Bill Miller: The 15-Year Streak Killer Who Bought Amazon at $5

From 1991 through 2005, Bill Miller's Legg Mason Value Trust beat the S&P 500 every single year — fifteen years in a row, an unmatched streak on Wall Street. Miller did it by buying Amazon (AMZN) at roughly $5 during the dot-com bust and holding through a 90% drawdown most "value" investors said was a death spiral.

How did a philosophy PhD become a Wall Street legend?

The short answer is by treating stocks the way philosophers treat arguments. William H. Miller III was an unlikely fund manager. He studied philosophy at Washington & Lee in the late 1960s, served as a U.S. Army intelligence officer in Vietnam, and earned a Ph.D. in philosophy at Johns Hopkins. He read Wittgenstein for fun.

He joined Legg Mason in 1981 as research director and took over the Value Trust in 1982 alongside founder Ernie Kiehne. Miller's philosophical training mattered more than it sounds — he was constantly asking "what could make this thesis wrong?" in a way that pure finance backgrounds rarely produce.

The fund's mandate was traditional value: undervalued securities trading below intrinsic worth. Miller stretched that mandate further than any of his peers. He argued that a great business at 25 times earnings was often a better "value" than a mediocre business at 8 times earnings — because the cash flow stream of the great business was worth materially more than 25 times.

That stretched mandate produced the streak. And eventually, it produced the blow-up.

Philosophy: what does Miller actually believe?

Miller believes the only definition of "value investing" that matters is paying less for a business than its discounted future cash flows. Everything else — low P/E, low P/B, dividend yield — is a screen, not a principle.

That sounds obvious. In practice it puts you in places traditional value investors will not go.

In the late 1990s, Miller bought Amazon, America Online, and Dell (DELL) at multiples that looked nothing like Ben Graham's screens. He argued that the present value of AMZN's eventual operating cash flow — if the customer acquisition flywheel and the marginal cost economics worked — was a multiple of the current price even after the dot-com bust crushed the multiple by 90%.

The philosophical commitment was: I don't care what label the market puts on this stock; I care about the math. Most "value" investors care about the label.

The five Miller principles

Principle 1 — Intrinsic value is the only anchor. Buy when price is below your estimate of intrinsic value; sell when it is above. The size of the gap determines position size. Multiples are output, not input.

Principle 2 — Lowest-cost-basis wins. Miller is famous for averaging down. When AMZN cratered from around $107 to under $6 in 2001-2002, he bought more — repeatedly. The discipline only works if the original intrinsic-value thesis is still intact; on a broken thesis, averaging down is doubling your error.

Principle 3 — Mean reversion is the friend. Miller argued that the market's tendency to over-extrapolate creates the bargains. Stocks that look bad usually do not stay bad forever, and stocks that look great usually do not stay great forever. The corollary: the best risk-adjusted opportunities live in the names everyone else has given up on.

Principle 4 — Diversification is for people who do not know what they own. Miller ran a concentrated book. Top 10 positions were routinely roughly 50%+ of fund assets. He used the Buffett line: if you have studied the position deeply enough, more diversification adds confusion, not safety.

Principle 5 — Probabilistic thinking, not certainty. Miller would talk about a stock as having a "70% chance of being worth $80" and a "20% chance of being worth $200" and a "10% chance of going to zero." Sum the expected values, compare to the current price. The math is the trade.

The Amazon bet: how to stay long through a 95% drawdown

Amazon (AMZN) is the trade Miller is remembered for, and it is worth dwelling on.

Miller started buying AMZN in 1999 around $40-90, on the thesis that the customer acquisition flywheel and incremental margin economics were misunderstood. The stock then fell to roughly $5.51 by October 2001 — an approximately 95% drawdown peak-to-trough. Most institutional managers would have been fired by their limited partners, the head of equity, or both, well before the bottom.

Miller did not sell. He bought more. The thesis was that AMZN's revenue base could re-accelerate once the dot-com bust ended, and that the customer cohort economics would prove the marginal economics. He was right by approximately a decade — but the path to being right took the fund through a stretch where every quarterly report defending the position required intellectual stamina most professional investors do not have.

The position came back. By 2014, AMZN had a roughly 100-fold return from the 2001 lows. Miller had been forced out of the Value Trust by then, but the trade is the case study every concentrated value investor studies.

The 2008 blow-up: when the principle broke

Miller's streak ended badly. In 2007-2008 he loaded the fund with American International Group (AIG)-adjacent names, Bear Stearns (before the collapse), Citigroup (C), and other levered financials.

The thesis was classic Miller mean-reversion: financials had cratered, the franchise values were still real, and the cycle would turn. The risk he underestimated was that some of these companies were not in a cyclical drawdown — they were in an existential one. Equity holders were going to be wiped.

The Value Trust lost approximately 55% in 2008 — far worse than the S&P 500's roughly 37% decline. The streak ended; the long-term track was tarnished; Miller stepped down as fund manager in 2012.

The lesson is durable: intrinsic-value investing assumes a business survives. When solvency is the question, intrinsic value math is the wrong tool. Miller has acknowledged this in subsequent letters and interviews.

Notable trades and holdings (Miller's later career)

After Legg Mason, Miller's personal vehicle Miller Value Partners is small enough to take asymmetric concentrated bets. He has spoken publicly about being a Bitcoin bull since approximately 2014, with MicroStrategy (MSTR) as the largest equity expression of that view. Reports from 2022 onward suggested approximately half his personal net worth sits in Bitcoin and Bitcoin-adjacent equities.

He has also held Amazon (AMZN) at various points across the last decade, Apple (AAPL) at various points, and concentrated bets in Alphabet (GOOGL) and Meta (META) when each was unloved.

What can you actually learn from Bill Miller?

The honest answer is three things, and they apply to anyone with a long-term horizon.

Lesson 1 — Define what you mean by value. If "value" means low P/E, you are running a strategy that has historically struggled in regimes where intangibles dominate the economy. If "value" means price below intrinsic value, you are running Miller's framework — and you should accept that the most interesting opportunities will rarely look "cheap" by simple multiples.

Lesson 2 — Concentration is not the enemy; bad analysis is. Miller ran 50%+ in top 10 positions for two decades and produced an unprecedented streak. The same concentration produced the 2008 blow-up. The variable is the depth of the underwriting, not the position size.

Lesson 3 — Most investors quit at the bottom. The single most defining feature of the AMZN trade was the willingness to keep buying through approximately a 95% drawdown. Almost no professional investor can do this because the institutional structure forces them out. As an individual, you have the option Miller did not always have. Use it carefully.

For a deeper framework on how to evaluate the underlying intrinsic-value math Miller relied on, our fundamental analysis index has the matching guides on discounted cash flow and ROIC. For more profiles of legendary value investors, see our investment strategies index.

How big was Miller's outperformance, really?

The answer depends on whether you stop the clock in 2005 or include 2008. The 15-year streak (1991-2005) is the single best-known statistic in U.S. mutual fund history. The compounding rate of the Value Trust during that period was roughly 16% per year versus the S&P 500's approximately 11% — a meaningful spread that, over 15 years, doubled an investor's relative wealth.

Including the 2008 collapse, Miller's full Value Trust track is closer to S&P 500 returns net of fees — a reminder that one catastrophic year can erase a decade and a half of outperformance, and that risk management is not optional even for the most-respected stock picker of a generation.

The post-Legg Mason chapter is harder to benchmark because the personal fund is small and the Bitcoin position dwarfs everything else. By certain reports Miller has produced one of the better post-financial-crisis records in concentrated equity management, but it lives outside the public mutual fund track.

Ready to analyze these stocks yourself? Search any ticker on MainRatios to see valuations from 6 legendary investors - free.