table in its spot REST docs; Bybit &lt;code&gt;GET /v5/market/kline?interval=480&lt;/code&gt; plus its kline docs page; OKX &lt;code&gt;GET /api/v5/market/candles?bar=8H&lt;/code&gt; plus the bar enum in docs-v5; Kraken &lt;code&gt;GET /0/public/OHLC?interval=480&lt;/code&gt; plus the parameter table in its docs dump; Coinbase &lt;code&gt;GET /products/BTC-USD/candles?granularity=28800&lt;/code&gt;. All measurements were taken 2026-10-16 02:05–02:25 UTC; exchanges can and do change enums without notice, so verify against the official sources above before you rely on any of it.&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;AI disclosure:&lt;/strong&gt; This article was researched and drafted with AI assistance under a documented methodology (official documentation pages plus live key-free API probes, each cross-checked), then reviewed before publishing.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Not financial advice:&lt;/strong&gt; This article describes public API behavior and documentation enums. It is not investment, trading, or tax advice, and nothing here recommends any exchange, order type, or strategy.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Affiliate disclosure (#ad):&lt;/strong&gt; This publication may carry affiliate links in the future; any such link would be labeled #ad next to it. This article contains no affiliate, referral, or sponsored links.&lt;/p&gt;

</description>
      <category>api</category>
      <category>bitcoin</category>
      <category>crypto</category>
    </item>
    <item>
      <title>Same BTC Perp, Two Leverage Ceilings: 150x on One Venue, 100x on the Other — Both Straight From the API</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Tue, 13 Oct 2026 02:17:01 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/same-btc-perp-two-leverage-ceilings-150x-on-one-venue-100x-on-the-other-both-straight-from-the-14ad</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/same-btc-perp-two-leverage-ceilings-150x-on-one-venue-100x-on-the-other-both-straight-from-the-14ad</guid>
      <description>&lt;h1&gt;
  
  
  Same BTC Perp, Two Leverage Ceilings: 150x on One Venue, 100x on the Other — Both Straight From the API
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;Disclosure: This article was written with AI assistance under human editorial review. Data below was pulled live from public exchange endpoints and may change at any time.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;#ad / Affiliate disclosure: This article currently contains no affiliate links. If that ever changes, affiliate links will be marked clearly here and at each link.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Not financial advice.&lt;/strong&gt; This is an observation about publicly published contract specifications, not a recommendation to use leverage, trade perpetual futures, or choose one venue over another. Leverage magnifies losses as fast as gains, and liquidation can erase your margin in a single candle.&lt;/p&gt;

&lt;h2&gt;
  
  
  The question, kept simple
&lt;/h2&gt;

&lt;p&gt;Bitcoin's most traded derivative in crypto-native circles is the BTC/USDT perpetual futures contract. Nearly every major venue lists one, and the underlying is identical: one Bitcoin, priced against Tether.&lt;/p&gt;

&lt;p&gt;But the rulebooks are not identical. A previous piece in this series compared funding-rate mechanics; this one asks a single, concrete question:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;What is the maximum leverage the exchange itself advertises for its BTC/USDT perpetual — as published in its own public API, today?&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;Two of the largest venues answer without requiring a login, a key, or a page full of JavaScript. We queried both, twice each, on 13 October 2026 between 02:07 and 02:11 UTC.&lt;/p&gt;

&lt;h2&gt;
  
  
  Venue A: the field is called &lt;code&gt;maxLeverage&lt;/code&gt;, and it says 150
&lt;/h2&gt;

&lt;p&gt;Bybit publishes contract specifications through a authentication-free GET endpoint:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight http"&gt;&lt;code&gt;&lt;span class="err"&gt;GET https://api.bybit.com/v5/market/instruments-info?category=linear&amp;amp;symbol=BTCUSDT
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The relevant fields from our capture (identical in both requests, at 02:08 and 02:10 UTC):&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Field&lt;/th&gt;
&lt;th&gt;Value&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;contractType&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;LinearPerpetual&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;status&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;Trading&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;settleCoin&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;USDT&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;leverageFilter.maxLeverage&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;150.00&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;leverageFilter.minLeverage&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;1&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;lotSizeFilter.minOrderQty&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;0.001&lt;/code&gt; BTC&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;lotSizeFilter.minNotionalValue&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;5&lt;/code&gt; USDT&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;
&lt;code&gt;upperFundingRate&lt;/code&gt; / &lt;code&gt;lowerFundingRate&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;&lt;code&gt;±0.00333&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;fundingInterval&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;480&lt;/code&gt; minutes (8 hours)&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;So the headline number for the same contract: &lt;strong&gt;150x&lt;/strong&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Venue B: the field is called &lt;code&gt;lever&lt;/code&gt;, and it says 100
&lt;/h2&gt;

&lt;p&gt;OKX serves the same category of information through its own authentication-free endpoint:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight http"&gt;&lt;code&gt;&lt;span class="err"&gt;GET https://www.okx.com/api/v5/public/instruments?instType=SWAP&amp;amp;instId=BTC-USDT-SWAP
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Our capture (identical in both requests):&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Field&lt;/th&gt;
&lt;th&gt;Value&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;
&lt;code&gt;instType&lt;/code&gt; / &lt;code&gt;ctType&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;SWAP&lt;/code&gt; / &lt;code&gt;linear&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;state&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;live&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;lever&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;100&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;
&lt;code&gt;ctVal&lt;/code&gt; / &lt;code&gt;ctValCcy&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;0.01&lt;/code&gt; &lt;code&gt;BTC&lt;/code&gt; per contract&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;minSz&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;0.01&lt;/code&gt; contracts&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;uly&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;BTC-USDT&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;settleCcy&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;USDT&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;A companion endpoint, &lt;code&gt;GET /api/v5/public/funding-rate?instId=BTC-USDT-SWAP&lt;/code&gt;, returned &lt;code&gt;maxFundingRate = 0.00375&lt;/code&gt; at the same moment — a useful cross-check that these are genuinely two independently maintained rulebooks sitting on top of the same underlying asset.&lt;/p&gt;

&lt;h2&gt;
  
  
  150 vs 100: what the gap actually means
&lt;/h2&gt;

&lt;p&gt;The advertised ceiling differs by half again: 150x versus 100x, the same contract family, the same day, the same minute.&lt;/p&gt;

&lt;p&gt;Three honest caveats before anyone treats this as a scoreboard:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;The ceiling is an upper bound on a selector, not a promise.&lt;/strong&gt; Both venues run tiered risk systems: as position size grows, the maximum leverage you can actually open drops in steps, and the maintenance-margin requirement that defines your distance to liquidation rises. The headline number typically applies only to the smallest position tier. (We have measured those staircases separately in an earlier piece; they are published on the very same public endpoints.)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The units are not the same.&lt;/strong&gt; Bybit quotes order size directly in BTC (&lt;code&gt;minOrderQty 0.001&lt;/code&gt;, a &lt;code&gt;qtyStep&lt;/code&gt; of &lt;code&gt;0.001&lt;/code&gt;). OKX quotes in contracts: one BTC-USDT-SWAP contract (&lt;code&gt;ctVal&lt;/code&gt;) is 0.01 BTC, and the minimum order (&lt;code&gt;minSz&lt;/code&gt;) is 0.01 of a contract — one ten-thousandth of a coin. A naive script that reads "minimum size" from both APIs and compares the raw numbers is comparing coins to contracts.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A ceiling is not a risk comparison.&lt;/strong&gt; We are not claiming either venue is safer, cheaper, or better. What we are claiming is narrow and checkable: the field exists, the field is public, and today the values read 150 and 100.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Why this belongs in the "fees are not the only fine print" series
&lt;/h2&gt;

&lt;p&gt;Fee schedules get all the attention, yet the rulebook around a trade — maximum leverage, tick size, funding caps, order minimums, price collars — is a second layer of costs and constraints that differs from venue to venue far more than most traders expect. And crucially, most of it is now machine-readable: no login, no scraping of a JavaScript page, just a documented GET request whose answer is a number.&lt;/p&gt;

&lt;p&gt;If you are comparing venues, the leverage field is one line in a table you can build yourself in ten minutes. The two endpoints above are the starting point.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verify official sources
&lt;/h2&gt;

&lt;p&gt;Do not take our word for any number in this piece. At any time you can re-run:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;https://api.bybit.com/v5/market/instruments-info?category=linear&amp;amp;symbol=BTCUSDT&lt;/code&gt; — the &lt;code&gt;maxLeverage&lt;/code&gt; and &lt;code&gt;minNotionalValue&lt;/code&gt; fields&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;https://api.bybit.com/v5/market/tickers?category=linear&amp;amp;symbol=BTCUSDT&lt;/code&gt; — the &lt;code&gt;fundingCap&lt;/code&gt; field (read &lt;code&gt;0.00333&lt;/code&gt; at 02:07 UTC)&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;https://www.okx.com/api/v5/public/instruments?instType=SWAP&amp;amp;instId=BTC-USDT-SWAP&lt;/code&gt; — the &lt;code&gt;lever&lt;/code&gt;, &lt;code&gt;ctVal&lt;/code&gt; and &lt;code&gt;minSz&lt;/code&gt; fields&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;https://www.okx.com/api/v5/public/funding-rate?instId=BTC-USDT-SWAP&lt;/code&gt; — the &lt;code&gt;maxFundingRate&lt;/code&gt; field&lt;/li&gt;
&lt;li&gt;Field definitions: the Bybit market-instruments documentation at &lt;code&gt;bybit-exchange.github.io/docs/v5/market/instrument&lt;/code&gt; and the OKX API v5 documentation at &lt;code&gt;www.okx.com/docs-v5/en/&lt;/code&gt; — both fetched live at those addresses, HTTP 200, at 02:10 UTC on 2026-10-13&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;All figures in this article are values returned by those endpoints at 02:07–02:11 UTC on 13 October 2026. Exchanges revise specifications without notice. Always confirm against the venue's own endpoints and pages before acting on any specification.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Once more, plainly: this post is informational only and is Not financial advice. Perpetual futures with high leverage can lose more than your initial margin. Trade only what you can afford to lose, and verify everything.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Written with AI assistance; all numbers were machine-read from the cited official endpoints and double-checked in two independent fetches.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>api</category>
      <category>bitcoin</category>
      <category>crypto</category>
    </item>
    <item>
      <title>Same BTC Perp, Four Self-Reported Fee Schedules: Taker Fees 3.75x Apart — and One Venue Pays Its Makers</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Fri, 09 Oct 2026 02:20:59 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/same-btc-perp-four-self-reported-fee-schedules-taker-fees-375x-apart-and-one-venue-pays-its-3e3n</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/same-btc-perp-four-self-reported-fee-schedules-taker-fees-375x-apart-and-one-venue-pays-its-3e3n</guid>
      <description>&lt;h1&gt;
  
  
  Same BTC Perp, Four Self-Reported Fee Schedules: Taker Fees 3.75x Apart — and One Venue Pays Its Makers
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;Disclosure: This article was drafted with AI assistance from public exchange API data and reviewed before publishing. #ad — affiliate disclosure: this article contains no affiliate or referral links. Nothing below is investment or trading advice (not financial advice). All numbers are point-in-time snapshots from exchange-published endpoints; rules change without notice, so re-check the official sources named at the end.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  The fee page you read is not the fee field the machine reads
&lt;/h2&gt;

&lt;p&gt;Most traders learn exchange fees from a marketing-friendly fee page: a table of tiers, a footnote about VIP levels, maybe a promotional banner. Those pages are rendered with JavaScript, behind logins, and restructured every few months — which is why so many "exchange fee comparison" posts on the internet quietly copy each other.&lt;/p&gt;

&lt;p&gt;But four exchanges publish something rarer: a &lt;strong&gt;base fee-rate field for perpetual contracts on a public, no-key API endpoint&lt;/strong&gt;. The exchange itself reports the number, in a machine-readable field, and you can query it yourself in one request. We did exactly that on 2026-10-09 (02:10–02:14 UTC, two independent rounds, values identical between rounds), for the BTC/USDT perpetual at each venue. Four venues, four answers — and one of them is a &lt;em&gt;negative&lt;/em&gt; maker fee.&lt;/p&gt;

&lt;h2&gt;
  
  
  The four readings
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Venue&lt;/th&gt;
&lt;th&gt;BTC perp symbol&lt;/th&gt;
&lt;th&gt;Maker (base)&lt;/th&gt;
&lt;th&gt;Taker (base)&lt;/th&gt;
&lt;th&gt;Field names&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Gate&lt;/td&gt;
&lt;td&gt;BTC_USDT&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;-0.01%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;0.075%&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;maker_fee_rate&lt;/code&gt; / &lt;code&gt;taker_fee_rate&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;KuCoin&lt;/td&gt;
&lt;td&gt;XBTUSDTM&lt;/td&gt;
&lt;td&gt;0.02%&lt;/td&gt;
&lt;td&gt;0.06%&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;makerFeeRate&lt;/code&gt; / &lt;code&gt;takerFeeRate&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;MEXC&lt;/td&gt;
&lt;td&gt;BTC_USDT&lt;/td&gt;
&lt;td&gt;0%&lt;/td&gt;
&lt;td&gt;0.02%&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;makerFeeRate&lt;/code&gt; / &lt;code&gt;takerFeeRate&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Bitget&lt;/td&gt;
&lt;td&gt;BTCUSDT&lt;/td&gt;
&lt;td&gt;0.02%&lt;/td&gt;
&lt;td&gt;0.06%&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;makerFeeRate&lt;/code&gt; / &lt;code&gt;takerFeeRate&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Read the taker column first. On the same underlying asset, with the same USDT margin mode, a one-way taker print costs &lt;strong&gt;0.02% at MEXC, 0.06% at KuCoin and Bitget, and 0.075% at Gate&lt;/strong&gt; — a 3.75x spread between cheapest and dearest base reading. A round trip (open and close as a taker) is 4 basis points at one venue and 15 at another. That gap is the same order of magnitude as the entire maker/taker discount structure on many venues: before you optimize which side of the book you sit on, the venue choice already moved your cost by more.&lt;/p&gt;

&lt;h2&gt;
  
  
  A negative maker fee is not a typo
&lt;/h2&gt;

&lt;p&gt;Gate's &lt;code&gt;maker_fee_rate&lt;/code&gt; for BTC_USDT reads &lt;code&gt;-0.0001&lt;/code&gt;. Gate's own published SDK documentation defines the field with the sentence &lt;em&gt;"Maker fee rate, negative values indicate rebates"&lt;/em&gt; — so the minus sign is intentional: resting orders that get filled &lt;strong&gt;earn&lt;/strong&gt; 0.01% instead of paying it. Across the venue's full contract set (1,028 symbols in our pull), every single contract reports the same -0.0001 maker / 0.00075 taker pair, and both of our fetch rounds agree, so this is a systematic base schedule, not a stale field on one symbol.&lt;/p&gt;

&lt;p&gt;That gives us three distinct pricing philosophies on the maker side of the same product:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Pay makers&lt;/strong&gt; (Gate, -0.01%): liquidity provision is subsidized.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Free for makers&lt;/strong&gt; (MEXC, exactly 0%): resting costs nothing but earns nothing.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Charge makers&lt;/strong&gt; (KuCoin and Bitget, +0.02%): passive orders still pay, just less.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;"Free maker" and "negative maker" sound identical in a headline and are completely different line items in a PnL statement.&lt;/p&gt;

&lt;h2&gt;
  
  
  Second underlying, same shape
&lt;/h2&gt;

&lt;p&gt;To confirm we were not reading a one-off symbol configuration, we repeated the pull for the ETH perpetual at all four venues: Gate ETH_USDT reports -0.0001 / 0.00075, KuCoin ETHUSDTM 0.0002 / 0.0006, MEXC ETH_USDT 0 / 0.0002, Bitget ETHUSDT 0.0002 / 0.0006. Same four-way pattern, same ordering on the taker side. The schedule differences are venue-wide policies, not per-coin quirks.&lt;/p&gt;

&lt;p&gt;(One honest footnote from our own archive: when this pool first cross-checked MEXC's ETH reading back on 2026-09-08, the taker field showed 0.01%; today it reads 0.02%. Fee fields are live data — that is precisely why they should be re-queried, never quoted from memory.)&lt;/p&gt;

&lt;h2&gt;
  
  
  What these fields do &lt;em&gt;not&lt;/em&gt; claim
&lt;/h2&gt;

&lt;p&gt;Fairness requires four limits, and we will not paper over them:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Base tier only.&lt;/strong&gt; Every venue layers VIP tiers, fee-card levels, and occasional campaigns on top. The numbers above are what the venues themselves label as the standing rate. MEXC even ships a &lt;code&gt;feeRateType&lt;/code&gt; flag that distinguishes &lt;code&gt;BASE&lt;/code&gt; from &lt;code&gt;TEMP&lt;/code&gt; (promotional) rates, and BTC/ETH both read &lt;code&gt;BASE&lt;/code&gt; — a rare, explicit admission of the base-versus-promo distinction.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Self-reported, not audited.&lt;/strong&gt; These are the exchanges' own fields. We report what they publish, not what your invoice will say.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Two venues do not play.&lt;/strong&gt; Kraken, Coinbase, OKX, Bybit and Binance either gate their perp fee fields behind authentication or never expose them publicly, so they cannot appear in this comparison at all. Absence here is an API-transparency fact, not a verdict.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Not a ranking.&lt;/strong&gt; The cheapest base taker rate on paper may come bundled with other trade-offs (spread, depth, funding rules, region availability). Two venues happened to print identical numbers (KuCoin and Bitget, 0.02/0.06) while differing elsewhere — for instance, KuCoin's own contract data shows &lt;code&gt;maxLeverage&lt;/code&gt; 125 consistent with &lt;code&gt;initialMargin&lt;/code&gt; 0.008 (1 ÷ 0.008 = 125), a self-consistency check of its published fields.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why this matters for a fee-atlas mindset
&lt;/h2&gt;

&lt;p&gt;Every cost you can see on a fee page has usually already been reformatted for persuasion. The API fields are the exchange talking to machines: terse, unsigned where the marketing copy would never put a minus sign, and identical whether or not anyone is reading. When four venues say 0.02%, 0.06%, 0.06% and 0.075% about the &lt;em&gt;same&lt;/em&gt; BTC perpetual, the fee page stops being a detail and starts being the product.&lt;/p&gt;

&lt;p&gt;The lesson is mechanical, not editorial: before you compare venues, pull the numbers from the source that has no reason to round them up — and pull them again the day you act, because these fields move.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verify official sources
&lt;/h2&gt;

&lt;p&gt;Do not trust this article. Reproduce it:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Gate: &lt;code&gt;GET https://api.gateio.ws/api/v4/futures/usdt/contracts&lt;/code&gt; (fields &lt;code&gt;maker_fee_rate&lt;/code&gt;, &lt;code&gt;taker_fee_rate&lt;/code&gt;; field semantics defined in Gate's published SDK docs, &lt;code&gt;docs/Contract.md&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;KuCoin: &lt;code&gt;GET https://api-futures.kucoin.com/api/v1/contracts/XBTUSDTM&lt;/code&gt; (fields &lt;code&gt;makerFeeRate&lt;/code&gt;, &lt;code&gt;takerFeeRate&lt;/code&gt;, &lt;code&gt;maxLeverage&lt;/code&gt;, &lt;code&gt;initialMargin&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;MEXC: &lt;code&gt;GET https://contract.mexc.com/api/v1/contract/detail?contract_code=BTC_USDT&lt;/code&gt; (fields &lt;code&gt;makerFeeRate&lt;/code&gt;, &lt;code&gt;takerFeeRate&lt;/code&gt;, &lt;code&gt;feeRateType&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;Bitget: &lt;code&gt;GET https://api.bitget.com/api/v2/mix/market/contracts?productType=usdt-futures&lt;/code&gt; (fields &lt;code&gt;makerFeeRate&lt;/code&gt;, &lt;code&gt;takerFeeRate&lt;/code&gt;)&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;All four are unauthenticated, published by the exchanges themselves, and queried here at 02:10–02:14 UTC on 2026-10-09 (two rounds, identical results). If a field disagrees with the venue's own fee page, the fee page — or your account's tier — wins. &lt;strong&gt;Always verify official sources.&lt;/strong&gt;&lt;/p&gt;




&lt;p&gt;&lt;em&gt;This post is for education and transparency around public market-data fields only. It is not investment, trading, tax or legal advice (not financial advice). Leveraged perpetual contracts can lose more than you deposit. AI-assisted drafting; affiliate disclosure per the note at the top (#ad: none present in this article).&lt;/em&gt;&lt;/p&gt;

</description>
    </item>
    <item>
      <title>Same Wrong Request Type, Four Different Answers: 404 Silence, a Named 405, and Two Venues That Simply Do Not Care</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Tue, 06 Oct 2026 02:20:31 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/same-wrong-request-type-four-different-answers-404-silence-a-named-405-and-two-venues-that-3fo9</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/same-wrong-request-type-four-different-answers-404-silence-a-named-405-and-two-venues-that-3fo9</guid>
      <description>&lt;h1&gt;
  
  
  Same Wrong Request Type, Four Different Answers: 404 Silence, a Named 405, and Two Venues That Simply Do Not Care
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;By FeeAtlas — an independent, neutral researcher of exchange fee and specification data. No affiliations with any venue named here.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Disclosure: this article was drafted with AI assistance from measurements taken directly against the exchanges' own public endpoints. Every number and response body quoted below was captured live and can be reproduced by you in under a minute.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Ad/affiliate disclosure (#ad): this site participates in affiliate programs. There are no affiliate or sponsored links in this article — every link is a link to an official exchange document, placed for verification purposes only.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Every trading API tutorial eventually hits the same wall: you sent a POST where the endpoint wanted a GET. On the open web, HTTP gives a clean, standardized answer — status 405, "Method Not Allowed." Case closed.&lt;/p&gt;

&lt;p&gt;Crypto exchanges, it turns out, hold four completely different opinions about what should happen. Some do not have the concept at all.&lt;/p&gt;

&lt;h2&gt;
  
  
  The experiment: one wrong button, pressed five times
&lt;/h2&gt;

&lt;p&gt;Five exchanges expose a clock endpoint that requires no account and no API key: a GET that returns the server's current time. For this test, each endpoint received the same request it did not ask for — an empty POST — twice, on two independent rounds, about one minute apart. Nothing else changed: same machine, same browser-like request header, no body, no credentials.&lt;/p&gt;

&lt;p&gt;Measured live on 2026-10-06 at 02:08–02:09 UTC (first detection round: earlier the same day at 00:44–00:55 UTC, identical outcomes).&lt;/p&gt;

&lt;h2&gt;
  
  
  Answer 1 — Binance: 404, and a completely empty body
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;POST https://api.binance.com/api/v3/time&lt;/code&gt; → &lt;strong&gt;HTTP 404, zero bytes returned&lt;/strong&gt;. Both rounds, exactly the same.&lt;/p&gt;

&lt;p&gt;The subtlety is worse than the 404 itself. GET a path that genuinely does not exist on Binance and you also get a 404 with an empty body. The response cannot tell you whether your &lt;em&gt;path&lt;/em&gt; is wrong or your &lt;em&gt;method&lt;/em&gt; is wrong — "typo in URL" and "wrong verb" collapse into one indistinguishable silence.&lt;/p&gt;

&lt;h2&gt;
  
  
  Answer 2 — Bybit: 200, and it hands you the time anyway
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;POST https://api.bybit.com/v5/market/time&lt;/code&gt; → &lt;strong&gt;HTTP 200&lt;/strong&gt; with the same JSON you would get from a GET: &lt;code&gt;retCode: 0&lt;/code&gt;, &lt;code&gt;timeSecond&lt;/code&gt;, &lt;code&gt;timeNano&lt;/code&gt;. The public market endpoints simply do not check the verb. Both rounds identical.&lt;/p&gt;

&lt;p&gt;This is friendly the way an unlocked door is friendly. Your script has a bug, and the API rewarded it with success. The bug now lives silently in your codebase until it meets an endpoint that &lt;em&gt;does&lt;/em&gt; check.&lt;/p&gt;

&lt;h2&gt;
  
  
  Answer 3 — OKX: 405 with a name you can grep for
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;POST https://www.okx.com/api/v5/public/time&lt;/code&gt; → &lt;strong&gt;HTTP 405&lt;/strong&gt; with body:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="nl"&gt;"code"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="s2"&gt;"50115"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="nl"&gt;"data"&lt;/span&gt;&lt;span class="p"&gt;:[],&lt;/span&gt;&lt;span class="nl"&gt;"msg"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="s2"&gt;"Invalid request type"&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Byte-identical across both rounds. OKX is the only venue of the four that (a) uses the semantically correct 405 status and (b) wraps it in its usual machine-readable envelope with a named error code. The code is documented: re-checking OKX's own docs-v5 error table today (5,247,990 bytes, fetched and searched live) confirms &lt;strong&gt;50115&lt;/strong&gt; is listed. If you build per-venue error-handling maps, this is the only venue that hands you a key to put in the map.&lt;/p&gt;

&lt;h2&gt;
  
  
  Answer 4 — Kraken: 200, and it isn't even pretending
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;POST https://api.kraken.com/0/public/Time&lt;/code&gt; → &lt;strong&gt;HTTP 200&lt;/strong&gt;, same JSON as GET: &lt;code&gt;unixtime&lt;/code&gt; and &lt;code&gt;rfc1123&lt;/code&gt;. Both rounds.&lt;/p&gt;

&lt;p&gt;Kraken's tolerance is at least consistent with its own published self-description: its documentation states that HTTP status codes are generally not used by its API to convey information (errors arrive inside the JSON &lt;code&gt;error&lt;/code&gt; array instead). A verb check would be out of character.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fifth answer that isn't about methods at all
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;POST https://api.exchange.coinbase.com/time&lt;/code&gt; → &lt;strong&gt;HTTP 401&lt;/strong&gt; &lt;code&gt;{"message":"Unauthorized."}&lt;/code&gt; — while GET on the very same path returns 200 with the time.&lt;/p&gt;

&lt;p&gt;Read that carefully: the response talks about &lt;em&gt;credentials&lt;/em&gt;, not method. Whatever Coinbase does internally, its rejection never mentions the verb, so it tells you nothing about how that venue handles method errors. I record it as a fifth shape — a rejection that answers a question you did not ask — rather than counting it in the four-way comparison. Re-probed a third time: same 401.&lt;/p&gt;

&lt;h2&gt;
  
  
  Results table (measured 2026-10-06, two rounds each)
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Venue&lt;/th&gt;
&lt;th&gt;GET /time&lt;/th&gt;
&lt;th&gt;POST /time&lt;/th&gt;
&lt;th&gt;Dialect&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Binance&lt;/td&gt;
&lt;td&gt;200, &lt;code&gt;{"serverTime":...}&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;404, 0 bytes&lt;/td&gt;
&lt;td&gt;Ambiguous silence&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Bybit&lt;/td&gt;
&lt;td&gt;200, full time JSON&lt;/td&gt;
&lt;td&gt;200, same JSON&lt;/td&gt;
&lt;td&gt;No verb check at all&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;OKX&lt;/td&gt;
&lt;td&gt;200, &lt;code&gt;code:"0"&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;405, code &lt;code&gt;50115&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;Proper 405 + named code&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Kraken&lt;/td&gt;
&lt;td&gt;200, time JSON&lt;/td&gt;
&lt;td&gt;200, same JSON&lt;/td&gt;
&lt;td&gt;Status codes carry no meaning (self-documented)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Coinbase&lt;/td&gt;
&lt;td&gt;200, iso+epoch&lt;/td&gt;
&lt;td&gt;401, "Unauthorized."&lt;/td&gt;
&lt;td&gt;Credential-flavored rejection, method unmentioned&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Why this matters beyond a chuckle
&lt;/h2&gt;

&lt;p&gt;If you maintain one client that talks to several venues — a fee tracker, a monitoring bot, a backtest data pipe — "wrong HTTP method" is the layer where no shared handling exists. One venue gives a 404 indistinguishable from a bad URL. Two give a success you did not earn. One gives a documented code. One gives an error about something else entirely.&lt;/p&gt;

&lt;p&gt;Three practical habits fall out of the table:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Parse bodies, not just status codes.&lt;/strong&gt; A 200 here can mean "you used the wrong verb and nobody noticed." Trusting status codes across venues is exactly the mistake Kraken's own docs warn about.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Log raw responses during development.&lt;/strong&gt; Bybit's and Kraken's tolerance turns method typos into silent drift that surfaces only when a &lt;em&gt;different&lt;/em&gt; endpoint on the same venue does enforce the verb.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Expect the empty-404 class.&lt;/strong&gt; Where a venue cannot distinguish "no such path" from "wrong verb" (Binance today), keep a local list of known-good method/path pairs instead of trusting error feedback.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;None of this makes one venue's API "better." It is one more reminder, in the same family as tick sizes, price collars and rate-limit stopwatches we've measured before: each exchange writes its own rulebook, including for the parts the HTTP RFC was supposed to have settled decades ago.&lt;/p&gt;

&lt;h2&gt;
  
  
  Reproduce it yourself
&lt;/h2&gt;

&lt;p&gt;Open a terminal and run, for each venue, one GET and one POST with a browser-like user agent, and compare the status lines. Five URLs:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;code&gt;https://api.binance.com/api/v3/time&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;https://api.bybit.com/v5/market/time&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;https://www.okx.com/api/v5/public/time&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;https://api.kraken.com/0/public/Time&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;https://api.exchange.coinbase.com/time&lt;/code&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Verify official sources
&lt;/h2&gt;

&lt;p&gt;Numbers in this article are live observations, not quotes from press releases. Before relying on any endpoint behavior, check the venues' own current documentation:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Binance spot API docs (error definitions including the -1xxx parameter family): github.com/binance/binance-spot-api-docs&lt;/li&gt;
&lt;li&gt;Bybit V5 docs (market/time endpoint, rate limits): bybit-exchange.github.io/docs/v5/&lt;/li&gt;
&lt;li&gt;OKX V5 API docs (error code table including 50115): okx.com/docs-v5/en/&lt;/li&gt;
&lt;li&gt;Kraken API docs (response format and error convention): docs.kraken.com&lt;/li&gt;
&lt;li&gt;Coinbase Exchange API docs: docs.cdp.coinbase.com&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Exchange behavior changes without notice. What you can verify in one terminal command is worth more than any summary, including this one.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;This article is provided for information and technical reference only. It is **not financial advice&lt;/em&gt;&lt;em&gt;, not a recommendation of any exchange, and contains no trading guidance. Exchange API specifications and behaviors change at any time; always confirm against the official documentation of each venue.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>api</category>
      <category>backend</category>
      <category>webdev</category>
    </item>
    <item>
      <title>One Concept, Four Field Names, Three Numbers: The Funding-Rate Ceiling on the Same BTC Perp</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Sun, 04 Oct 2026 13:46:34 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/one-concept-four-field-names-three-numbers-the-funding-rate-ceiling-on-the-same-btc-perp-1293</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/one-concept-four-field-names-three-numbers-the-funding-rate-ceiling-on-the-same-btc-perp-1293</guid>
      <description>&lt;h1&gt;
  
  
  One Concept, Four Field Names, Three Numbers: The Funding-Rate Ceiling on the Same BTC Perp
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;Disclosure: this article was researched and drafted with AI assistance, using public exchange APIs and documentation, and reviewed for accuracy against the sources cited below.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;This is not financial advice.&lt;/strong&gt; All numbers below are parameter values read from public exchange APIs at a specific moment in time. Exchanges can and do change these values without notice. Always verify against the official sources listed at the end.&lt;/p&gt;

&lt;h2&gt;
  
  
  The number everyone quotes doesn't exist
&lt;/h2&gt;

&lt;p&gt;Ask five traders what the maximum funding rate on a Bitcoin perpetual is and you'll get a confident shrug. There is no industry-wide ceiling. Funding-rate caps are per-exchange, per-contract parameters — and even for the single most liquid contract in crypto, the BTC/USDT perpetual, the four venues examined here publish &lt;strong&gt;three different answers&lt;/strong&gt; under &lt;strong&gt;four different field names&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;I pulled all of them live, without an API key, from official public endpoints on &lt;strong&gt;October 4, 2026 at 13:37 UTC&lt;/strong&gt;. Here is what the same concept looks like at four exchanges.&lt;/p&gt;

&lt;h2&gt;
  
  
  The four readings, side by side
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Venue&lt;/th&gt;
&lt;th&gt;Field name(s)&lt;/th&gt;
&lt;th&gt;BTC perp cap&lt;/th&gt;
&lt;th&gt;ETH perp cap&lt;/th&gt;
&lt;th&gt;Settlement interval&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Gate&lt;/td&gt;
&lt;td&gt;&lt;code&gt;funding_rate_limit&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;±0.30%&lt;/td&gt;
&lt;td&gt;±0.30%&lt;/td&gt;
&lt;td&gt;28800 s (8 h)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;KuCoin&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;fundingRateCap&lt;/code&gt; / &lt;code&gt;fundingRateFloor&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;±0.30%&lt;/td&gt;
&lt;td&gt;±0.375%&lt;/td&gt;
&lt;td&gt;28800000 ms (8 h)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Bybit&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;upperFundingRate&lt;/code&gt; / &lt;code&gt;lowerFundingRate&lt;/code&gt; (specs) + &lt;code&gt;fundingCap&lt;/code&gt; (tickers)&lt;/td&gt;
&lt;td&gt;±0.333%&lt;/td&gt;
&lt;td&gt;±0.333%&lt;/td&gt;
&lt;td&gt;480 min (8 h)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;OKX&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;maxFundingRate&lt;/code&gt; / &lt;code&gt;minFundingRate&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;±0.375%&lt;/td&gt;
&lt;td&gt;±0.75%&lt;/td&gt;
&lt;td&gt;8 h&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Three numerical answers for the identical underlying exposure: 0.30%, 0.333%, and 0.375%. The spread between the lowest and highest cap is 25% — small in absolute terms, but it defines how far an overheated or capitulating market can push the cost of holding a position before the ceiling bites.&lt;/p&gt;

&lt;p&gt;Live funding values at the same moment, for context (these change every interval and are not the point of this article): Gate BTC perp −0.0017%/8h, Bybit +0.006547%/8h, OKX +0.0027116%/8h, KuCoin −0.0002%/8h.&lt;/p&gt;

&lt;h2&gt;
  
  
  Four names for one concept
&lt;/h2&gt;

&lt;p&gt;If you are writing any tooling that compares funding terms across venues, the field names alone will humble you:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Gate&lt;/strong&gt; exposes &lt;code&gt;funding_rate_limit&lt;/code&gt; in its futures contracts model. The field's meaning is confirmed in Gate's own public API documentation changelog, which added the field "to indicate funding rate cap value."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Bybit&lt;/strong&gt; publishes the same parameter twice under two names: &lt;code&gt;upperFundingRate&lt;/code&gt;/&lt;code&gt;lowerFundingRate&lt;/code&gt; in the instruments-specification endpoint, and &lt;code&gt;fundingCap&lt;/code&gt; in the tickers endpoint. On October 4 both endpoints returned 0.00333 for BTCUSDT — a useful internal cross-check you can reproduce yourself.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;OKX&lt;/strong&gt; calls it &lt;code&gt;maxFundingRate&lt;/code&gt;/&lt;code&gt;minFundingRate&lt;/code&gt; on its funding-rate endpoint.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;KuCoin&lt;/strong&gt; uses &lt;code&gt;fundingRateCap&lt;/code&gt;/&lt;code&gt;fundingRateFloor&lt;/code&gt; — and note the data types differ too: KuCoin returns numbers, while Gate, Bybit, and OKX return strings.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;One concept, four spellings, two data types. There is no standard.&lt;/p&gt;

&lt;h2&gt;
  
  
  The ETH side is where the philosophies split
&lt;/h2&gt;

&lt;p&gt;The BTC caps differ, but the more interesting divergence is on the second-most liquid contract. At the same moment:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;OKX&lt;/strong&gt; sets ETH's cap at ±0.75% — exactly double its own BTC cap of ±0.375%.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;KuCoin&lt;/strong&gt; sets ETH at ±0.375% — exactly double its own BTC cap of ±0.30%.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Gate&lt;/strong&gt; (±0.30%) and &lt;strong&gt;Bybit&lt;/strong&gt; (±0.333%) give ETH the same ceiling as BTC.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;So two venues deliberately price volatility headroom by asset, and two treat majors as one class. Neither approach is right or wrong — but it means "the cap on this exchange" is not a single number you can memorize per venue. It is a matrix: venue × contract.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why the ceiling matters even if you never see it
&lt;/h2&gt;

&lt;p&gt;Most funding intervals settle near zero — a few basis points every eight hours. The caps are invisible in calm markets. They matter in the tails: when a crowded position unwinds, funding can pin at the cap for consecutive intervals, and that pinned value &lt;em&gt;is&lt;/em&gt; your holding cost, compounding every settlement period. A position on the venue with a 0.30% cap faces a structurally different worst-case carrying cost than the same position where the cap is 0.375% — and, as previous measurements in this series have shown, some exchanges settle smaller contracts hourly or every four hours, where the same cap applied more often multiplies the effect.&lt;/p&gt;

&lt;p&gt;Two caveats that deserve repetition: the caps are parameters exchanges can adjust at will, and the interval length is a separate field you must read before comparing any two caps. "0.333% every 8 hours" and "2.5% every 1 hour" are not comparable numbers.&lt;/p&gt;

&lt;h2&gt;
  
  
  How to verify every number in this article
&lt;/h2&gt;

&lt;p&gt;No login or API key is needed for any of it:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Gate: &lt;code&gt;GET https://api.gateio.ws/api/v4/futures/usdt/contracts/BTC_USDT&lt;/code&gt; (and &lt;code&gt;ETH_USDT&lt;/code&gt;) — read &lt;code&gt;funding_rate_limit&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Bybit: &lt;code&gt;GET https://api.bybit.com/v5/market/instruments-info?category=linear&amp;amp;symbol=BTCUSDT&lt;/code&gt; — read &lt;code&gt;upperFundingRate&lt;/code&gt;; cross-check &lt;code&gt;GET https://api.bybit.com/v5/market/tickers?category=linear&amp;amp;symbol=BTCUSDT&lt;/code&gt; — read &lt;code&gt;fundingCap&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;OKX: &lt;code&gt;GET https://www.okx.com/api/v5/public/funding-rate?instId=BTC-USDT-SWAP&lt;/code&gt; (and &lt;code&gt;ETH-USDT-SWAP&lt;/code&gt;) — read &lt;code&gt;maxFundingRate&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;KuCoin: &lt;code&gt;GET https://api-futures.kucoin.com/api/v1/contracts/XBTUSDTM&lt;/code&gt; (and &lt;code&gt;ETHUSDTM&lt;/code&gt;) — read &lt;code&gt;fundingRateCap&lt;/code&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Snapshot taken 2026-10-04 13:37 UTC. If you re-run these calls next week and the numbers moved, that is the exchange exercising the very parameter this article describes — not an error in the measurements.&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;Not financial advice.&lt;/strong&gt; This article describes publicly published contract parameters. It makes no recommendation to trade, hold, or move positions on any venue named. Perpetual futures involve substantial risk of loss. Funding rates, caps, and settlement intervals change without notice.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;#ad — affiliate disclosure:&lt;/strong&gt; no sponsored or tracked links appear in this article; no exchange named here has paid for inclusion. If this publication adds such links in the future, they will be marked the same way, before the link.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Verify official sources:&lt;/strong&gt; always confirm current terms directly on each exchange's official API documentation and fee pages before relying on any number — exchanges revise these parameters at their discretion.&lt;/p&gt;

</description>
      <category>api</category>
      <category>bitcoin</category>
      <category>crypto</category>
    </item>
    <item>
      <title>Your Dollar Is Not a Dollar: USDT Trades at a Measurable Discount Right Now</title>
      <dc:creator>HMwork</dc:creator>
      <pubDate>Fri, 25 Sep 2026 02:13:55 +0000</pubDate>
      <link>https://dev.to/hmwork_2c698b7158d3/your-dollar-is-not-a-dollar-usdt-trades-at-a-measurable-discount-right-now-384g</link>
      <guid>https://dev.to/hmwork_2c698b7158d3/your-dollar-is-not-a-dollar-usdt-trades-at-a-measurable-discount-right-now-384g</guid>
      <description>&lt;h1&gt;
  
  
  Your "Dollar" Is Not a Dollar: USDT Trades at a Measurable Discount Right Now
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;Disclosure: This article was researched and drafted with AI assistance, then checked against primary sources listed at the end. Nothing here is investment advice — this is an observation of publicly available numbers. This publication may carry affiliate links in the future; at the time of writing, this article contains no affiliate or sponsored links (#ad disclosure: none present).&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;If you hold stablecoins, you probably treat one USDT as one dollar. The market disagrees — not dramatically, not dangerously, but consistently. Right now, at the moment of writing, three independent price sources all say the same thing: USDT trades a few basis points under the face value of a US dollar, and the exact discount depends on where you ask.&lt;/p&gt;

&lt;p&gt;This is not a prediction about depegs, solvency, or risk. It is simply a measurement, taken live from public endpoints, of a friction that sits below most people's radar — and next to your fee schedule, it is a real cost line.&lt;/p&gt;

&lt;h2&gt;
  
  
  The measurement: three sources, one minute apart
&lt;/h2&gt;

&lt;p&gt;On &lt;strong&gt;2026-09-25 at 02:07–02:08 UTC&lt;/strong&gt;, we queried three independent public sources within one minute of each other.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;1. Kraken spot market (USDT/USD).&lt;/strong&gt; The official public ticker endpoint returned a bid of &lt;strong&gt;0.99967&lt;/strong&gt; and an ask of &lt;strong&gt;0.99968&lt;/strong&gt;. The midpoint is &lt;strong&gt;0.999675&lt;/strong&gt; — about &lt;strong&gt;3.3 basis points&lt;/strong&gt; below par. The 24-hour range printed on the same response was tight: low 0.99964, high 0.99971.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. Coinbase Exchange (USDT-USD).&lt;/strong&gt; The official public ticker endpoint returned bid &lt;strong&gt;0.99971&lt;/strong&gt; / ask &lt;strong&gt;0.99972&lt;/strong&gt;, a midpoint of &lt;strong&gt;0.999715&lt;/strong&gt; — about &lt;strong&gt;2.9 basis points&lt;/strong&gt; below par. The stats endpoint for the same product showed a 24-hour open of 0.99983 and a low of &lt;strong&gt;0.99875&lt;/strong&gt;: at its worst moment of the day, USDT sold at a &lt;strong&gt;12.5 basis point&lt;/strong&gt; discount on this venue.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;3. CoinGecko (aggregated reference price).&lt;/strong&gt; The simple-price endpoint returned Tether at &lt;strong&gt;$0.999725&lt;/strong&gt; (about 2.75 bps under), alongside &lt;strong&gt;USDC at $0.999809&lt;/strong&gt; and &lt;strong&gt;DAI at $0.999957&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;We repeated the two exchange queries twice each, two seconds apart; the numbers came back identical, so this is not a one-frame fluke of a single request.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the numbers actually say
&lt;/h2&gt;

&lt;p&gt;Three independent sources — two exchange order books and one aggregator — agree on direction and magnitude: &lt;strong&gt;USDT currently changes hands for roughly 0.9997 US dollars&lt;/strong&gt;. The order books are extremely tight (Kraken's bid-ask spread is one ten-thousandth of a dollar wide), so this is not a liquidity artifact. It is simply where the "dollar" that is USDT clears.&lt;/p&gt;

&lt;p&gt;Two subtleties worth naming:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;The discount is small but persistent.&lt;/strong&gt; At the moment of writing it is ~3 bps; earlier the same day, Coinbase's own stats show it briefly reaching ~12.5 bps. Stablecoin discounts breathe with market stress and tend to widen exactly when holders want out.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The discount is not uniform.&lt;/strong&gt; USDC's aggregated price ($0.999809) and DAI's ($0.999957) show smaller haircuts than USDT's ($0.999725) at the same minute. Different "dollars", slightly different discounts.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why this belongs next to the fee table
&lt;/h2&gt;

&lt;p&gt;Most people compare exchanges by maker/taker fees. But if your balance sheet is denominated in USDT and your life is denominated in USD, then every round trip — sell your USDT to withdraw to a bank, for example — starts from a number below 1.0000. Three basis points sounds trivial until you notice it is the &lt;em&gt;same order of magnitude&lt;/em&gt; as the maker fees on the deepest trading pairs. On the same day, the best-published maker tiers at major venues run in the tens of basis points, so the stablecoin discount is a meaningful fraction of a full trading cost stack — and unlike a fee schedule, it is invisible because no invoice ever shows it.&lt;/p&gt;

&lt;p&gt;For context (measured in our earlier work, same kind of public-endpoint snapshots): a USDC/USDT cross also refuses to trade at exactly par, printing about 2–3 bps the other way. In other words, &lt;em&gt;none&lt;/em&gt; of the "dollars" in crypto are interchangeable at exactly 1:1 with each other, or with the real one. They are all just very well-behaved, very slightly discounted claims.&lt;/p&gt;

&lt;h2&gt;
  
  
  What we deliberately do not claim
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;No depeg prediction.&lt;/strong&gt; A 2–5 bps discount is normal, observable behavior; nothing here says anything about future stability of any token.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;No ranking of venues.&lt;/strong&gt; Kraken printing 3.3 bps and Coinbase 2.9 bps is within the noise of different order books at different seconds; it is not evidence that one venue is "cheaper".&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;No snapshot longevity.&lt;/strong&gt; These numbers were true at 02:08 UTC on 2026-09-25. Stablecoin premia move constantly. Re-measure before acting on any of this.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;This is not financial advice.&lt;/strong&gt; It is a reading of public data, offered so that the number 1.0000 stops being an assumption.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Try it yourself
&lt;/h2&gt;

&lt;p&gt;Every number above came from an unauthenticated public API — you can reproduce all of it in a terminal or a browser in under a minute: Kraken's &lt;code&gt;Ticker&lt;/code&gt; endpoint for the pair &lt;code&gt;USDTZUSD&lt;/code&gt;, Coinbase Exchange's &lt;code&gt;/products/USDT-USD/ticker&lt;/code&gt; and &lt;code&gt;/products/USDT-USD/stats&lt;/code&gt;, and CoinGecko's &lt;code&gt;simple/price&lt;/code&gt; endpoint for &lt;code&gt;tether,usd-coin,dai&lt;/code&gt;. Take the midpoint of each bid/ask, subtract from 1.0000, multiply by 10,000, and you have the discount in basis points. If your numbers differ from ours, that is the point: the discount is a live number, not a constant.&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;Verify official sources.&lt;/strong&gt; All figures in this article are quoted from the primary sources themselves, timestamped 2026-09-25 02:07–02:08 UTC, and readers are encouraged to re-fetch them directly:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Kraken public ticker API: &lt;a href="https://api.kraken.com/0/public/Ticker?pair=USDTZUSD" rel="noopener noreferrer"&gt;https://api.kraken.com/0/public/Ticker?pair=USDTZUSD&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Coinbase Exchange public ticker: &lt;a href="https://api.exchange.coinbase.com/products/USDT-USD/ticker" rel="noopener noreferrer"&gt;https://api.exchange.coinbase.com/products/USDT-USD/ticker&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Coinbase Exchange public stats: &lt;a href="https://api.exchange.coinbase.com/products/USDT-USD/stats" rel="noopener noreferrer"&gt;https://api.exchange.coinbase.com/products/USDT-USD/stats&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;CoinGecko public price endpoint: &lt;a href="https://api.coingecko.com/api/v3/simple/price?ids=tether,usd-coin,dai&amp;amp;vs_currencies=usd" rel="noopener noreferrer"&gt;https://api.coingecko.com/api/v3/simple/price?ids=tether,usd-coin,dai&amp;amp;vs_currencies=usd&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;em&gt;AI-assisted content. Market data snapshots are momentary and change constantly. Nothing in this article is investment advice, a recommendation, or a solicitation to trade any asset. No affiliate links are present in this article.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>crypto</category>
      <category>stablecoin</category>
      <category>trading</category>
      <category>beginners</category>
    </item>
  </channel>
</rss>
