Warren Buffett: four crashes and the case against borrowed money
On 18 September 2026 Warren Buffett stepped down as chairman of Berkshire Hathaway, sixty years after he took control of it. The tributes will all quote the compounding. The most useful thing he ever published is the opposite: a four-line table of the times his own shares fell by more than a third, and what he concluded from it.
KEY TAKEAWAYS
- Berkshire shares fell 59.1%, 37.1%, 48.9% and 50.7% — one of them in 25 days. Source: Buffett’s own 2017 letter.
- Each fall needed a bigger gain to undo: the 2008–09 drop of 50.7% needed +102.8%. Our arithmetic, checked on our calculator.
- Owned at 2× with borrowed money, all four falls would have gone through the 33.3% drop that triggers a margin call at 25% maintenance. Our arithmetic.
- The record is subtler than the slogan: researchers estimate Berkshire ran at about 1.7-to-1 leverage — but from insurance float, not from a broker who could call it in.
Who is Warren Buffett?
Warren Edward Buffett was born on 30 August 1930 in Omaha, Nebraska, and has lived and worked there for most of his life. He studied under Benjamin Graham at Columbia, ran his own investment partnership from 1956, and in 1965 took control of Berkshire Hathaway, a declining New England textile company that he turned into a holding company for insurers, railways, utilities and large stakes in public companies.
The number everyone quotes comes from his letters: from 1965 to 2017, Berkshire’s per-share market value compounded at 20.9% a year, against 9.9% for the S&P 500 with dividends included. He handed the chief executive role to Greg Abel on 1 January 2026, and on 18 September 2026 Berkshire announced that he was stepping down as chairman, to remain a director with the title chairman emeritus, with his son Howard Buffett taking the chair.
None of that is a trading lesson. You cannot copy sixty years, an insurance company or a reputation. What you can copy is the way he talked about losses — in writing, every year, with numbers.
How far did Berkshire’s own shares fall?
Further than most people who quote him would guess. In the 2017 shareholder letter, Buffett printed what he called “the gory details”: four periods in which Berkshire stock suffered a truly major dip, with the high, the low and the percentage.
| Period | High | Low | Fall | Gain needed to get back (ours) |
|---|---|---|---|---|
| March 1973 – January 1975 | $93 | $38 | −59.1% | +144.5% |
| 2 – 27 October 1987 | $4,250 | $2,675 | −37.1% | +59.0% |
| 19 June 1998 – 10 March 2000 | $80,900 | $41,300 | −48.9% | +95.7% |
| 19 September 2008 – 5 March 2009 | $147,000 | $72,400 | −50.7% | +102.8% |
The first four columns are his. The last one is ours: the gain needed to return to the old high is the fall divided by what is left, so a 50.7% fall needs 50.7 ÷ 49.3 = +102.8%. Put the two side by side and the asymmetry stops being a slogan.
We put the 2008–09 fall into our own drawdown calculator to check the arithmetic and to give it a time scale. For the monthly return we used 1.6%, which is what 20.9% a year works out to per month — in other words, Berkshire’s own long-run pace.
That is the part the compounding headline hides. A 20.9% average is made of years like these. A $10,000 position became $4,930 in under six months, and at the very pace that made Buffett famous, the climb back is close to four years.
What would borrowed money have done to the same shares?
This is where Buffett drew his conclusion, and it is worth reading in full:
This table offers the strongest argument I can muster against ever using borrowed money to own stocks. There is simply no telling how far stocks can fall in a short period.Warren Buffett, letter to Berkshire Hathaway shareholders, 2017 (published February 2018).
He goes on: even if your borrowing is small and your positions are not immediately threatened, “an unsettled mind will not make good decisions.”
He did not run the numbers in the letter. We did. Take the same four falls and apply them to someone who owned Berkshire with borrowed money — $1 of their own for every $1.50 or $2 of stock — ignoring interest, so this is the kind version.
| Fall in the stock | Your loss, no borrowing | Your loss at 1.5× | Your loss at 2× |
|---|---|---|---|
| −37.1% (1987) | −37.1% | −55.7% | −74.2% |
| −48.9% (1998–2000) | −48.9% | −73.3% | −97.8% |
| −50.7% (2008–09) | −50.7% | −76.1% | wiped out |
| −59.1% (1973–75) | −59.1% | −88.7% | wiped out |
And the table flatters the borrower, because nobody gets to sit through a fall like that on margin. With half the position borrowed and a 25% maintenance requirement, the broker calls for more cash once the stock is 33.3% below where you bought (the value falls to 0.667 of the start, the loan is 0.5, so equity is 25% of the position). Every one of Berkshire’s four falls went deeper than that. The person who believed most in the company and borrowed to own more of it would have been sold out near the bottom four times, and would not have owned the shares for the recoveries.
That is the whole lesson in one sentence: the stock survived, and the borrowed position did not. Being right about the asset is worth nothing if the structure you hold it in cannot survive the path.
PRACTICE CORNER
Buffett’s rule is about structure, not stock-picking, and you can apply it before your next trade. On whichever exchange you use, open the order screen and, before you size anything, set leverage to 1× and look at the liquidation price: on a spot or unlevered position there isn’t one. Then raise it step by step and watch how close that line moves to the current price. If a fall smaller than the smallest of Berkshire’s four — 37.1% — would close you out, the position is sized for a market that has never existed. Leverage and margin shows the same trade at 3×, 10× and 50×.
These are affiliate links: we earn a commission if you sign up, it costs you nothing, and it does not change what we write. See our disclosure and the full exchange comparison.
Did Buffett really never use leverage?
No, and this is the part of the story the slogan leaves out.
The something else is insurance float. Buffett’s 2017 letter describes it as “money that belonged to others but was held by our two insurers”: premiums are paid up front, and claims are paid out months, years or decades later. At the end of 2017 that pool was $114.5 billion. It is borrowed money in the sense that it is not Berkshire’s. It is not borrowed money in the sense that matters on a bad day: no lender can demand it back because Berkshire’s share price fell 50%.
So the honest version of the lesson is narrower and more useful than “never use leverage”. It is: never hold an asset with money that can be taken away from you on the way down. The one that can be taken away is exactly the one a crypto exchange offers you.
All of this is checkable. The letters are free on Berkshire’s own site, every year since 1977, and the two we quote most are the 2017 letter (the four falls, the float) and the 2007 letter (the worst deal).
What did his worst mistakes cost?
He tells you, which is rarer than it should be. In the 2007 letter he called Dexter Shoe, bought in 1993, the worst deal he had made. He paid $433 million — in Berkshire stock rather than cash — for a shoe business that he later called worthless. Because he paid in shares, the cost to Berkshire shareholders was not $400 million but, in his words, $3.5 billion: he had given away 1.6% of a business then worth $220 billion. The lesson he drew is a sizing lesson, not a stock-picking one: a mistake paid for in your best asset compounds against you.
The second was about reputation, not money. In 1991 Salomon Brothers, in which Berkshire had a large investment, admitted to submitting false bids at US Treasury auctions, and Buffett became its interim chairman to save it. Testifying to a House subcommittee that September, he set the rule he later repeated to Berkshire’s managers:
Lose money for the firm, and I will be understanding; lose a shred of reputation for the firm, and I will be ruthless.Warren Buffett, testimony to a US House subcommittee on the Salomon Brothers Treasury bid scandal, September 1991.
A trader’s version: a loss you can explain in your journal is recoverable; breaking your own rule to avoid it is not.
What does Buffett’s table mean for a crypto trader?
The comparison is ours, not his. Asked about bitcoin in May 2018, he told CNBC it was “probably rat poison squared”. You do not have to agree with that to use his arithmetic, because the arithmetic does not care what the asset is.
Two things change in crypto, and both make his point sharper. First, the falls are larger and faster: a 37% decline that took Berkshire 25 days in 1987 is not an unusual month for many coins. Second, the leverage is closer. On a perpetual futures contract at 5×, a move of about 20% against you erases the margin before fees and maintenance requirements, which bring the liquidation price closer still. That is shallower than every fall in Buffett’s table. Here is the deepest of them on a phone:
Buffett could wait 56 months because nothing forced him to sell. Your position on an exchange will be closed for you at the liquidation price, long before your thesis has a chance to be right. That is the line his table draws: patience is only an option for capital that cannot be taken away.
Where this lesson stops working
Berkshire recovered every time. The table describes a company that kept compounding underneath its falling share price. Most assets, and most coins, are not that. Holding through a 59% fall was correct for Berkshire; holding through a 59% fall in a token that never recovers is just a 59% loss. The table argues against borrowed money; it does not argue for holding anything forever.
He was not a trader. Buffett holds businesses for years and does not use stop-losses. If you trade a defined setup with a stop, your protection is the stop and the position size, not patience. Read his table as a limit on structure — how much you borrow — not as permission to sit through losses your plan says to cut. Avoiding the big loss (Stage 11) is the trader’s version.
His leverage was not your leverage. Float that cannot be called in, at low cost, held by an insurer with decades of claims history, is not available to individuals. Quoting the 1.7-to-1 figure as a reason to use 2× on an exchange misses the only feature that made it safe.
Common mistakes when learning from Buffett
Quoting the 20.9% without the four falls. The average is made of those years; you only earn it if you are still holding at the bottom. Reading “be greedy when others are fearful” as a timing signal. His own letter says no one can tell you when declines will come, only that they will. Taking “he never uses leverage” as fact. The record says he used a lot of it — the kind that cannot be margin-called. Treating the bitcoin remark as the lesson. His opinion of an asset is the least transferable part of his work; his arithmetic about drawdowns is the most. Sizing for the average fall. His smallest major decline was 37.1%; plan for the worst one in your own market, not the typical one, as why risk comes before strategy (Stage 9) explains.
FAQ
How much did Berkshire Hathaway stock fall in its worst crashes?
Buffett listed four major declines in his 2017 shareholder letter: 59.1% from March 1973 to January 1975, 37.1% from 2 to 27 October 1987, 48.9% from June 1998 to March 2000, and 50.7% from September 2008 to March 2009. By our arithmetic each needed a larger gain to undo, from +59.0% for the 1987 fall to +144.5% for the 1973 to 1975 one.
Why was Warren Buffett against buying stocks with borrowed money?
Because there is no telling how far stocks can fall in a short period, as he put it in the 2017 letter, and because borrowed money can force you to sell at the bottom. By our arithmetic, a stock owned at 2 to 1 with a 25% maintenance requirement gets a margin call after a 33.3% fall, and all four of Berkshire's major declines went deeper than that.
Did Warren Buffett use leverage?
Yes, but not the kind he warned against. The 2018 Financial Analysts Journal study Buffett's Alpha estimated Berkshire's average leverage at about 1.7 to 1. Much of Berkshire's borrowed capital was insurance float, premiums held before claims are paid, which a lender cannot call in because the share price fell. Margin loans and exchange leverage can be called in, which is the risk his table describes.
What did Warren Buffett say about bitcoin?
In May 2018 he told CNBC that bitcoin was probably rat poison squared. His view of the asset is the least useful part of his work for a crypto trader. His arithmetic on drawdowns and borrowed money applies to any asset, and it applies more sharply to one that moves faster and is offered with higher leverage.