Warren Buffett's Market Indicator Has No Single Value: Four Data Choices Move It by More Than 40 Points

Ask four websites for the warren buffett market indicator today and you will get four different numbers. One says 244%. Another says 288%. A third says 277%. A fourth, written earlier this year, quoted 217%. None of them is lying, and none of them is broken. Each one made a different set of data choices and then gave the result the same name.

This post lines those readings up, attributes every gap to a specific input, and shows you how to recompute each variant yourself from public series. We make no call on whether the market is expensive. The point is narrower and more useful: until you know which four choices sit behind a reading, the number can't be compared to anything.

What the Buffett Indicator is, in one line

The Buffett Indicator is the total value of US-listed stocks divided by annual economic output. That's it. Market value on top, GDP underneath, expressed as a percentage. A result of 288% means the stock market is valued at 2.88 times a year of national output.

The name comes from a 2001 Fortune article. According to Financer's explainer, Buffett used the aggregate value of US stocks from Federal Reserve data and divided by gross national product, not gross domestic product. That detail matters later, because it's the oldest of the four divergences and, as it turns out, the smallest.

Why do published Buffett Indicator readings disagree?

Because the ratio has two inputs, and each input has more than one reasonable definition. The disagreement comes from four choices:

1. Which stock-market series is the numerator. The Federal Reserve's corporate equities line, a nonfinancial-only version of that line, or a Wilshire 5000-based total.
2. Which GDP print is the denominator. The figure that was current when the Fed posted its data, the latest revised figure, or a newer quarter than the equity data.
3. GDP or GNP. Buffett's original used GNP. Most modern pages use GDP.
4. The as-of date. A quarter-end official figure and a daily market-close estimate are different readings even when they share a methodology.

Our methodology for this post was simple. We collected every reading we could attribute to a named publisher, then we pulled the underlying public series from FRED, the St. Louis Fed's data service, and recomputed the variants ourselves. Where we couldn't recompute a variant cleanly, we left it out rather than estimate.

The reconciliation table: every reading we cite, with its four choices

Here are the published readings, exactly as the publishers state them. The "as of" column matters as much as the number.

Data from thetrading.tools and Current Market Valuation is quoted from their pages as we read them on November 7, 2026. The three early-2026 readings come from Financer's article, dated November 6, 2026, which doesn't give an exact day for them.

Look at the top three rows. They all describe roughly the same stretch of 2026, and they span 244% to 288%. That's a gap of about 44 points on a ratio that most people treat as a single fact.

Numerator choices: where most of the gap lives

The biggest mover by far is the stock-market series. We pulled two Federal Reserve series from FRED for the second quarter of 2026 and divided each by the same GDP figure.

• FRED series BOGZ1LM883164115Q, the series thetrading.tools uses: $93.61 trillion, which gives 287.5% on current GDP.
• FRED series NCBEILQ027S, nonfinancial corporate business equities only: $83.05 trillion, which gives 255.0% on the same GDP.

That is a 32.5-point difference from one choice. Nothing else changed. Same quarter, same Federal Reserve release, same denominator. The nonfinancial series leaves out the equity of financial companies, so it is a narrower slice of the market by construction.

Then there's the Wilshire route. Current Market Valuation, per its own page, takes the Wilshire 5000 as the most common measure of aggregate US market value and uses the Fed's nonfinancial series only for data before 1970. Its June 30, 2026 total was $78.10 trillion, which is $15.5 trillion below the Fed-based figure for the same quarter. Dividing $78.10 trillion by $32.06 trillion gives 243.6%, which matches the 244% it publishes.

We tried to recompute the Wilshire variant ourselves. When we requested the Wilshire 5000 series from FRED's CSV endpoint on November 7, 2026, nothing usable came back. So we left that row out of our own recomputation and rely on Current Market Valuation's stated inputs instead. Dropping a row is better than faking one.

A fair question is why a Fed total and a Wilshire total differ so much. We can describe the construction of each, but we haven't reconciled the $15.5 trillion line by line, and we won't pretend to. What we can say is that they are different series from different publishers, and the ratio inherits that difference in full.

Denominator choices: vintage, nowcast and GNP

The denominator looks like the safe half of the ratio. GDP moves slowly, and it's published by one agency. It's still a choice, and it's revised.

We pulled the Q2 2026 GDP level twice from FRED's archive. The vintage on record on September 11, 2026 was $32,486.066 billion. The current vintage is $32,563.030 billion. That revision is about $77 billion, or 0.24%. On the Fed equity numerator it moves the ratio from 288.2% to 287.5%, which is under one point.

The Bureau of Economic Analysis, which publishes these figures, revises them in later releases as source data arrives. You can read its release schedule on the BEA GDP page. The Federal Reserve revises its side too. The Z.1 Financial Accounts release is rewritten each quarter, and thetrading.tools says openly that it doesn't restate either side: each quarter keeps the values on record when it entered their dataset.

Both sides move, then. That's the honest summary. The denominator revision is small today, but it's real, and it means a reading you saved in September won't match one you compute in November from the same series.

A newer quarter in the denominator

The second denominator choice is bigger. The Fed's Q2 equity level is the newest official number, but BEA has already published an advance Q3 2026 GDP of $33,010.036 billion. Divide the Q2 Fed equity total by that newer GDP and you get 283.6%, nearly four points under the same-quarter figure. This is the logic behind thetrading.tools' 277% estimate: a fresher market value over the newest available GDP.

A nowcast takes this one step further. The Atlanta Fed's GDPNow model publishes a running estimate of real GDP growth for the current quarter. It is a growth rate, not a dollar level, so using it as a denominator means rolling the last published level forward yourself. Few public Buffett pages do that, which is one reason we didn't add a nowcast row to our table.

GDP or GNP: the oldest divergence is the smallest

This is the one that surprises people. Buffett's 2001 article used GNP, and plenty of explainers make a point of it. We compared the two on the same numerator. For Q2 2026, GNP was $32,559.355 billion against GDP of $32,563.030 billion. Both give 287.5%. In Q1 2000 the Fed equity total of $17.21 trillion over GDP of $10,002 billion gives 172.1%, while the GNP version of $10,035 billion gives 171.5%.

So the swap from GNP to GDP costs about 0.6 of a point at the dot-com peak and effectively nothing now. It makes a good footnote and a poor explanation of why anyone's number differs from yours.

The as-of date is a choice too

Official figures arrive quarterly. The latest one covers Q2 2026 and posted on September 11, 2026. The next Z.1 release is due December 10, 2026, according to thetrading.tools. In between, anyone who wants a current number has to estimate.

That's why the same site shows two values. Its official reading is 288%. Its market-close estimate, which scales the official equity value by the movement of the VTI fund and divides by the latest GDP, is 277%. The site also reports that its scaling method reproduced 66 later official quarter-end values with a median absolute error of 0.32%. That is a good validation figure, but the estimate stays an estimate, and 11 points separate it from the official print published a few weeks earlier.

When two readings carry different dates, you can't tell whether they disagree about method or just about time. Pin the date first.

Putting a number on each choice

Here's our own recomputation for Q2 2026, from FRED data pulled on November 7, 2026. Every row changes one thing from the first.

Add Current Market Valuation's Wilshire-based 243.6% and the same-quarter field runs from about 244% to 288%. That is the 44-point spread in the title, and the table shows where it comes from. Roughly three quarters of it is the numerator. The denominator vintage and the GNP swap together barely register.

If you put the early-2026 readings on the same line, the quoted range stretches from 217% to 288%, which is 71 points. We don't treat that as a clean spread, because those readings carry different dates and methods at once. It's useful mainly as a warning about what happens when you mix sources.

Why the series drifts upward on its own

Set the four choices aside for a moment. Even with every choice pinned, the ratio has a structural upward drift that has little to do with valuation.

The reason is mismatched scope. US-listed companies earn a large share of their revenue abroad. Their market value reflects all of those earnings. The denominator, though, counts only domestic output. Financer notes this bias and does not quantify it. As more of the profit pool moves offshore, the numerator can grow faster than the denominator without anyone paying a higher price for a dollar of earnings.

Thetrading.tools gives a clean way to see it. It splits the ratio into two parts: market value over after-tax corporate profits, multiplied by profits as a share of GDP. For Q2 2026 it reports 21.8 times profits and a 13.2% profit share, which multiply to the 288% reading. For the dot-com peak in Q1 2000 it reports 30.5 times profits and a 5.6% share, which gives 172%. The market was pricier against profits in 2000. Profits just captured less than half the share of the economy they capture now.

The same page lists a long-run median of 85% and calls the current official value a record. Both statements are fine. They just shouldn't be read as a verdict.

Why comparing today's number to an old threshold is a category error

You'll often see a rule of thumb like "above 200% is playing with fire" next to a modern reading. Financer attributes that line to Buffett and quotes 70 to 80% as his buy zone. Those thresholds belong to a specific series with a specific vintage and a specific start date.

Take a threshold from one source and a current reading from another, and you've compared two different series. Here's a quick check to run before trusting any comparison:

1. Find the numerator series name on the page. If it says "Wilshire" and the threshold came from a Fed-based chart, stop.
2. Find the GDP vintage or the as-of date. If the page doesn't state one, treat the number as unpinned.
3. Check whether the history was restated. Some sites keep original values, others rewrite the past.
4. Only then compare the level to a historical band, and only to bands from the same source.

Our data from the Fed-based numerator shows why this matters. On that one series, the same ratio reads 172% at the dot-com peak and 288% now. On the nonfinancial-only version, the dot-com peak was 162.6% and the current value is 255.0%. Same event, same economy, different levels. A threshold only means something inside the series that produced it.

What the Buffett Indicator is actually evidence about

It is evidence about long-horizon starting valuation. It is not a timing tool, and the sources we read agree on this. Financer calls it a risk management tool rather than a timing tool. Thetrading.tools says high readings set no crash date.

The sample problem is worth stating plainly. Thetrading.tools grades each starting valuation against the next ten years of inflation-adjusted S&P returns. Its 170%-and-above band has 27 starting quarters, but only 1 has a completed ten-year window, and it comes from a single era. The site itself says the record can't yet grade a 288% starting point over a full decade. The middle bands aren't monotonic either. When the sample is that thin, any confident forecast built on it is a story, not a measurement.

Current Market Valuation, for its part, frames its 244% as strongly overvalued relative to a trend line, about 2.6 standard deviations above trend. That's one defensible reading of one series. Our job here isn't to pick between those framings, and we won't.

How a chart trader can use it, if at all

A ratio that updates four times a year, with revisions after the fact, can't belong in an entry or exit rule. Intraday and swing signals react in minutes or days. This number is stale before the next release and rewritten after it.

The one defensible use is as a slow regime label that sets context for risk. Think of it the way you'd use a long-run volatility regime. Quantzee's guide to global volatility regimes across VIX, VSTOXX and crypto covers the faster version of that idea, a label that changes how you read signals without generating any. If you size positions off a regime label, our primer on position sizing and the definition of max drawdown are the places to anchor the risk side of it.

Be wary of the opposite temptation. A big number feels like it should predict something, and a chart trader reaching for an edge can talk themselves into one. Whether a slow signal adds anything is a testable claim, and the discipline in testing an indicator against a random walk applies here as much as anywhere. For breadth-style context that updates daily rather than quarterly, the McClellan Oscillator is a closer fit for market-wide reads, and our ranking of indicators for US stock trading shows where each tool sits. None of this is advice. It is analytical software and analysis, and nothing here predicts a return.

The reproducibility recipe: compute every variant yourself

You don't need to trust our table. Every input is public. FRED serves each series as a CSV at https://fred.stlouisfed.org/graph/fredgraph.csv?id=SERIES, where SERIES is one of the identifiers below.

• BOGZ1LM883164115Q: the Fed Z.1 corporate equities level used by thetrading.tools. Values are in millions of dollars.
• NCBEILQ027S: nonfinancial corporate business equities. Same units.
• GDP: BEA nominal GDP, in billions of dollars, annual rate.
• GNP: BEA nominal GNP, same units.
• GDPNOW: the Atlanta Fed nowcast, a growth rate in percent.

Divide the equity series by GDP, remembering the unit difference (millions against billions). Then change one input at a time and watch the result move. For past GDP vintages, the St. Louis Fed's archive version of the same service accepts a vintage date, which is how we got the September 11 figure.

When you finish, you'll have a spread of your own. That's the real lesson. The next time you see a Buffett Indicator number quoted without a numerator series, a GDP vintage, a GDP-or-GNP statement and an as-of date, you'll know exactly how much weight it deserves.

Frequently Asked Questions

This article is analytical commentary and educational material, not investment advice. Figures are quoted from the named publishers and from FRED series as retrieved on November 7, 2026; they are revised over time.