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
Top comments (0)