Li Lu and Burry win the same share of trades, their expectancy differs 6x
Two investors with almost identical win rates post expectancy figures nearly six times apart, and a third with the highest win rate of the three lands in the middle. The gap comes from how big a typical win is next to a typical loss, not from how often either investor is right.
Li Lu wins 64.29 percent of his disclosed trades. Michael Burry wins 64.12 percent, a gap of 0.17 percentage points, close enough to call the same. Li Lu's expectancy per trade is 40.61 percent. Burry's, on his own disclosed record, is 6.93 percent, less than a fifth as large. Two investors who win at nearly the same rate post an expectancy figure that differs by close to 6 times.
Expectancy is the average return a trade actually delivers once wins and losses are both counted, and it has not appeared in a Tenachine guide before this one. Warren Buffett sits between the other two on this figure at 14.14 percent, and he has the highest win rate of the three, 68.57 percent. Rank these three investors by win rate and Buffett leads, Li Lu is second, Burry is third. Rank the same three by expectancy and the order is completely different: Li Lu leads by a wide margin, Buffett is second, Burry is third. Win rate does not even preserve the ranking, let alone the size of the gap.
What is actually driving the gap
Median return on a winning trade against median return on a losing trade tells the real story. Li Lu's median winner returns 69.78 percent, his median loser costs 11.90 percent, a ratio of 5.86 to 1. Burry's median winner returns 16.23 percent, his median loser costs 9.69 percent, a ratio of 1.67 to 1. Buffett's median winner returns 30.46 percent, his median loser costs 21.46 percent, a ratio of 1.42 to 1, the smallest of the three despite Buffett having the highest win rate. Win rate answers how often a trade closes positive. This ratio answers how much bigger a typical win is than a typical loss, and it is this number, not the win rate, that expectancy actually tracks.
Why the sample size matters here more than usual
Li Lu's numbers rest on 14 disclosed trades, 9 winners and 5 losers. Burry's rest on 170, 109 winners and 61 losers. Buffett's rest on 140, 96 winners and 44 losers. Tenachine's guide on Li Lu's own Kelly criterion output already flags this same 14-trade record as too thin to treat its payoff ratio as a fixed fact, and the same caution applies directly here. A 5.86 to 1 median win to loss ratio built on 14 trades can move a great deal with one more disclosed position closing out. A 1.67 to 1 ratio built on 170 trades, Burry's case, is a far more stable number simply because more trades sit behind it. The size of the gap between Li Lu and the other two is real and worth naming. Whether it says something durable about how Li Lu trades, or mostly reflects a small sample that has not yet had a bad quarter show up in it, is a different question, and this data cannot answer it either way.
None of these figures cover a full trading history. A 13F filing discloses long US equity positions at one point in time, not every trade a manager made, so a position that was opened and closed between filings would never appear in this data at all, and only a subset of each manager's real activity, the positions that happened to still be open or recently closed when a filing was captured, is what these expectancy and ratio figures are built from. Past performance does not predict future results, and matching a win rate is not the same as matching an outcome, which is the plain lesson sitting underneath all three of these numbers.