The Most Famous Warning Light in Finance
On January 28, the S&P 500 closed at a record 7,002.28. That same week, the most famous valuation gauge in the world - the Buffett Indicator, the total value of the US stock market divided by US GDP - printed roughly 230%, the highest reading ever recorded.
Then the bombs started falling. American and Israeli strikes on Iran began on the last Saturday of February, oil spiked, and the S&P 500 strung together four straight losing weeks. Tonight the index sits at 6,581 - down 6% from the January record - after a 1.15% Monday rally on word of talks and a pause in strikes on Iranian energy infrastructure.
So the internet has rediscovered the Buffett Indicator, and the framing is everywhere: it predicted every crash for sixty years, it has never been higher, you do the math.
We decided to actually do the math. We went back to the 2001 Fortune essay where Buffett laid out the metric, rebuilt six decades of readings, and used Barebone to pull every Berkshire Hathaway filing since 2018 - the 10-Ks, the 13Fs, and one very interesting 8-K from three weeks ago.
The short version: the record is real. The track record is messier than the legend. And Berkshire's own filings tell a sharper story than the cash-pile headlines - $373.3 billion in cash, twelve straight quarters of net selling, and then, on March 4, the first share repurchases in nearly two years.
One Division, Three Zones
The indicator is a single division: take everything the US stock market is worth, divide it by everything the US economy produces in a year. As of tonight, that is roughly $69.5 trillion of equity value against about $31.6 trillion of annualized GDP - call it 220%, down from January's 230% record but still miles above anything in the historical record.
Buffett's endorsement, in the December 2001 Fortune essay he wrote with Carol Loomis, is the reason the metric carries his name. He called it "probably the best single measure of where valuations stand at any given moment." And he gave it zones:
"If the percentage relationship falls to the 70% or 80% area, buying stocks is likely to work very well for you. If the ratio approaches 200% - as it did in 1999 and a part of 2000 - you are playing with fire."
One correction to the version going around short-form video: the popular "120% means expensive" threshold is an analyst convention added later. Buffett's own essay names two lines - the 70 - 80% buy zone and the 200% fire zone. Here is where those lines sit today:
| Zone |
Reading |
Whose label |
| "Likely to work very well" |
70 - 80% |
Buffett, Fortune 2001 |
| "Expensive" |
~120% |
Analyst convention, not Buffett |
| "Playing with fire" |
~200% |
Buffett, Fortune 2001 |
| January 2026 record |
230% |
Uncharted territory |
| Tonight |
~220% |
Still above every prior extreme |
Look at the right side of that chart. Even after a 6% war discount, the reading sits above the 200% fire line - and roughly 30 points above the February 2021 peak, which was itself the all-time record until last year. On a consistent series, today's market is more expensive relative to the economy than at any moment in six decades, including the top of the dot-com bubble.
The Track Record the Legend Skips
The viral claim is that the indicator "predicted every major crash for sixty years." The honest version: January was only the fourth time in roughly sixty years that the gauge stretched about two standard deviations above its long-run trend. Here are the previous three, and what followed:
| Episode |
Indicator at peak |
What the S&P 500 did next |
Trough |
| 1968 - 69 |
~100% |
-36% |
May 1970 |
| 2000 dot-com |
159% |
-49% |
October 2002 |
| 2021 - 22 |
190% |
-25% |
October 2022 |
| January 2026 |
230% |
-6% so far |
Open |
Three for three, each followed by a decline of at least 25%. That is the part of the legend that survives a fact-check, and it deserves respect.
Now the two parts that do not.
The indicator missed the biggest crash of the modern era. In 2007, on the eve of the global financial crisis, the gauge peaked around 118% - the mildest of the major peaks, nowhere near its prior extreme. The S&P 500 then lost 57%. The worst drawdown in the dataset arrived with the warning light barely flickering, because 2008 was a leverage crisis, not a valuation crisis. An indicator that sleeps through Lehman is not a crash predictor. It is a valuation thermometer.
And the false-positive problem is worse. The gauge has trended upward for thirty years - the lows of 2009 would have counted as average readings in the 1950 - 1995 era. It cleared its dot-com extreme years ago and just kept climbing. Anyone who went to cash the first time it crossed Buffett's 200% fire line spent the entire melt-up to January's record watching from the sidelines. A signal that has been flashing red for most of a decade cannot tell you when anything happens.
Three precedents is not a dataset. The thermometer is real; the stopwatch is fiction.
The $373 Billion Tell
Which brings us to the more interesting question the script writers love: is Buffett putting his money where his metric is?
The positioning is unambiguous. Berkshire Hathaway ended 2025 with $373.3 billion in cash, equivalents and short-term Treasuries - the largest liquidity position any company has ever held, bigger than the entire market value of roughly 95% of the companies in the S&P 500. The filings behind it read like a slow-motion exit: Berkshire sold more stock than it bought for twelve consecutive quarters through September 2025, and repurchased zero of its own shares in all of 2025 - nothing, in fact, since May 2024.
The shape of that chart is the argument. Cash roughly doubled in 2024 and kept building through 2025 - exactly the years the indicator went vertical past 200%. Whatever Buffett says about the metric these days, Berkshire behaved like a firm that could not find prices it liked.
Two footnotes the doom edit leaves out. First, "stopped buying stocks" was never literally true: Berkshire opened a $4.3 billion Alphabet position in the third quarter of 2025 and held all 17.8 million shares through the year-end 13F filed February 17. Selective, not absent. Second, caution has a bill: Berkshire shares returned 10.9% in 2025 against 17.9% for the S&P 500. Holding a third of a trillion dollars in T-bills during a melt-up is how you underperform by eight points.
And the man himself has changed seats. Warren Buffett handed the CEO role to Greg Abel on January 1 and stays on as chairman. Abel's first annual letter landed on February 28 - the same Saturday the first strikes hit Iran. The cash pile he inherited was built by Buffett. What happened next was the first real signal of the new era.
Then March Happened
Here is the twist the crash narrative cannot absorb.
On March 4, with the S&P 500 down roughly 6% from its record and Berkshire's own stock caught in the war selloff, Berkshire disclosed it had resumed share repurchases - about $226 million of Class A and B stock on the first day alone, the first buybacks since May 2024. Under Berkshire's policy, that can only happen when both Abel and Buffett agree the shares trade below a conservative estimate of intrinsic value. Abel went further and bought 21 Class A shares personally - roughly $15 million of his own money - and said he plans similar purchases.
Read that sequence carefully, because it is the whole worldview:
The record cash pile was never a prophecy that markets would crash. It was a price list. Berkshire spent three years refusing to buy at 200%-and-climbing - and then started buying the moment something it understood (its own stock) got 6% cheaper amid a panic about something else entirely (oil and war). They did not wait for the -49% scenario. They did not call a bottom. They had a number written down, and the number arrived.
The cash was never about whether. It was about at what price.
Where the Indicator Breaks
Before you wire your conviction to one ratio, the known flaws - and they are structural, not nitpicks:
The numerator is global; the denominator is domestic. US-listed companies earn a large and growing share of their revenue abroad - Apple in Europe, NVIDIA in Asia - none of which shows up in US GDP. That mismatch alone biases the ratio upward over time, which means comparing today's 220% against 1968's 100% is not apples to apples. Some of the "overvaluation" is just globalization.
It ignores interest rates. A stock market competes with bonds for capital, and the ratio has no opinion about the competition. Buffett himself made this the core caveat when he walked back his endorsement at the 2017 shareholder meeting, declining to bless any single metric and noting that "everything in valuation gets back to interest rates" - rates act on asset prices, in his words, the way gravity acts on matter. The dot-com extreme formed with 10-year Treasuries near 6.5%; the current one formed with rates meaningfully lower. Same ratio, different gravity.
The measurement itself is squishy. Depending on whose market-value series you use, the 2000 peak reads anywhere from 137% (Wilshire 5000 basis) to nearly 200% (the series Buffett cited in Fortune). We used one consistent GDP-based series throughout this piece, but a gauge whose historic landmark moves 60 points depending on the vendor is not precision machinery.
And the economy's composition drifted. Far more of American business is publicly listed and intangible-heavy than in 1970. A software economy capitalizes differently than a steel economy, and the ratio's long upward trend partly reflects that - not just mania.
None of this makes 220% comfortable. It makes 220% imprecise - a number that says returns from here are likely to be poor, while saying nothing about the path.
What This Means
The indicator and the cash pile are answering two different questions, and the mistake is collapsing them into one.
The indicator is a forward-returns thermometer, not a timing device. At roughly 220% of GDP, history's message is that the next decade of returns is likely to be far thinner than the last one. Its message about next month is precisely nothing - it missed 2008, and it screamed through years of gains.
Watch the denominator, not just the numerator. The war has markets repricing recession odds. If GDP stalls while stock prices merely hold, the ratio gets worse without a single up-day - one more reason extreme readings can persist or resolve in strange ways.
The Berkshire playbook is the actionable read. Hold liquidity when nothing clears your bar; act the moment something does, even if the headlines are still on fire. The next checkpoints are all dated: Berkshire's first-quarter 13F in mid-May will show whether the buying extended beyond its own shares, the Q1 report in early May will show the buyback pace, and the indicator itself will tell you whether January's 230% stands as the high-water mark of the cycle.
Every time this gauge reached two standard deviations above trend, the market eventually fell at least 25%. That sentence is true. So is this one: the people who actually run the world's largest cash pile just spent it on the way down, six percent off the top, while everyone else was quoting their indicator as a reason to hide.
The question the cash pile asks is not whether a crash is coming. It is whether you know - in advance, in writing - the price at which you would act. Berkshire wrote theirs down. March showed us the number.
Data: Barebone | Sources: Fortune, "Warren Buffett on the Stock Market" (December 2001), Berkshire Hathaway FY2025 Annual Report and 10-K (SEC EDGAR), Berkshire Hathaway Q4 2025 13F (SEC EDGAR), Berkshire Hathaway March 2026 buyback disclosure | Data as of March 23, 2026
