DEV Community: ABROWNFOX001

US Crypto Up/Down, Part 1: Polymarket US Just Opened a 15-Minute BTC Book

ABROWNFOX001 — Fri, 02 Oct 2026 12:46:02 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Polymarket US launched Bitcoin Up or Down in late September 2026. On Sep 28 the official account said 15-minute crypto up/downs are live in the US app. Hourly windows are on the same product.

This is not the offshore btc-updown-5m-* book I trade. It is a second venue, a second clock, and a second settlement story.

What shipped

The US contract is simple:

BTC only for now
a fresh price to beat at the start of each window
Up if BTC finishes above that level, Down if it finishes below
15-minute rounds, plus 60-minute rounds
resolution from Chainlink BTC/USD
early cash-out is part of the app flow
24/7, matching BTC’s clock

Offshore short-duration crypto already settles on Chainlink TWAP (30s on 5m, 60s on 15m). The US product is a consumer event contract on the same idea: beat the open, or don’t.

Do not assume the books are the same instrument.

Why this matters for a bot operator

The volume chart from the last series already showed the split. Offshore monthly notional fell from the June peak toward ~$2.9B in the September print. The US exchange crossed above it.

15-minute BTC is how the US app tries to keep that flow.

For @abrownfox001 the implications are narrow:

Offshore (live loop)	Polymarket US (new)
5m BTC core, some 15m	15m and 60m BTC
CLOB v2, public tape	US app, separate book
TWAP vs frozen open	price-to-beat vs close
scratch near 50¢	early cash-out is the exit
self-custody / API	regulated rails

A calibrated 5-minute TWAP read does not automatically become a 15-minute US edge. The window is longer, the exit is a cash-out, and the book is not the one in src/clob.ts.

What this series will cover

Part 1 — what actually launched
Part 2 — 15m vs 5m: why the signal contract changes
Part 3 — price-to-beat, Chainlink, and early cash-out as execution
Part 4 — what not to port from the offshore bot
Part 5 — how I would paper a US 15m loop without touching the live 5m engine

The live account stays on the offshore 5-minute book. The US product is a new market to study, not a rewrite of the frozen loop.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

Polymarket #PolymarketUS #BTC #TradingBot #PredictionMarkets #Chainlink #UpDown

Settlement Math — BRTI Window vs Last Tick vs Offshore TWAP

ABROWNFOX001 — Tue, 29 Sep 2026 07:49:10 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Part 1: the US app launched 15-minute Bitcoin Up/Down.

Part 2 is the only question that matters for a bot:

What number does the venue actually settle on?

If that number is not your model target, you do not have a strategy. You have a candle opinion.

Three closes people confuse

Source	What it is	Legal for which book?
Last CEX print	One trade at expiry	Almost never
Offshore Chainlink TWAP	Official short window on the on-chain crypto binaries	Offshore 5m / 15m
US posted rules	Example US 15m text: last-minute average of 60 BRTI prints, rounded to 2 decimals, at both open and end	That specific US contract

Same “Up or Down” label. Different (Y).

On a posted Polymarket US 15m rule set I checked, settlement is not “was Binance ticking up.” It is closer to:

[
\text{Up} \iff \text{avg}(60 \text{ BRTI prints before } t_{\text{end}}) \ge \text{avg}(60 \text{ BRTI prints before } t_{\text{open}})
]

That is a one-minute window average at two timestamps — CF Benchmarks’ Bitcoin Real-Time Index — not a wick.

Ask-box chatter also points at Chainlink TWAP language for some 15m crypto binaries. Do not assume. Read the market card you are trading. US and offshore cards are allowed to disagree.

Why last tick dies here

A last-tick bot wins on the spike that the average ignores.

A window-average bot dies on the spike that the chat is celebrating.

If BTC wicks +$80 with 8 seconds left and the last-minute mean barely moved, the US contract can still resolve Down. Your screenshot of “price now” is not the judge.

That is the same lesson as the offshore TWAP cutover: train on the settlement object, or your win rate is a backtest against the wrong close.

What `@abrownfox001` is allowed to reuse

Reuse the contract, not the feed:

freeze open_ref from the official family
score live value from the same family
halt when that feed is stale
never invent open from a random Binance candle

Do not reuse:

5-minute slot clock
30-second TWAP topic as if it were BRTI
CLOB v2 order IDs on the US app
“if CEX leads, override the official average”

CEX lead-lag can still be a feature. It cannot be the oracle.

Practical check before any US 15m code

1. Copy the resolution source off the live card
2. Name the open statistic
3. Name the end statistic
4. Name the rounding rule
5. Only then estimate P(up)

If step 1 is “I assumed Chainlink because offshore uses it,” stop.

Part 3: why US 15m flow will look like chase, not like a 50¢ specialist tape.

Not financial advice. Paper or micro-size first. Rules differ by venue and by market card.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

#Polymarket #PolymarketUS #TWAP #BRTI #Bitcoin #TradingBot #PredictionMarkets

15-Minute Bitcoin Markets Are Not the Offshore 5m Bot

ABROWNFOX001 — Tue, 29 Sep 2026 07:44:10 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

On Sept 28, 2026, Polymarket flipped a switch that changes distribution more than it changes microstructure:

15-minute crypto Up/Down is now live in the Polymarket US app.

Americans can now trade Bitcoin’s next candle on a CFTC-regulated venue. Offshore traders have had short-horizon crypto binaries since 2025. This is the US copy of that habit — not a clone of the on-chain 5-minute engine behind @abrownfox001.

What actually launched

Right now the US product is narrow:

Asset: Bitcoin
Window: 15-minute Up/Down (guides also mention 60-minute rounds)
Venue: Polymarket US app / polymarket.us
Question: finish above or below the price-to-beat

It is a consumer loop: pick Up or Down, size it, wait for the candle, collect $1 or $0. Some UI paths also allow an early cash-out. That is a different instrument from a hold-to-redeem specialist bot.

Two books, two settlement stories

Do not paste an offshore 5m strategy onto the US app.

	Offshore on-chain	Polymarket US app
Access	Non-US / on-chain wallet	KYC, regulated US app
Core short window I trade	BTC 5m	BTC 15m (for now)
Book	Public CLOB, public tape	Separate US book
Bot surface	CLOB + RTDS + Data API	App-first, different APIs/docs
Settlement family	Chainlink TWAP (30s on 5m; 60s on 15m)	Official US rules (BRTI-style window average on posted US markets; confirm per contract)

Same question — “up or down?” — is not the same contract.

If you train on Binance last tick and the venue averages the last minute, you built a confident wrong bot. That was the whole point of the TWAP series.

What this means for `@abrownfox001`

I am not porting the live engine to the US app in week one.

The offshore loop stays frozen:

BTC 5m
official TWAP vs frozen open
45–55¢ clips
scratch decay
halt on stale data

The US launch matters as flow and regime, not as a new GitHub module.

More US users on 15m binaries will:

pull attention and sports-adjacent volume onto Polymarket US
create a second BTC-direction price that will not always match the on-chain 5m book
tempt people to copy-trade across venues

Cross-venue copy is how you buy 54¢ on one book for a signal that belongs to the other.

Series map

Part 1 — the launch, and why it is a different instrument
Part 2 — settlement: TWAP / BRTI window math vs last tick
Part 3 — why 15m US flow will look more like chase than 50¢ specialist flow
Part 4 — what a bot is allowed to reuse from the 5m engine
Part 5 — how I will watch both tapes without merging strategies

Short-horizon BTC binaries just became a US retail product. That increases noise. It does not give the 5m engine a new edge.

Not financial advice. Paper or micro-size first. Read the contract rules on the venue you actually trade.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

Polymarket #PolymarketUS #Bitcoin #UpDown #TradingBot #TWAP #PredictionMarkets

Scale the Clip, Not the Product

ABROWNFOX001 — Sun, 27 Sep 2026 11:10:35 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Parts 1–4 froze the book, the 50¢ math, the skip list, and the public-tape tests.

Part 5 is the question that wrecks specialists: “it works, so now what?”

The wrong answer is more markets.

The right answer is a slightly larger clip on the same contract, only after the tape still passes.

What scaling is allowed to mean

Allowed:

raise base clip inside the same $ band
raise per-slot cap slowly
keep BTC 5m, 45–55¢, scratch-on-decay

Not allowed:

ETH/SOL “because correlation”
15m “because more time”
99¢ “because easier”
two slots at full size just because both look good
turning a taker clip into a maker grid to force size

Those are product changes. Product changes reset the measurement clock.

The scale gate

Do not touch size unless all of these were true on the last tape window:

Entries still clustered in 45–55¢
Scratch spread still ~0
Buys still earlier than sells
No double-sends
Halt rules actually fired when the feed was stale
Held-to-resolution math still cleared fees

Fail one gate → size stays.

Fail two → size down, not sideways into a new market.

How size should move

if tape_pass and ev_after_fees_ok:
    clip = min(clip * 1.1, visible_ask * 0.25, slot_cap)
else:
    clip = clip * 0.7
    slot_cap stays or drops

Small steps. The 5m book is thin at the open and thinner when you are the clip.

Taking the whole ask is not scaling. It is becoming the next trader’s worse price — the thing Part 2 said kills 50¢ EV.

Correlation is already maxed

Every ticket is short-horizon BTC. Ten simultaneous slots is not diversification. It is the same impulse ten times.

So the real cap is not “how many markets can the API see.” It is:

max live BTC 5m inventory
max daily loss before HALT
max consecutive error slots before size-down

If the Rust CLOB later improves p99, that may allow a larger clip on the same book. It still does not allow a second product.

The account should look boring as it grows

A scaled specialist is still recognizable:

same wallet
same 5m BTC slug family
same band
slightly larger notionals
same scratch shape

If growth makes the tape look like a different bot, you did not scale. You shipped a rewrite and kept the old name.

That is the whole playbook: one settlement definition, one band, one skip list, one public test, and size that only moves after the test passes.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

#Polymarket #TradingBot #PositionSizing #TWAP #BTC #AlgoTrading #RiskManagement #PredictionMarkets

The Public Tape Is the Unit Test

ABROWNFOX001 — Sun, 27 Sep 2026 11:08:53 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Part 3 said most slots should be skips.

Part 4 is how you check that without the private weights.

@abrownfox001 publishes fills, not SKIP reason=... logs. That is enough. The activity feed is a delayed unit test for the whole stack: TWAP target, 50¢ band, scratch policy, halt rules.

Pull the tape

curl "https://gamma-api.polymarket.com/public-profile?address=0x12b712029efeee7efc4066e98e4e1b6beb8b1b0b"
curl "https://data-api.polymarket.com/trades?user=0x12b712029efeee7efc4066e98e4e1b6beb8b1b0b&limit=1000"

Do not start with P&L. Start with shape.

Pass / fail checks

Test	Pass	Fail
Product	BTC 5m is the book	Sports / spray / random longs
Band	Most entries 45–55¢	Cluster at 60–70¢ or 94–99¢
Timing	Buys earlier, sells later	All size in last 20s
Cull	Same-market buy+sell near 0 spread	Almost no sells, or 3¢ scratch leaks
Size	Boring $50–175 clips	One slot eats the day
Style	Directional specialist	Sudden 99¢ farm

If three of those fail in a week, the code drifted or the gates were bypassed. A green day of P&L does not override a red tape.

What you cannot see — and what that implies

The tape will not show:

skipped slots
TWAP age at decision time
trusted=false
rate-limit warnings that blocked a clip

So a quiet hour can be health (skips working) or death (process down).

That is why local logs still need reason. The public feed tests the trades that escaped. The private log tests the trades that should never have existed.

Use both. Trust neither alone.

How I read a bad week

Did entries walk up in price? → execution chasing, not signal
Did sells vanish? → cull disabled
Did late-slot size explode? → clock rule broken
Did non-BTC share jump? → specialist contract broken
Did scratches start paying a spread? → flatten is now a fee

Fix the first broken test. Do not add ETH 5m to “diversify” a process that already failed its own unit tests.

Part 5: how to scale clip size without changing the bot.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

#Polymarket #TradingBot #TWAP #BTC #AlgoTrading #PredictionMarkets #RiskManagement

Slots You Must Skip

ABROWNFOX001 — Sun, 27 Sep 2026 11:07:08 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Part 2 was the 50¢ identity after fees.

Part 3 is the other half of that math: most 5-minute slots should produce no order.

A directional engine dies when it treats every btc-updown-5m-* print as mandatory. The edge is not “trade 288 slots a day.” The edge is refusing the ones where the contract is broken.

Skip is a first-class output

src/signal.ts is allowed to return side = none and trusted = false.

The engine should log that as success:

SKIP reason=stale_twap
SKIP reason=no_open_ref
SKIP reason=ask_outside_band
SKIP reason=late_slot
SKIP reason=no_edge_after_fees

If your logs are only fills, you cannot see the strategy. You can only see the accidents.

The skip list

Condition	Why the slot is illegal
TWAP age over the freshness cap	You are scoring a snapshot
Reconnect with no new stamped TWAP	Last value is dead
`open_ref` missing or guessed late	Wrong binary
Ask / bid outside 45–55¢	You are paying up for a coin-flip
`P(up)` close to mid after fees	No trade, just turnover
First ~30s, book thin	Starter only or wait
Last ~30s, no position yet	Too late to start a thesis
Rate-limit warning up	You may not be able to scratch
Unacked clip already in flight	Double-send risk
Inventory ≠ last fill	Blind

Two of those in the same slot is not “be careful.” It is NO_NEW_ENTRIES or HALT.

Pretty `P(up)` is not an override

The most expensive skips are the ones that feel cowardly.

61% read at t=12s on a 3-lot ask
58% read at 59¢
57% read after the socket just resumed
64% read at t=286s with no prior position

Those are how a 50¢ engine becomes a chase engine.

The previous series already defined the rule: the signal estimates settlement. The engine decides whether the book will pay you. When the book will not pay you, the correct P(up) still produces SKIP.

What the tape should show

A healthy day has more skipped slots than filled slots.

On the public activity feed you will not see the skips. You will see the residue:

fewer late-bell dumps
fewer 60¢+ entries
scratches still near entry
clip sizes boring

If every slot shows a fill, the skip list is documentation, not code.

Part 4: use the public tape as a unit test for this policy.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

#Polymarket #TradingBot #RiskManagement #TWAP #BTC #AlgoTrading #PredictionMarkets

Edge Math at 50¢ After Fees

ABROWNFOX001 — Sun, 27 Sep 2026 11:05:50 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Part 1 froze the book: BTC, 5 minutes, coin-flip band.

Part 2 is why that band is the whole business.

A 99¢ farmer lives on tiny clips and huge volume.

@abrownfox001 lives on a different equation: buy near 50¢, win more than half of the tickets that actually reach resolution, scratch the rest near flat.

If that math does not clear fees, there is no bot. There is a hobby.

The raw identity

Hold a binary to $1 at entry price (p), with held-side win rate (w):

[
EV = w(1-p) - (1-w)p
]

At (p \approx 0.50):

[
EV \approx 0.50 \times (2w - 1)
]

Held-side win rate	EV / stake at 50¢
50%	0%
55%	+5%
60.7%	+10.7%
65%	+15%

That +10.7% is the schematic from the public resolved sample — before scratches and before fees.

Scratches are how the live book can look better than raw (w): losers get flattened near entry instead of going to $0. That only helps if scratch spread stays ~0.

Fees change the threshold

On short-dated crypto, taker fees are not a rounding error once you stop pretending every fill is free.

Write the real hurdle as:

[
w_{\min} \approx \frac{p + f}{1 + f_{\text{net}}}
]

You do not need a perfect fee model to use the rule:

at 50¢, a 1% round-trip leak already eats a chunk of a 55% read
at 99¢, the same leak can erase the whole clip
scratches that leak 2–3¢ are just fees with extra steps

This is why the engine refuses 57–62¢ “favorites.”

You paid up, so (p) rose, so (w) has to rise with it. Most short-horizon reads are not that good.

What a legal trade looks like

trusted P(up)
ask still inside 45–55¢
edge after fees >= threshold
clip <= slot cap
else skip

P(up) = 0.56 vs a 56¢ ask is not edge.

P(up) = 0.56 vs a 49¢ ask might be.

The signal estimates settlement.

The book decides whether you get paid to say it.

Why I do not farm 99¢

Convergence EV is small by design. Fees, rebates, and queue position become the strategy.

That is a different account:

massive volume
almost no sells
tiny per-ticket edge
rate-limit budget spent on sweep size

This account spends the budget on one fair clip and one clean scratch. Mixing the two styles on one wallet makes the public tape unreadable and the fee drag incoherent.

The only number that matters this week

Not one-day P&L.

After fees and scratches, did held-to-resolution tickets still clear the 50¢ identity?

If yes, keep the loop.

If no, the problem is usually fill price or scratch leak — not “we need more markets.”

Part 3: slots you must skip, even when (P(up)) looks pretty.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

#Polymarket #TradingBot #EV #TWAP #BTC #AlgoTrading #PredictionMarkets

Why @abrownfox001 Stays on BTC 5m

ABROWNFOX001 — Sun, 27 Sep 2026 09:57:55 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

The first series was the engine.

The second was the venue (Rust CLOB, indexer, Q4 rails).

This series is the strategy choice that sits on top of both: stay a specialist.

@abrownfox001 still trades one object. BTC. 5 minutes. Coin-flip band. Official TWAP vs frozen open. Scratch the weak tickets.

That looks conservative while the platform is splitting into an offshore book and a US exchange. It is the point.

The temptation after every venue change

When volume migrates, bots usually expand:

add ETH/SOL/XRP 5m
add 15m
add sports
add 99¢ convergence
add copy-trading

Each add looks cheap. Each add is a new clock, a new book, a new failure mode.

A 5m BTC engine already has four hard dependencies:

slot clock
official TWAP
45–55¢ liquidity
cancel/scratch tails

Sports does not use that TWAP contract.

15m uses a different window.

99¢ farming is a different P&L machine.

The US book is a different venue entirely.

If you bolt those on during a rails season, you do not have one bot. You have four half-built bots sharing a wallet.

Why 5m BTC is still the right book

Short-dated crypto is one of the few places the on-chain venue still has a native edge:

24/7 clock
public activity tape
TWAP settlement you can code against
enough 50¢ depth for small clips
a loop that can halt cleanly

The US exchange can win sports and fiat users.

Convergence farmers can win on other people’s near-certain leftovers.

This account is trying to turn a market people treat as 50/50 into a filtered ~60/40 book at fair entry. That only works if the instrument stays the same long enough to measure.

Change the instrument every week and you cannot tell whether the signal, the feed, or the venue broke.

What specialization buys you

Specialist	Multi-market spray
One `open_ref` rule	One rule per product
One freshness gate	“fresh” means something else on each feed
Scratch policy you can audit	Mixed exits you cannot read on the tape
Public profile looks like one process	Tape turns into noise

The public check is still the same:

mostly BTC 5m
mostly 45–55¢
sells later than buys
many near-breakeven scratches
no sudden 99¢ farm

If that shape holds, the strategy is intact.

If it drifts, the “new markets” were not diversification. They were a rewrite.

This series

Part 1 — why the book stays BTC 5m
Part 2 — edge math at ~50¢ after fees
Part 3 — slots you must skip
Part 4 — using the public tape as a unit test
Part 5 — how to scale without becoming a different bot

The calibrated weights stay private. The constraint does not: one market family, one settlement definition, one halt policy.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

Polymarket #TradingBot #BTC #TWAP #AlgoTrading #PredictionMarkets #QuantTrading

Audit the Public Tape Before You Touch the Model

ABROWNFOX001 — Fri, 25 Sep 2026 23:02:51 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Parts 1–4 were the specialist contract: one book, 50¢ entries, scratch the weak tickets, wait when the clock or the ask says no.

Part 5 is how you know the live process still matches that contract.

Do not start by retuning P(up). Start with the tape everyone else can already see.

The audit commands

Wallet: 0x12b712029efeee7efc4066e98e4e1b6beb8b1b0b

curl "https://gamma-api.polymarket.com/public-profile?address=0x12b712029efeee7efc4066e98e4e1b6beb8b1b0b"
curl "https://data-api.polymarket.com/trades?user=0x12b712029efeee7efc4066e98e4e1b6beb8b1b0b&limit=1000"

Also just read: https://polymarket.com/@abrownfox001?tab=activity

P&L for one afternoon is not the test. Shape is the test.

Healthy shape

BTC 5m is still the core
most fills still in 45–55¢
buys exist without a sell on every market — some tickets redeem
sells exist — scratches are real
sells land later than buys
clip sizes stay boring
same-market round trips sit near a zero spread
no sudden 99¢ farm, no sports spray, no close-bell dump

That shape is the public proof of Parts 1–4.

Unhealthy shape

entries walking to 60–70¢
almost no sells (hold-everything)
all size in the last 20 seconds
one slot many times normal clip size
scratch spreads of 2–4¢ as a habit
random non-BTC volume becoming the book

Any of those means the process broke. Changing signal weights will not fix a missing price cap or a dead scratch rule.

What to change when the tape breaks

Tape	Fix
Prices too high	restore the 45–55¢ gate
No sells	restore decay → scratch
Late-only size	restore the slot clock
Double clips	fix `intent_id` / fill reconcile
Random tickers	delete the extra markets
Stale-looking bursts	halt-on-stale, not a new model

The private layer is the TWAP → P(up) map.

The public layer is the only layer you should debug first.

Series close

A specialist Polymarket bot is not a dashboard with twenty markets.

It is one sentence you can still defend after a bad day:

TWAP-anchored 5m BTC direction, bought near 50¢, scratched when belief dies, skipped when the clock or the book is unclean.

If the public tape still says that, leave the model alone.

If it does not, do not add features. Put the gates back.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

#Polymarket #TradingBot #TWAP #BTC #AlgoTrading #RiskManagement #PredictionMarkets

The Slot Clock — Waiting Is a Trade

ABROWNFOX001 — Fri, 25 Sep 2026 23:01:01 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Part 3 was the scratch.

Part 4 is the other unpaid action that keeps a 5-minute book alive: do nothing.

A BTC Up/Down slot is 300 seconds. Most of those seconds should not contain an order.

The clock is part of the signal

The same P(up) = 0.56 is not the same object at t=20s and t=280s.

Early, the 30-second TWAP can still flip. The open is frozen, the window is not.

Late, the official path is already committing and the fair 50¢ size is usually gone.

So the engine does not only ask “is there edge?”

It asks “is this second allowed to express it?”

Four windows

Time in slot	Default action
0–30s	Discover slot, freeze `open_ref`, starter only if book is already fair
30–150s	Main accumulation inside 45–55¢
150–270s	Re-score. Add less. Scratch more.
270–300s	No new thesis. Hold or flatten. Redeem path.

Public tape on this wallet matches that clock: buys median ~47s, sells median ~226s. That is not a flourish. It is the schedule.

When waiting is the trade

Wait if any of these are true:

TWAP is stale
open_ref was not captured cleanly
ask is outside the band
visible size cannot fill a min clip without walking the book
an intent_id is still in flight
rate-limit headers say the next cancel might not land
t is past the no-entry cutoff

src/signal.ts may still publish a number.

src/engine.ts can answer none.

A skipped slot is a successful slot. Missing a 5-minute BTC print is cheap. Paying 61¢ because you were bored is not.

Close-bell entries are a process break

Last-20-second size looks brave on a heatmap. On TWAP settlement it is usually:

worse price
worse scratch
higher chance the model is just reading the book back to itself

The close is for resolution, not for discovering religion.

One rule

If you cannot write the current second’s legal actions on one line, you are improvising.

t=41  trusted + ask 0.51 → clip
t=41  trusted + ask 0.59 → wait
t=188 decayed + bid 0.505 → scratch
t=282 any new buy → forbidden

The specialist book is a clock with a price cap, not a stream of opinions.

Part 5: audit the public tape before you touch the model.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

#Polymarket #TradingBot #TWAP #BTC #AlgoTrading #Execution #PredictionMarkets

The Scratch Is the Strategy

ABROWNFOX001 — Fri, 25 Sep 2026 22:58:30 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Part 2 was the 45–55¢ cap.

Part 3 is why this account sells at all.

A hold-everything 5-minute bot is easy to write. It is also how a decent P(up) becomes a 50/50 P&L. The resolved win rate on @abrownfox001 is a filtered book, not a magic forecast.

What a scratch is

A scratch is a same-market sell near the entry price when belief died, not when the mark-to-market flickered.

In public samples this shows up as hundreds of buy+sell pairs with spreads like 0.504 in / 0.503 out.

That is not profit-taking.

That is culling a ticket so it never pays $0.

Why resolved WR can beat raw accuracy

Raw signal: sometimes right, often noisy.

Engine:

trusted + fair ask → buy clip

re-score 10–30s

belief holds → carry to redeem

belief decays → flatten near entry

feed stale → no add, no panic dump

Only the tickets that stay trusted reach $1 / $0.

Losers that would have gone to zero get turned into almost-flat exits while the 5m book is still deep near 50¢.

That is the whole economic trick. Entry EV at 50¢ assumes you hold. Scratches change the mix of what you hold.

What a scratch is not

Not this	Why
Selling +8¢ “winners”	Different strategy
Flattening every 48¢ dip	Noise-stop, not conviction cull
Panic sell on stale TWAP	Blindness is a halt
Crossing a 3–4¢ hole by habit	Cull is no longer free

If scratch slippage becomes a habit, the “≈ 0 spread” tape is gone and the win-rate story stops being true.

Timing

Buys earlier (median ~47s).

Sells later (median ~226s).

That clock is the re-score. Early in the slot, TWAP can still flip. Later, a compressed P(up) is information. Flatten then, not at t=8s because the token ticked.

One line

The entry finds the fair 50¢ expression.

The scratch decides whether that expression is allowed to meet settlement.

Without the second step, a specialist 5m bot is just another hold-and-hope wallet with extra logs.

Part 4: the slot clock — when waiting is the trade.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

Polymarket #TradingBot #TWAP #RiskManagement #BTC #AlgoTrading #PredictionMarkets

Fair Value Beats Buying the Favorite

ABROWNFOX001 — Fri, 25 Sep 2026 22:54:52 +0000

Polymarket: https://polymarket.com/@abrownfox001?tab=activity

GitHub: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

Telegram: https://t.me/abrownfox001

Part 1 said run one book.

Part 2 is the price band that makes that book viable.

Most 5-minute bots do not die from a bad BTC call. They die from paying 58–65¢ for a thought that was only worth expressing at 50¢.

Two different games

Style	Zone	Edge
Convergence farm	94–99¢	tiny clip × huge volume
Favorite chase	58–70¢	“it already looks right”
This account	45–55¢	calibrated direction × fair entry

@abrownfox001 is the third game.

At (p \approx 0.50), held-to-resolution EV is roughly:

[
EV \approx 0.50 \times (2w - 1)
]

A 60% held-side win rate at 50¢ is about +10% of stake before scratches.

The same 60% read bought at 62¢ is close to noise after fees and a few bad scratches.

Direction without a price cap is not a strategy. It is a mood.

Favorite price is not confirmation

When Up is 61¢, the book is already saying Up is more likely. That is not extra information for a TWAP model. That is the crowd mid.

Legal question:

P(up) vs ask
not
P(up) vs 0.50 ignoring the ask

P(up) = 0.58 against a 58¢ ask is no trade.

P(up) = 0.58 against a 49¢ ask can be a clip.

If the fair zone is gone, wait. The slot still has minutes. Chasing turns a specialist into a late taker on someone else’s conviction.

How the tape should look

Public fills on this wallet cluster in 45–55¢ on purpose.

early seconds: book often thin → wait
30–150s: main accumulation if ask stays fair
after that: add less, scratch more
last 30s: no new thesis

A sudden pile of 60¢+ takes means the cap was removed. That is a process break, not “more confidence.”

One rule

No trusted signal may lift a price outside the band.

The model can still print P(up).

The engine does not have to obey it.

Fair value is how a 5m specialist stays solvent while generalist bots pay for the favorite and call it momentum.

Part 3: why the scratch — not the entry — is what makes the resolved book look better than raw accuracy.

Not financial advice. Paper or micro-size first.

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity


**Relevant Hashtags/Tags:**

Polymarket #TradingBot #TWAP #BTC #FairValue #AlgoTrading #PredictionMarkets

DEV Community: ABROWNFOX001

US Crypto Up/Down, Part 1: Polymarket US Just Opened a 15-Minute BTC Book

What shipped

Why this matters for a bot operator

What this series will cover

Polymarket #PolymarketUS #BTC #TradingBot #PredictionMarkets #Chainlink #UpDown

Settlement Math — BRTI Window vs Last Tick vs Offshore TWAP

Three closes people confuse

Why last tick dies here

What @abrownfox001 is allowed to reuse

Practical check before any US 15m code

15-Minute Bitcoin Markets Are Not the Offshore 5m Bot

What actually launched

Two books, two settlement stories

What this means for @abrownfox001

Series map

Polymarket #PolymarketUS #Bitcoin #UpDown #TradingBot #TWAP #PredictionMarkets

Scale the Clip, Not the Product

What scaling is allowed to mean

The scale gate

How size should move

Correlation is already maxed

The account should look boring as it grows

The Public Tape Is the Unit Test

Pull the tape

Pass / fail checks

What you cannot see — and what that implies

How I read a bad week

Slots You Must Skip

Skip is a first-class output

The skip list

Pretty P(up) is not an override

What the tape should show

Edge Math at 50¢ After Fees

The raw identity

Fees change the threshold

What a legal trade looks like

Why I do not farm 99¢

The only number that matters this week

Why @abrownfox001 Stays on BTC 5m

The temptation after every venue change

Why 5m BTC is still the right book

What specialization buys you

This series

Polymarket #TradingBot #BTC #TWAP #AlgoTrading #PredictionMarkets #QuantTrading

Audit the Public Tape Before You Touch the Model

The audit commands

Healthy shape

Unhealthy shape

What to change when the tape breaks

Series close

The Slot Clock — Waiting Is a Trade

The clock is part of the signal

Four windows

When waiting is the trade

Close-bell entries are a process break

One rule

The Scratch Is the Strategy

What a scratch is

Why resolved WR can beat raw accuracy

What a scratch is not

Timing

One line

Polymarket #TradingBot #TWAP #RiskManagement #BTC #AlgoTrading #PredictionMarkets

Fair Value Beats Buying the Favorite

Two different games

Favorite price is not confirmation

How the tape should look

One rule

Polymarket #TradingBot #TWAP #BTC #FairValue #AlgoTrading #PredictionMarkets

What `@abrownfox001` is allowed to reuse

What this means for `@abrownfox001`

Pretty `P(up)` is not an override