DEV Community: Jaroslav Šmarda

“The Market” Isn't One Thing: Watching Sectors Take Turns

Jaroslav Šmarda — Tue, 29 Sep 2026 11:03:51 +0000

Python for programmers, prompts for data analysts.

"The market was up 20% this year" is one of the most compressed sentences in finance. It blends technology companies, oil producers,
banks, utilities, and drugmakers into a single number — as if they all moved together.

They don't. Some years one corner of the market carries everything; other years that same corner drags. Traders call the pattern "sector
rotation." Let's pull the eleven sectors apart and watch who led, who lagged, and whether last year's winner tells you anything about
next year's.

Want to try this yourself? The EODHD MCP Server is what
pulled the data above — connect it to Claude and ask your own questions (affiliate link).

New here? The Setup covers
everything you need before your first prompt.

Pull the eleven sectors apart

🟧 PROMPT

Use the eodhd MCP to get yearly returns for the
eleven SPDR sector ETFs from 2021 to today, and
show the best and worst sector each year.

Best and worst sector, year by year:

2021 — Energy (XLE) +53.3% · Communication (XLC) +16.0%

2022 — Energy (XLE) +64.3% · Communication (XLC) −37.6%

2023 — Technology (XLK) +56.0% · Utilities (XLU) −7.2%

2024 — Communication (XLC) +34.7% · Materials (XLB) +0.2%

2025 — Technology (XLK) +24.6% · Consumer Staples (XLP) +1.5%

2026 so far — Energy (XLE) +42.8% · Consumer Discretionary (XLY) −7.0%

Look at 2022 alone: the best sector gained 64%, the worst lost 38%. That's a gap of more than 100 percentage points, within the same
"market" and the same twelve months. Any single number describing "how the market did" that year was an average of two completely
different experiences.

Does last year's winner tell you anything?

🟧 PROMPT

For each year, where did the previous year's best
and worst sector finish the following year?

Last year's best sector, ranked the next year (out of 11):

Energy, best in 2021 → #1 in 2022

Energy, best in 2022 → #9 in 2023

Technology, best in 2023 → #5 in 2024

Communication, best in 2024 → #2 in 2025

Technology, best in 2025 → #2 in 2026 so far

Last year's worst sector, ranked the next year:

Communication, worst in 2021 → #11 in 2022 (worst again)

Communication, worst in 2022 → #2 in 2023, up 52.8%

Utilities, worst in 2023 → #4 in 2024

Materials, worst in 2024 → #7 in 2025

Consumer Staples, worst in 2025 → #6 in 2026 so far

There's no clean rule here. Energy won two years running, then fell to ninth. Communication finished last two years in a row, then
jumped to second with a 53% gain. Across these five handoffs, last year's winners averaged roughly fourth place the following year, and
last year's losers averaged sixth, which is about where a random draw would put them.

Five handoffs is far too few to call that a pattern. What the table shows clearly is how fast the order reshuffles, and how badly "buy
whatever led last year" or "avoid whatever lagged" could have gone in the wrong year. Chasing 2022's winner meant riding Energy into a
flat 2023. Abandoning 2022's loser meant missing Communication's rebound.

A word of honesty

A few things about this comparison deserve naming.

Calendar years are an arbitrary way to slice the data. Measure the same eleven sectors from July to July, or quarter by quarter, and the
winners and losers would shuffle differently — some of 2022's 100-point gap between Energy and Communication came from where January 1st
happened to fall. The pattern of sectors taking turns holds up; the exact rankings depend on where you draw the lines.

The window is also short: under six years, five year-to-year handoffs. That's enough to show how much the order moves, and nowhere near
enough to prove whether it moves randomly or follows a cycle. People who study sector rotation seriously look at decades of data,
alongside interest rates and the business cycle, and still argue about how predictable it is.

And a sector ETF is less diversified than its name suggests. XLK, XLC, and XLY are each dominated by a handful of giant companies, so
"Technology had a great year" can mostly mean "two or three stocks had a great year." Treat each line in the table as the story of a few
heavyweights, not of an entire industry. Not investment advice.

Final Thoughts

This is the thirty-second article in the series Unlock Real-Time Market Intelligence with EODHD and Claude, and it takes apart a number every other article
has quietly leaned on: "the market."

Two prompts split one index into eleven sectors, then checked whether the leaders stayed in front.

So here's what they found: in 2022, the best sector gained 64% and the worst lost 38% — a gap of more than 100 percentage points
inside the same market, in the same year. And the sector that finished last that year finished second the next. "The market" is an
average of very different stories, and the order of those stories keeps reshuffling. Knowing which one led last year tells you
surprisingly little about which one leads next.

If this made you curious, the MCP Server is free to
try — The Setup walks you through
it.

← Part 31: Before You Panic About Insider Selling, Check the Code

Before You Panic About Insider Selling, Check the Code

Jaroslav Šmarda — Fri, 25 Sep 2026 11:19:10 +0000

Python for programmers, prompts for data analysts.

Every time a company's CEO or CFO sells a big chunk of stock, someone posts about it — usually with a headline implying the insider
knows something bad is coming. SEC Form 4 filings, the disclosures that trigger those headlines, are public and easy to pull. Reading
them correctly is a different matter.

Let's pull real Form 4 filings and see what the raw numbers actually say, once you check the fine print.

Want to try this yourself? The EODHD MCP Server is what
pulled the data above — connect it to Claude and ask your own questions (affiliate link).

New here? The Setup covers
everything you need before your first prompt.

Pull the raw filings

🟧 PROMPT

Use the eodhd MCP to get Amazon's most recent
Form 4 insider filings.

The last ten filings include: CEO Andy Jassy sold $5,180,213 worth of Amazon stock on August 21, 2026, across four separate
transactions. Executive Chair Jeff Bezos disposed of 415,580 shares over two filings on August 25–26. AWS CEO Matt Garman, CFO Brian
Olsavsky, and three other senior executives all show sales in the same window, several exceeding $1 million each.

Read that paragraph alone, and it sounds like the entire leadership team is heading for the exits at once — millions of dollars in
stock, sold within days of each other, right around the same price level.

Ask what the transaction codes actually mean

🟧 PROMPT

Check the transaction codes and footnotes on those
filings — were these discretionary decisions?

Every single "sell" transaction in that batch — Jassy's, Garman's, Olsavsky's, Reynolds's, Zapolsky's, Herrington's — carries
transaction code "S" alongside a footnote citing a Rule 10b5-1 trading plan, each one adopted months before the sale: Jassy's plan dates
to November 14, 2025; Garman's to May 4, 2026. These aren't decisions made the week of the sale. They're pre-scheduled disposals, set up
long in advance specifically so they can't be timed around inside information.

Bezos's 415,580 "disposed" shares carry an entirely different code: "G," for gift. Price per share: $0. Total value: $0. He didn't
sell anything — he donated the shares to non-profit organizations.

Every alarming number in the first prompt turned out to be something other than what it looked like. Not one of the ten filings in this
sample was a same-week, no-notice, "I know something you don't" sale.

A word of honesty

This sample — ten filings, one company, one narrow window — doesn't prove that insider selling never signals anything.

Rule 10b5-1 plans exist precisely because some insider selling would otherwise be suspicious, or at least look that way; they're a
legal shield, not a guarantee of pure intent. An executive can still adopt, modify, or cancel a plan when they choose to — and a plan
set up right before bad news becomes public knowledge has drawn real regulatory scrutiny in other cases, just not in the filings checked
here. The absence of same-week discretionary selling in this batch is a fact about this batch, not a law of insider trading.

It's also worth noting what this data doesn't answer: whether the volume of scheduled selling itself means anything — a plan selling 5,
000 shares a month reads very differently from one liquidating a large stake all at once, even though both would carry the identical
10b5-1 footnote. Reading the code correctly is the first step, not the whole analysis. Not investment advice.

Final Thoughts

This is the thirty-first article in the series Unlock Real-Time Market Intelligence with EODHD and Claude, and it closes the loop this series opened with
SEC EDGAR back in Part 18: the raw filing has the truth, but only if you read past the headline number.

Two prompts pulled the alarming version first, then checked what the codes actually said.

So here's what they found: every one of the ten Form 4 filings checked — millions of dollars in apparent selling — turned out to be
either a Rule 10b5-1 plan set up months in advance, or, in Bezos's case, a charitable gift worth $0 in proceeds. None of it was a
same-week decision made on fresh information. The filing told the truth the whole time. The headline just skipped the footnote.

If this made you curious, the MCP Server is free to
try — The Setup walks you through
it.

← Part 30: The Volatility Crush: What Happens to IV the Day After Earnings

The Volatility Crush: What Happens to IV the Day After Earnings

Jaroslav Šmarda — Fri, 25 Sep 2026 11:09:58 +0000

Python for programmers, prompts for data analysts.

Implied volatility isn't constant — Part 15 already showed it shifts across strikes on the same day. It also shifts across time, and
nowhere more dramatically than around a single, predictable event: earnings.

Before a company reports, nobody knows what the numbers will say, and options pricing reflects that uncertainty with elevated IV. The
moment the report lands, the uncertainty resolves — whatever the numbers are, they're no longer unknown — and IV tends to collapse,
often within hours. Traders call it the "volatility crush." Let's find a real one.

Want to try this yourself? The EODHD MCP Server is
what pulled the data above — connect it to Claude and ask your own questions (affiliate link).

New here? The Setup covers
everything you need before your first prompt.

Find a real earnings crush

🟧 PROMPT

Use the eodhd MCP to find AAPL's most recent
earnings date, then get implied volatility for the
near-the-money call before and after the report.

AAPL reported on July 30, 2026 (after market close) — EPS $2.02 vs. a $1.88 estimate, a 7.4% positive surprise. The near-the-money
$335 call, expiring August 21:

July 29 (day before): IV 30.72%

July 31 (day after): IV 27.73%

A drop of 3 percentage points — about a 10% relative decline — in a single trading day, on an option that didn't even expire for
another three weeks.

The earnings beat was good news. AAPL's stock fell 7.4% the next day anyway — a move already covered back in Part 11 of this series. IV
fell too, despite that sharp move, because it isn't pricing "will this go up or down." It's pricing "how much don't we know yet" — and
the moment the report landed, that uncertainty was gone, whichever direction the stock went.

Is 10% a big crush, or a small one?

🟧 PROMPT

Is a 10% relative drop in implied volatility a
dramatic earnings crush, or a mild one?

It's on the mild end. The sharpest volatility crushes usually show up in options expiring within days of the report — weekly
contracts where the earnings-day uncertainty makes up most of their remaining time value, so resolving it can cut IV by 30% or more in a
single session. This $335 call had 20 days left even after the report — three more weeks of ordinary market uncertainty still baked into
the price, diluting how much of its IV was ever earnings-specific to begin with.

The lesson isn't "the crush didn't happen." It's that how much crush you see depends on how much of an option's remaining life the event
actually accounted for — a fact this series has now touched twice: Part 15 showed IV varies by strike; this shows it varies by how close
expiration sits to the event you're pricing for.

A word of honesty

Two things make this one measurement messier than the clean number suggests.

The $335 strike started exactly at-the-money on July 29, but AAPL's 7.4% drop pushed the stock away from it — by August 3, that same
strike was roughly 10% out-of-the-money. Part 15 already showed IV varies by moneyness, not just by time. So some of the 3-point decline
is genuine earnings crush, and some of it is just this contract drifting into a lower-IV part of the skew as the stock moved beneath it.
This data can't cleanly separate the two effects from each other.

And this is one earnings report, for one company, with one particular kind of surprise (a beat, followed by a drop for unrelated
reasons). A report that missed estimates, or one where price and expectations moved together, could show a different-shaped crush
entirely. Treat the mechanism — uncertainty resolves, so IV falls — as reliable; treat any single measured number, including this one,
as a snapshot of how it played out this one time. Not investment advice.

Final Thoughts

This is the thirtieth article in the series Unlock Real-Time Market Intelligence with EODHD and Claude, and it revisits a stock this series has met
before. Part 11 watched AAPL's options amplify a 7.4% price drop. This time, the same event, seen through implied volatility instead of
price.

Two prompts found the crush, then sized it honestly.

So here's where it lands: the near-the-money AAPL call lost 3 points of implied volatility — about 10% — in the single session after
earnings, despite the stock itself moving sharply. That's a real but modest crush, smaller than a weekly option would have shown,
because three weeks of ordinary uncertainty were still baked into the price alongside the earnings-specific kind. IV didn't fall because
the news was good or bad. It fell because, either way, nobody had to wonder anymore.

If this made you curious, the MCP Server is
free to try — The Setup walks you
through it.

One Number, Two Meanings: What Beta Actually Measures

Jaroslav Šmarda — Fri, 25 Sep 2026 10:50:55 +0000

Python for programmers, prompts for data analysts.

Beta is one of the most quoted numbers in finance — "this stock has a beta of 1.5" gets said constantly, usually as shorthand for
"risky" or "volatile." But beta isn't actually a measure of how much a stock moves. It's a measure of how much a stock moves relative
to the market — and that distinction turns out to matter more than the number itself.

Let's ask for a stock's beta, and then ask what it would take to prove that number means what everyone assumes it means.

Want to try this yourself? The EODHD MCP Server is what pulled
the data above — connect it to Claude and ask your own questions (affiliate link).

New here? The Setup covers
everything you need before your first prompt.

Ask for the number, then build it yourself

🟧 PROMPT

Use the eodhd MCP to get AMZN's beta, then
calculate it independently from five years
of daily returns against SPY.

EODHD's reported beta: 1.443. Calculating it independently from 1,254 trading days of AMZN and SPY daily returns: 1.491.

The two numbers are close — both say AMZN moves roughly 1.4 to 1.5 times as much as the market — but they aren't identical, even
though they're trying to measure the same thing.

Beta of 1.45-ish means: on days the market moves 1%, AMZN has historically moved about 1.45% in the same direction. That's the headline.
The gap between 1.443 and 1.491 is where the interesting part starts.

Change the window, get a different answer

🟧 PROMPT

Recalculate beta using 1 year, 2 years, and
5 years of daily returns, and compare.

1-year beta: 1.432

2-year beta: 1.353

5-year beta: 1.491

EODHD's reported figure: 1.443

Four numbers, all claiming to answer "what's AMZN's beta?", ranging from 1.353 to 1.491 — a spread of nearly 14%. None of them are
wrong. Beta isn't a fixed property of a stock, like its ticker or its founding date. It's a statistical estimate, and every estimate
depends on choices someone made: which benchmark counts as "the market," how far back to look, and how often to sample returns. Change
any of those choices, and the "same" number moves.

That's the second meaning hiding behind the first. Beta answers "how sensitive is this stock to the market" — but only after you've
quietly answered a second question first: sensitive according to which window, against which benchmark. Skip that second question, and
the first one has no single right answer.

A word of honesty

Even after picking a window and a benchmark, beta has real limits worth naming.

It's backward-looking by construction — every number above describes how AMZN has moved relative to the market, not how it will. A
company's risk profile can shift for reasons beta can't see coming: a new business line, a management change, a shift in what investors
expect from the stock next. The 5-year beta of 1.491 already blends together very different chapters of Amazon's story, smoothed into
one coefficient that no single day of that history actually looked like.

Beta also assumes a straightforward linear relationship — that AMZN moves proportionally with the market in both directions, up and down
alike. In practice, many stocks move more sharply on the downside than the upside (or vice versa), a pattern a single beta number
averages away rather than reveals. If you want to know how a stock behaves specifically during market crashes, beta calculated across
all conditions isn't answering that question, even though it's often used as if it does. Not investment advice.

Final Thoughts

This is the twenty-eighth article in the series Unlock Real-Time Market Intelligence with EODHD and Claude, continuing where Part 26 left off — from "do two
things move together" to "how much does one thing move relative to another."

Two prompts pulled one reported number, then rebuilt it four different ways.

So here's where it lands: AMZN's beta ranged from 1.353 to 1.491 depending only on which window you chose — a 14% spread hiding behind
a statistic most people quote as if it were a single, fixed fact. Beta isn't wrong to trust. It's incomplete to trust without asking
which beta, over what period, against what benchmark. The number was never the whole answer. The choices behind it were.

If this made you curious, the MCP Server is free to try
— The Setup walks you through it.

← Part 26: Two Lines That Move Together, For No Reason At All

Two Lines That Move Together, For No Reason At All

Jaroslav Šmarda — Fri, 25 Sep 2026 09:51:24 +0000

Python for programmers, prompts for data analysts.

A high correlation number looks like proof of something. Two lines on a chart, rising and falling together — surely that means one
drives the other, or at least that something real connects them.

Sometimes it does. Sometimes it's a coincidence dressed up as a statistic — two numbers that happened to trend the same way over the
same stretch of time, with no relationship between them at all. Let's find a pair like that, and see what correlation alone can't tell
us.

Want to try this yourself? The EODHD MCP Server is what pulled
the data above — connect it to Claude and ask your own questions (affiliate link).

New here? The Setup covers
everything you need before your first prompt.

Find two things that shouldn't be related

🟧 PROMPT

Use the eodhd MCP to compare AMZN's monthly price
against global sugar prices from May 2023 to
January 2026, and calculate the correlation.

Correlation coefficient: −0.887. Over those 33 months, as sugar prices fell, Amazon's stock rose — and it did so with a consistency
that would usually be read as a strong, meaningful relationship. On a scale where 0 means no relationship and ±1 means perfect lockstep,
−0.887 is close to as tight as real-world financial data ever gets.

There is no supply chain, no economic mechanism, and no plausible causal story connecting the world sugar market to Amazon's stock
price.

A correlation this strong, between two things with nothing to do with each other, is exactly the trap this article is about.

Zoom out and watch it fall apart

🟧 PROMPT

Now calculate the same correlation over the full
period, January 2021 to July 2026, instead of
just that one window.

Over the full 67 months, the correlation between AMZN and sugar prices drops to −0.579 — still negative, but noticeably weaker than
the −0.887 we found by picking the 33-month stretch that happened to line up best.

That's not a data error. It's what happens when you search enough time windows: somewhere in almost any two unrelated series, a
stretch exists where they happen to move together more tightly than chance alone would suggest. The tighter number wasn't a discovery —
it was the result of looking until something matched.

This is the trap correlation-hunting sets for itself. Scan enough starting points, enough window lengths, enough pairs of tickers, and
you will find an impressive number eventually — not because you found a relationship, but because with enough attempts, coincidence
starts looking exactly like one.

A word of honesty

A correlation number, however strong, never tells you why two things moved together — and that's the whole problem with stopping here.

A more rigorous test exists: cointegration, which checks not just whether two series move together over one window, but whether they
share a stable, long-run equilibrium relationship that persists across time and would be expected to hold up out of sample. AMZN and
sugar prices would almost certainly fail that test — there's no economic reason to expect one to anchor the other. Correlation asks "did
these move together, this one time?" Cointegration asks a harder, more honest question: "is there a real, structural reason to expect
they'll keep doing it?"

The deeper lesson: any time a correlation is discovered by scanning many windows, many pairs, or many parameters — rather than proposed
first and tested once — treat it as suspect by default. The sugar-and-Amazon relationship above wasn't found because someone
hypothesized a link and checked. It was found because the exact opposite happened: we looked for a coincidence, and finance has an
almost unlimited supply of them. Not investment advice.

Final Thoughts

This is the twenty-sixth article in the series Unlock Real-Time Market Intelligence with EODHD and Claude, and the first one that manufactured its own trap
on purpose — deliberately hunting for a coincidence to show how easy coincidences are to find.

Two prompts built the illusion, then took it apart.

So here's where it lands: the same two series — AMZN's stock price and the world price of sugar — showed a correlation of −0.887 in
one carefully chosen 33-month window, and only −0.579 across the full six years. Neither number means Amazon's stock is driven by
sugar. What the gap between them means is simpler and more useful: a correlation you found by searching is a different kind of evidence
than a correlation you predicted and then checked. Ask which one you're looking at before you trust either.

If this made you curious, the MCP Server is free to try —
The Setup walks you through it.

← Part 25: Hundreds of Lines, One Sentence: Backtesting Without the Code

Hundreds of Lines, One Sentence: Backtesting Without the Code

Jaroslav Šmarda — Thu, 24 Sep 2026 12:30:43 +0000

Python for programmers, prompts for data analysts.

Testing a trading strategy on historical data used to mean opening a library like Backtrader: define a strategy class, loop through
price bars, compute indicators, simulate buys and sells, track the resulting equity curve. A simple test could easily run to a hundred
lines of Python before you saw a single result.

Let's test one of the oldest, most recognizable strategies there is — the "golden cross," buying when a 50-day moving average crosses
above the 200-day — the same way we've tested everything else in this series: by asking.

Want to try this yourself? The EODHD MCP Server is what
pulled the data above — connect it to Claude and ask your own questions (affiliate link).

New here? The Setup covers
everything you need before your first prompt.

Run the backtest with a single sentence

🟧 PROMPT

Use the eodhd MCP to backtest a golden cross strategy on AMZN 
from 2020 to today: buy when the 50-day moving average 
crosses above the 200-day, sell when 
it crosses below. Compare to buy-and-hold.

Over roughly six and a half years (2020–2026), the golden cross strategy triggered eight signals — four buys, four sells — the most
recent being a golden cross buy at $274.99 on May 6, 2026.

$10,000 invested using the strategy grew to $10,526.69 — a 5.3% total return. $10,000 in plain buy-and-hold grew to $27,057.81 — a
170.6% total return.

Buy-and-hold outperformed the "golden cross" strategy by more than 30 times over.

No Backtrader class, no manual loop through six years of daily bars, no hand-rolled equity curve calculation. One sentence, and the
strategy's entire track record came back — including the fact that it lost badly.

Ask why it lost so badly

🟧 PROMPT

Break down the returns during each period 
the strategy was out of the market, 
to see what it missed.

Between each sell signal and the next buy signal — the stretches spent in cash, waiting for a new uptrend to confirm — AMZN moved:

Apr 2021: +2.9% (brief, minor)
Jan 2022 – May 2023: −14.2% (the strategy correctly avoided this stretch)
Apr–Jul 2025: +26.7% (missed entirely, sitting in cash)
Mar–May 2026: +29.3% (missed entirely, sitting in cash)

The strategy did its job during the one genuine bear stretch — sidestepping a 14% decline. But the other three times it sat out, the
stock wasn't falling, it was recovering fast — and a moving-average crossover, by design, only confirms a new uptrend after it's already
well underway. Two rallies worth more than 25% each happened entirely while the strategy was waiting for confirmation that never came in
time.

A word of honesty

This backtest, like most quick ones, simplifies reality in ways worth naming.

It ignores transaction costs and taxes — eight trades over six years isn't expensive, but it's not free either, and the comparison to
buy-and-hold (zero trades) is already generous to buy-and-hold before costs are even considered. It also tested exactly one stock over
exactly one stretch of history; a 50/200-day crossover might behave completely differently on a different asset, a longer history, or a
market that chops sideways instead of trending hard in one direction, which is the environment crossover strategies are actually
designed for. AMZN's pattern here — long, powerful trends with sharp V-shaped recoveries — is close to the worst-case scenario for a
strategy that only confirms a trend after it's already moved.

None of that changes what happened on this specific test. It's a reminder that "backtested it once" and "proved it works" are different
claims — the same reminder Backtrader would have given you, just after considerably more setup. As always: not investment advice.

Final Thoughts

This is the twenty-fifth article in the series Unlock Real-Time Market Intelligence with EODHD and Claude, and the one that finally tested what this whole
series has been implying: that a plain-English prompt can replace not just a data fetch, but the code around it too. Six and a half
years of daily bars, eight signals, a full equity curve — no Backtrader class, no manual loop.

Two prompts ran the backtest and then explained exactly where it went wrong.

So here's what those two prompts found: buy-and-hold turned $10,000 into $27,058. The golden cross strategy turned it into $10,527 —
more than 30 times less growth, mostly lost across two rallies, worth 27% and 29%, that happened entirely while the strategy sat in cash
waiting for confirmation. The code that used to take an afternoon to write took two sentences to run. The lesson it revealed — that a
famous strategy can lose badly to doing nothing — was exactly the same either way.

If this made you curious, the MCP Server is free
to try — The Setup walks you
through it.

← Part 24: Max Pain: A Real Pattern or a Coincidence With a Good Story?

Free, Unrestricted, and Almost as Easy: Reading SEC Filings with a Prompt

Jaroslav Šmarda — Thu, 24 Sep 2026 11:45:00 +0000

Python for programmers, prompts for data analysts.

Every article in this series has leaned on one server: EODHD. But the underlying numbers — revenue, profit, shares outstanding — don't
originate there. They come from a company's own filings to the SEC, published for free, for anyone, the moment they're submitted.

So here's a natural question: could we skip the middleman entirely, and just ask the source?

Turns out, mostly yes.

New here? The Setup covers
everything you need before your first prompt.

Ask the source directly

🟧 PROMPT

Fetch data.sec.gov/api/xbrl/companyfacts/CIK0001018724.json 
for Amazon and tell me their most recent annual revenue.

Amazon's CIK (Central Index Key) is 1018724 — padded to 10 digits for the URL. Fetching that page returns a single JSON document
containing every financial fact Amazon has disclosed to the SEC since it started filing electronically: shares outstanding, accounts
payable, revenue, and hundreds more, each as a full time series stretching back years.

No login, no API key, no special header required to read it — the page opened on the first request.

No MCP server. No Docker. Not even a signup. The exact same government database that powers every "10-K" citation you've ever seen in a
finance article is sitting behind a plain URL, and Claude can just fetch it — the same way it's fetched news articles and pricing pages
throughout this series.

Pull a real number

🟧 PROMPT

From that data, what was Amazon's 
most recent quarterly accounts receivable balance?

The most recent figure in the file: $88.09 billion, for the quarter ending June 30, 2026, filed August 1, 2026. The same concept —
"AccountsReceivableNetCurrent" — stretches back to 2008, so you can trace the exact same line item across nearly two decades in one file.

That's the pattern for anything company-wide: revenue, cash, debt, shares outstanding — one concept name, one flat time series, every
quarter and every year a company has filed, going back as far as their electronic records exist. No parsing a PDF, no scraping a table
out of a press release. It's already structured, and it's already sitting there.

Ask for the one thing it can't give you

🟧 PROMPT

From that same file, break out Amazon's revenue 
by segment — AWS, North America, and International.

Looking through the file, revenue is there — but only as one number per period, the company-wide total. There's no "AWS revenue" or
"North America revenue" concept in this list. The segment breakdown genuinely exists in Amazon's 10-K — it's just not in this particular
document.

That's the wall. Everything company-wide — one number per concept, per quarter, forever — comes through cleanly. But a segment split
isn't a company-wide fact; it's a number tied to which part of the business it belongs to, and that extra dimension doesn't travel
through this endpoint. The data hasn't vanished. It's just filed under a different door than the one we just walked through.

A word of honesty

Two things worth being precise about, and they cut in opposite directions.

First, the good news is real: anything company-wide — revenue, cash, debt, shares outstanding, hundreds of concepts — comes through a
plain URL, no signup, no key, no rate-limit dance. That's rarer than it sounds; most financial APIs gate at least some of this behind a
paid tier. SEC EDGAR doesn't, because it's not a product — it's a public record.

Second, "no MCP server needed" only holds for exactly this kind of question — one concept, one company, over time. The moment you want
something structurally different, like a segment split, you've left the territory this endpoint covers. That's not a flaw in the API;
it's a reminder that "free and unrestricted" and "covers everything you might want" are different promises, and this only makes the
first one.

So: a genuinely good tool for the numbers a company reports about itself as a whole — and a dead end, at least here, for the numbers it
reports about its parts. As always: not investment advice.

Final Thoughts

This one sits outside the main numbered series — a detour to check something the series has leaned on all along without ever looking
directly at it: the government record every EODHD number ultimately traces back to.

Two prompts got us company-wide facts going back nearly two decades, no setup beyond a URL. A third hit a wall the moment we asked for
something structural instead of company-wide.

So here's where this leaves things: SEC EDGAR is as open as data gets — free, unrestricted, no key, no rate-limit dance — and it will
answer almost any question about a company as a whole. Ask about its parts, and you've asked a different kind of question than this
endpoint was built to answer. That's not a flaw to work around. It's the shape of the thing, worth knowing before you go looking for
it.

Originally published on CodeToDeploy. Part of the series Unlock Real-Time Market
Intelligence with EODHD and Claude.