Three Stopped Calls Before a US$6bn Technology Purchase
A famous fund spent months preparing for a technology sell-off. Then its manager tried three times to buy back in, stopped himself, and finally made the call.
By the team at Banyantree Investment Group
In March 2000, in the offices of Soros Fund Management in Midtown Manhattan, the manager of the most famous hedge fund in the world sat watching two of his own employees make money he had told the firm was not worth making. Their account was small, a corner of the firm deliberately left alone. It was invested in the technology stocks he had spent January telling everyone had lost their minds, and, in the account he gave 15 years later, it was making about 3% a day. By March it stood up roughly 50% for the year. The main fund, moved toward the safety he had chosen for it, was up 7.¹
The two managers running the small account were his own hires. When the firm's technology holdings were sold in January, their positions had been left alone; the account was too small to hurt anyone. So one firm now held two answers to the same question: the large answer had left the market on judgement, and the small answer had stayed in and was being paid for staying, every day, in public. The comparison proved nothing on its own; the two books carried different capital and different risk. What the comparison did was arrive. Every morning the market opened, marked the distance between fifty and seven, and delivered it to the desk of the man who had chosen.
He was Stanley Druckenmiller. He had run George Soros's Quantum Fund, the large answer, for more than a decade, alongside his own firm, Duquesne, where he would retire years later with returns of roughly 30% a year across three decades and not one losing calendar year.² The January sale had been his, and he had made it in person, in Soros's office, saying out loud what half the market privately suspected and putting a number on it. In his telling the sale was total, the technology book gone in a stroke, and the fund would step aside and wait.¹
That March, with the number on the small account's screen climbing again, he picked up the phone to put the position back on, telling himself as he did it not to. Then he put the phone down.
So far this is the anecdote the industry knows; he has told it himself, at least twice, and it has worn smooth in the telling. The record that survives around it is not smooth, and in the one place where the anecdote is cleanest, it disagrees with him.
The instinct he was resisting had a history. He had already made both of the mistakes available in this market, once each, at full price, and the first of them had taught him something real.
In February 1999, Druckenmiller put on a short of about US$200m against the internet names. Yahoo and America Online had risen past anything his valuation work could carry, and he expected them to fall. They rose instead, violently. Within weeks the short had lost roughly three times the position he had started with, and Quantum was deep in the red for the year.¹ The stocks he had shorted would in fact collapse, eventually, completely. It didn't matter. A market can stay expensive long enough to impose losses that are real and immediate, and the eventual verdict on the internet would say nothing about whether Quantum could afford to keep paying for the wait.
What he did next turned the loss into a diagnosis. The firm, he concluded, knew technology the way it knew everything else it traded: it understood IBM, it understood Hewlett-Packard, and it did not understand the new companies at all. It had been shorting businesses it couldn't analyse. That was a finding, and it had a date on it, and it pointed at a remedy. He hired two young managers who did understand the new names, and then he did the thing that looks, from a distance, exactly like what he would do a year later: he followed them back into the very stocks he had been shorting, and rode the sector up. The fund ended 1999 up 35%.¹
The market moved against him, and the pressure that movement generated sent him looking for what he had missed. A search provoked by pressure usually returns whatever its owner needs it to; his returned a fact, a gap in the firm's competence, specific enough to name and specific enough to fix. The re-entry followed the fact. The position was downstream of the finding, and the finding was downstream of honest work, and the work had been provoked, legitimately, by price. That sequence is what a genuine change of mind looks like, the thing every investor claims to be doing when they reverse, and in 1999 the file supports the claim.
So when he sold in January 2000, he was not a man who fought tapes on principle. He had been punished for shorting the bubble too early, and then rewarded, richly, for joining it on better information. The receipts from both were still on the desk. If experience could inoculate anyone against what came next, it should have been him.
The sale itself took place in person, in January 2000, and his account of it is compact. He walked into George Soros's office and said he was selling the technology book, the whole of it. The number he attached to the decision was 104 times earnings.¹ That was what the market was asking him to pay for the sector's profits, and he would not pay it. The fund would step aside and wait for the next fat pitch.
The 2009 telling of the same decision adds a second leg. The market's own internals had turned ugly on his screens, he recalled: something like 13 new highs against 242 new lows, an advance carried by 10 or 15 technology names while the rest of the market quietly left.⁶ Valuation said the sector was priced beyond defending; his own charts said the rise was hollowing out beneath it. The two legs agreed, and they were his two oldest instruments, the fundamental work and the tape-reading he had built his career on, pointing the same way at once.
Everything later turns on the design of the decision. A bet that technology would fall the following week would have been the 1999 short again, and he had already paid dearly for the difference between being right and being right on time. The January sale was built to survive the thing the short could not: the market continuing to rise. Continuation was in the analysis from the start, the stated cost of the position, accepted in advance the way a buyer of insurance accepts the premium. If the market kept going up, Quantum would keep earning 7 and watching, and that would be the decision working. There was, in the design of the thing, almost no news the tape on its own could deliver that January hadn't already counted.
The two specialists kept their account. It was small, he reasoned, too small to do the wider fund any damage.¹ Whatever the intention, the effect was to build a controlled experiment and install it inside the same offices. One firm, one question, two live answers: the large book out on valuation, the small book in on the specialists' conviction, and a market that would score both every day at the close. In February 1999 the gap between his view and the market's had accumulated in other people's portfolios, at a distance. This time the firm would generate the evidence against him in-house.
He was, by every account including his own, done with technology at those prices. All he had to do now was nothing.
Doing nothing turned out to be work. The firm's own rooms preserve what that work looked like. When The Wall Street Journal reconstructed the fund's final months that May, from people who had been inside them, it found an institution that had made its decision and was still, week after week, managing the consequences. Through late 1999 and into 2000, the Monday afternoon research meetings ran at a long table in a room overlooking Central Park, and for months they kept returning to the same subject: the technology sell-off that had not yet happened. Which signs would confirm it was starting. Which shares would go first. How fast the selling would need to be. Druckenmiller sat at the head of the table and warned the group that the turn might be close, and hard when it came. Soros, often travelling, telephoned senior staff with the same message in plainer words: technology was a bubble.³
This is where the record turns on the anecdote. At Lost Tree the sale is total: one meeting, the whole book, done, the fund out and waiting. In 2009 he told a room of traders the same thing in fewer words: he sold everything and sat flat, with the fund up around 13%.⁶ The contemporaneous record is messier. The Journal's reconstruction describes only modest selling in the new year, with the funds still carrying much of their technology book as the meetings ran,³ and the April accounting would agree: he had sold some, he said then, not enough and not early enough, and the year's early technology gains had for a while covered its macro losses, gains a fund that left in January could not have earned.⁴
The retellings agree with each other, and together they disagree with the record of the time. How far out Quantum actually was in any given week cannot be settled from the public record, and the difference is not small: a clean exit rebuilt in one US$6bn stroke and a technology book never fully dismantled and then enlarged near the peak are different portfolio events, carrying different exposure already in place and different quantities of new risk. The mechanism survives both. In every version his stated judgement ran one way, out, and by the top the fund's exposure pointed the other, at a scale his own analysis condemned. The account he has kept telling is the one most damning of himself, and what no version supplies is a sufficient reason for the buying.
In early March, with the Nasdaq still climbing, Druckenmiller told the group: "I don't like this market. I think we should probably lighten up."³ The judgement of January was not eroding. Weeks of rising prices had arrived, and the man who had led the firm out was still, on the record of people present, urging less exposure, later, at higher prices. Whatever was about to change his position, it had not changed his mind.
And every one of those Mondays ended, and the market opened on Tuesday, and the small account went up again. The arithmetic of the gap was visible inside the firm in a way few investment mistakes ever are: no waiting for a quarterly attribution note or a year-end peer table, just a number, on a screen, most days a few per cent better than yesterday, produced by two people he had hired, in stocks he had sold.
By March the small account stood near 50% for the year against the fund's 7, and the 7 was the number that travelled, into Quantum's published performance, into every conversation with investors about what the fund was doing with their capital, into the cheerful fact of the two people who had ignored him being proved right every day. In February 1999 the market had punished his view and the punishment had at least produced a discovery. This time there was, on his own accounts, no discovery to go looking for. He knew exactly why he had sold. The stocks were going up, and the people who had stayed were being paid for it, and that, in every account that survives, was very nearly the entire dossier.
In his telling, 15 years afterwards, what he remembered of those weeks was the need. He had to play. The account in front of him was minting money in a market he had publicly renounced, and three times in one week he picked up the phone to put the position back on, saying to himself, as he dialled, the words of a man who already knew the verdict: don't. Three times he put the phone down.¹ Each abandoned call was the January analysis holding its ground against everything the screen was sending. Nothing had crossed the tape, in any account he gave, that he would have accepted from one of his own analysts as a reason. The case for buying was on the screen, and it was denominated overwhelmingly in other people's gains.
On the fourth attempt, he let the call go through.
He bought roughly US$6bn of technology stocks. The purchase cannot be dated to the day from the public record, but it can be dated by its owner: he reckoned afterwards that he had missed the top of the market by about an hour.¹ The Nasdaq's closing peak came on 10 March 2000, at 5,048.62, a number it would not close above again for 15 years.⁵ If his reckoning is right, the fund that had spent months meeting about the crash re-entered the market it was braced for inside the final hour of its rise.
What followed took six weeks. The Nasdaq fell through the rest of March, steadied, then broke properly in April, and the positions he had bought within an hour of the top fell with it. By his own attribution, that one trade cost the fund about US$3bn.¹ The wider damage was worse, because the purchase had not been Quantum's only problem: there had been macro losses early in the year, hidden for a while by the technology gains, and there was now the cost of selling into a falling market to raise cash against the redemptions a year like this invites.⁴ On 28 April 2000, Soros and Druckenmiller sat down with a small group of journalists at the firm's offices and gave the numbers. Quantum was down roughly 22% for the year. Druckenmiller was leaving, and so was Nicholas Roditi, who ran the Quota Fund. Soros's letter to investors recast the firm itself: the flagship would be reorganised as the Quantum Endowment Fund, run at lower risk, its capital spread across smaller teams, because a fund of its size, the letter said in substance, could no longer move the way the old Quantum had moved.⁴ The new structure addressed the size of future positions and the distribution of authority; it could not recover whatever analysis had connected the March reconsideration to the trade already made.
"I overplayed my hand," Druckenmiller told the reporters.⁴ His explanation that day was the explanation of a man accounting for a fund's year, because that was the question in the room: the macro losses, the masking gains, the selling costs, a fund too big to leave a falling Nasdaq quickly. He did not disown the sector. The firm's technology programme, he estimated, had made about US$5bn over its life against some US$2bn lost at the end.⁴ His regret, that April, ran to timing: he should have sold in February, he said, when the exuberance was already visible; they had treated the market as an eighth-inning affair when it was in its ninth.⁴
Fifteen years later, asked a different question, he gave a different accounting. At Lost Tree the subject was a career's worst mistake, and for that he did not choose the macro book or the selling costs. He chose the purchase. The two accounts look like rivals, US$2bn against US$3bn, February against March, until you notice they answer different questions at different distances. What is notable is what they share. In April 2000, explaining the year, he offered no development that had justified the re-entry. In 2015, explaining the mistake, he offered none either, and was not looking for one. Just once, between the two, in 2009, talking to a room of professional traders, did he put anything at all in the column marked what changed.⁶
What changed. The question sounds forensic, and it is. A genuine change of view rests on at least one fact the original decision had not already digested, and facts of that kind carry dates. A capitulation rests on the reasons the decision-maker already knew, rearranged by pressure into something that feels like analysis, and those carry no dates at all, because the world changes on dates, and the room changes by degrees, which leave no entries. From inside, the two are near-impossible to tell apart. On paper, dated, they separate at a glance. His own record supplies the dates.
January's file is full and dated. The valuation, 104 times earnings. The internals, hollowed to a handful of leaders, per the 2009 telling. The lesson of the 1999 short, that being early is its own catastrophe, which is why the new position was an abstention rather than a bet against. The stated cost, accepted in advance: the market might keep rising, and the fund would watch. Every entry carries a January date or older.
Now the additions of March, entered one at a time. The price had kept going up. But continuation was in the January file already, priced as the premium; weeks more of it added magnitude, and magnitude is not information about a thesis that never claimed to know the date. The gap had reached fifty against seven. But the gap was the price's shadow, and its address gives it away: it was a fact about the fund's year, the investors' questions, the small account's daily arithmetic, none of which appears anywhere in the January analysis of what technology earnings were worth.
And his own conduct dates the file better than any reconstruction could. In early March, at the long table, he was still urging less exposure. Through the week of the stopped calls he was, on his own account, arguing the other side to himself, mid-dial. There was a need, and he has said so, in public, more than once, in words that require no gloss.
The March additions reduce, at first reading, to a single entry: the premium named in January arriving, on schedule, in public, in front of the two people collecting the other side of it. Expected pain is not, by itself, new evidence. But the composition of pain can carry information its arrival does not, and the facts of that March did not divide neatly into analysis on one side and need on the other: the specialists' performance might carry information about the companies, and it also produced the comparison that made exclusion hard to bear. Two entries in the file deserve better than the summary, one built by any careful reader, one entered by the man himself.
The reader's candidate: the specialists were still winning. The two of them were the correction to a proven research failure, hired precisely because the firm's old valuation lens could not see the new companies, and their account's performance might be read as the continuing signal of that repaired understanding, the 1999 discovery still paying out. Perhaps the sector really did contain businesses the January framework mispriced, and the people best placed in the world to know were on his own payroll, compounding. That is a real argument, and in 1999 its exact shape had justified a real reversal. By March, on that evidence, January deserved another hearing.
But follow what the hearing licenses. If the new fact was their insight, the position it supported was their kind of book: their names, at a size scaled to the insight, built the way the 1999 re-entry had been built, on their work. And even that book would have needed translating into flagship terms. A position a small account can enter quietly grows expensive as Quantum buys it, and hard to leave at Quantum's size, and several individually attractive companies become one crowded exposure once correlations rise. A company-level finding can support a stock without supporting the sector; a reason to participate can support a staged entry without supporting US$6bn. What went on instead was roughly US$6bn of technology exposure, its composition undisclosed in any account, taken in his telling in one decision, at a size that dwarfed the specialists' entire account, by a man who has never claimed to understand the companies any better in March than he had in January, with the top, on his reckoning, an hour away. The trade he describes is the wrong shape for the evidence said to justify it, and the right shape for the gap, which was denominated in sector performance and could only be closed by buying technology at scale.
The record's candidate comes from its owner, once. In the 2009 conversation, describing the buyback, he paused the confession to give the other side its due: "my momentum charts were bullish, but they were bullish for an equity chart."⁶ The entry is the genuine article and deserves to be taken seriously. It is dated, it sits between the old position and the new one, and it touches a January input directly: the internals that had helped argue him out in January had, on his own system, turned. If the file holds a fact, this is it.
Now watch what its author does with it. The entry arrives with the words to be fair, the phrasing of a man being scrupulous about a case he does not believe. The qualification arrives inside the same sentence: bullish for an equity chart, a shape he had watched commodity charts trace on the way to falling out of the sky, on a system built from rate of change and second derivative, designed to distrust exactly this. His standing rule, stated in the same conversation, was that he would never invest on a chart alone. And the frame he chose for the episode, before the entry and after it, was that he had completely lost his cool and his discipline. The line is real, and it stays in the file; its weight is the weight its owner gave it. The world column, on the fullest accounting its owner ever gave, holds one line, entered under protest, qualified as it was written, and overruled by its author in every telling before and since.
So the file stands as its owner left it. On why January became unbearable it is thick: the small account's daily gains, the widening gap, the Mondays at the long table, the phone. On what was new it holds a single line with its rebuttal attached.
That file is the surviving public account, not the firm's internal one. A speech, a conversation with traders, and a press conference were never built to preserve research, and their silence cannot establish that no company work, no sizing deliberation, existed inside Soros Fund Management that March. What the record establishes is narrower, and stranger: in a quarter of a century of retellings, its owner has never once supplied that work, never named a security, a horizon, or a route from reasonable doubt to US$6bn, and in every retrospective account has characterised the trade as a loss of discipline rather than a change of analysis.
What the file proves is less than a morality tale. It does not prove the purchase was wrong in prospect. Reasonable readers of the March market existed who thought the run had years left; had the turn come three months later, this trade would sit today in the long list of his celebrated reversals, cited as flexibility, and no one would ask about the dates. Nor does the record hand back the opposite moral, that the answer was to hold, because the same market was disposing of the manager who did: on 30 March 2000, 20 days past the Nasdaq's peak, Julian Robertson closed Tiger Management, having refused these stocks to the end and run out of investors just before the refusal paid, so that one month broke one of the era's two great funds for reversing and the other for refusing.⁷
What the record demonstrates is narrower and, for anyone who sits on an investment committee, worse. A decision changed under pressure, the only analytical support its maker ever named a signal he discounted even as he cited it, inside a structure where one man could reread the evidence, choose the size, and act. If anything in that structure required the difference between a changed world and a changed room to be stated before US$6bn moved, or required the size of the answer to bear a stated relationship to the evidence that reopened the question, it left no mark on the outcome. The safeguard everyone assumes was present, experience, judgement, self-knowledge, was present here in historic concentration. He knew the test. He was running it, call by call, and passing. What he could not do was hold the result.
Fifteen years passed. The fund he left became the family office Soros wanted; Duquesne ran on, compounding, until Druckenmiller closed it too, undefeated, in 2010.² The episode settled into the anecdote everyone in the industry thinks they know.
In January 2015, at the Lost Tree Club in North Palm Beach, Ken Langone asked him the standard question, the one every veteran gets asked and every audience expects to profit from: name the biggest mistake of your career, and say what you learned from it.
He named the purchase. And then, where the format called for the lesson, he declined to supply one.
"I didn't learn anything. I already knew that I wasn't supposed to do that. I was just an emotional basket case and couldn't help myself."¹
Status at publication
28 January 2026. Druckenmiller returned investors' capital and closed Duquesne Capital Management in August 2010, and has managed his own capital through a family office since. No regulatory proceeding arose from the events described.
General information only. Not personal advice. Past performance is not indicative of future performance. This material is intended for wholesale and professional investors.
Related Practice Note
Continue with What a Reversal Must Write Down Before It Trades, a practice note on separating investment evidence, portfolio conditions, accepted carrying costs, and institutional pressure before a material position changes.
Notes
1. Stanley Druckenmiller, address and Q&A at the Lost Tree Club, North Palm Beach, 18 January 2015, per the transcript published by Cove Street Capital, March 2015. The account is retrospective throughout, delivered 15 years after the events. Supporting, in order of appearance:
about 3% a day: and the small account standing up roughly 50% on the year against Quantum's 7. The 2009 account at note 6 gives 6% to 7% a day.
the sale was total: the January 2000 decision, related to Soros in person, to sell the technology positions and step aside to await a better opportunity.
roughly three times the position: the February 1999 short of approximately US$200m against internet stocks, losing approximately US$600m by mid-March 1999 in this telling. The 2009 account gives approximately US$800m over 13 days.
The fund ended 1999 up 35%: the hiring of the two technology managers, the fund's return to the sector, and its recovery for the year. Four accounts of the 1999 path differ: this telling describes a deep early-1999 drawdown recovering to roughly 35%; the 2009 telling (note 6) gives 42% net, with the fund still down 8% in early November; the Journal's 2000 reconstruction (note 3) has the fund down about 20% by July before finishing up 35%; and contemporaneous reporting of the April 2000 press conference records Soros describing a recovery from roughly 20% down to a 33% year. No public source reconciles the four. Of them, 35% is common to this telling and the Journal's reconstruction.
104 times earnings: the valuation he attached to the sale. The transcript marks a word or phrase immediately before the figure as unintelligible.
too small to do the wider fund any damage: the retention of the two managers' account.
Three times he put the phone down: the three aborted attempts in a single week to re-enter, and the repeated self-instruction not to.
missed the top of the market by about an hour: his reckoning, and the purchase of approximately US$6bn of technology stocks that preceded it.
about US$3bn: the loss he attributed to that trade over the following six weeks.
I didn't learn anything: the closing exchange, given in answer to Ken Langone's question.
The public record does not establish that investment, valuation, or sizing work was absent from the decision process inside the firm. What survives of March 2000 is the account its participants have given publicly.
2. Contemporaneous reporting of the closure of Duquesne Capital Management, announced 18 August 2010 (Bloomberg News and The Wall Street Journal, both 18 August 2010): assets above US$12bn at closure, average annual returns of approximately 30% over the firm's roughly three-decade history, and no losing calendar year. The characterisation is consistent across that reporting and Druckenmiller's own later accounts.
3. Gregory Zuckerman, "How the Soros Funds Lost Game of Chicken Against Tech Stocks", The Wall Street Journal, 22 May 2000, published roughly ten weeks after the events it describes and sourced to people inside the firm. Supporting, in order of appearance:
technology was a bubble: the Monday afternoon research meetings through late 1999 and early 2000, held at a long table in a room overlooking Central Park; their recurring agenda of preparing for a technology sell-off, including the indicators to watch, the shares to sell, and the speed of selling; Druckenmiller's warning to the group that the turn might come soon and hit hard; and Soros's telephoned warnings to senior staff.
still carrying much of their technology book: the reconstruction's description of only limited selling in the new year, with the funds retaining substantial technology exposure as the market peaked.
I don't like this market: the early-March recommendation, attributed to people present.
The report does not date the stopped telephone calls of Druckenmiller's later accounts, which rest on those accounts alone for their placement in March; the early-March warning and the calls cannot be sequenced against each other from the public record.
4. Brett D. Fromson, "Soros Speaks", TheStreet, 29 April 2000, a partial transcript of the press conference of 28 April 2000; "So Long, Soros: An Inside Look at the End of an Investing Era", TheStreet, April 2000; Katherine Burton, Bloomberg News, 28 April 2000; Walter Hamilton, Los Angeles Times, 29 April 2000; "Top Two Managers at Soros Resign", The Washington Post, 29 April 2000. Supporting, in order of appearance:
gains a fund that left in January could not have earned: his account that he had sold some positions but not enough and not early enough, and that the year's early technology gains had for a while masked its macro losses.
the redemptions a year like this invites: the costs of selling to raise cash against potential withdrawals.
could no longer move the way the old Quantum had moved: Quantum down approximately 22% for the year to that date; the departures of Druckenmiller and of Nicholas Roditi, manager of the Quota Fund; the reorganisation of the flagship as the Quantum Endowment Fund at a lower risk profile, with capital allocated among smaller teams; and the substance of Soros's letter to investors on the fund's size and visibility. The letter is available in substance through that coverage rather than as a primary document.
I overplayed my hand: the remark as reported.
US$5bn ... US$2bn: his estimates, per the partial transcript, on the firm's technology investing over its life.
an eighth-inning affair when it was in its ninth: his statement that he should have sold in February, when the exuberance was already visible.
The April accounts and the retrospective tellings differ on the completeness of the January exit. These fund-level and firm-level accounts do not isolate the economics of the approximately US$6bn purchase on a common basis.
5. Federal Reserve Bank of St Louis, FRED series NASDAQCOM, Nasdaq Composite Index, daily closing data sourced from Nasdaq, Inc.: the record closing high of 5,048.62 on 10 March 2000, with a lower close in the following session; the index did not close above that level again until April 2015. It fell approximately 39% over calendar 2000 and approximately 78% from peak to its October 2002 trough. The intraday high of 10 March 2000 was above the closing figure. No public source dates the US$6bn purchase; its placement near 10 March is an inference from Druckenmiller's own reckoning that he missed the top by about an hour.
6. Stanley Druckenmiller, in conversation with Jeff Feig for the Citi Foreign Exchange and Local Markets team, 2009, in two independently published transcriptions on Substack: Kevin Gee, "Letter #300: Stan Druckenmiller and Jeff Feig (2009)", A Letter a Day, 8 October 2025, whose publisher states the transcription and any errors are his own; and Arya Deniz, "Stan Druckenmiller and Jeff Feig (2009): FXLM Fireside Chat", 19 October 2025, in condensed form. Neither has been verified against an authenticated recording; the second transcriber states the transcription was made because video recordings of the conversation uploaded publicly are repeatedly removed. Supporting, in order of appearance:
13 new highs against 242 new lows: the market internals he recalled behind the sale, with the advance carried by 10 or 15 names. The condensed version does not reproduce the breadth figures, which rest on the fuller transcription alone.
he sold everything and sat flat, with the fund up around 13%: his description of the January 2000 exit.
the column marked what changed: his account of the re-entry, including the interval of a few weeks out during which the two managers gained a further 30% or so; the framing of the episode as a complete loss of cool and discipline; and the fund's fall of about 18% in roughly six weeks.
bullish for an equity chart: the momentum charts he described as bullish at the re-entry and qualified in the same sentence, a shape he had seen commodity charts trace before falling, on a system built on rate of change and second derivative; and his standing rule against investing on a chart alone. The passage rests on the fuller transcription alone.
This telling places the onset of the 1999 episode in March of that year without separately dating the short; it does not quantify the March 2000 purchase and does not place it against the market's peak, both of which rest on the 2015 account alone. Its other figures differ from that account as recorded at note 1. A later account is consistent in substance: Stanley Druckenmiller in conversation with Iliana Bouzali, Hard Lessons, episode 4, "Stan Druckenmiller: Invest, Then Investigate", an original Morgan Stanley series, recorded 30 January 2026 and published 27 February 2026, per the video and extended-audio transcripts published by Morgan Stanley, in which, asked for a lesson learned the hard way, he cites having played the Nasdaq melt-up of 1999, sold in January, and then bought the exact top; says he learned nothing from it, having known not to do it two decades earlier; and attributes the trade to emotion. That account likewise supplies no analytical basis for the purchase.
7. Julian Robertson, letter to the investors of Tiger Management, 30 March 2000, announcing the return of outside capital, as reproduced in contemporaneous reporting (The New York Times and The Wall Street Journal, 31 March 2000), including his statement that earnings and price considerations had taken a back seat to "mouse clicks and momentum" and that he would not expose investors to a market he did not understand. Tiger's assets are contemporaneously reported at approximately US$22bn in mid-1998 and near US$6bn at closure, after redemptions of roughly US$7.7bn over the prior eighteen months; the fund lost approximately 19% in 1999 against a rising index, with further losses in early 2000. The timing of the closure against the market's turn is not a complete account of Tiger's decline, to which specific positions, including US Airways, and the fund's size also contributed on the contemporaneous record.
