Li Lu's highest-IRR trade was held six months. His biggest raw gain took six years
Sort Li Lu's 14 disclosed trades by raw return and by IRR and the order changes. A six-month CROX position with a modest 26 percent return jumps to the top by IRR, ahead of a six-year GOOG position that returned nearly ten times as much. The gap is annualization, not performance.
Li Lu's disclosed record lists CROX, held for 6 months, returning 26.03 percent. It also lists GOOG, held for 6 years, returning 250.36 percent, nearly ten times as much. Rank the same 14 trades by IRR instead of raw return and CROX comes out on top at 58.83 percent, ahead of GOOG's 23.24 percent. The trade with the smaller gain outranks the trade with the much larger one, on the same investor's own published numbers.
The mechanism is arithmetic, not a mystery. IRR annualizes a return over however long the position was held, which rewards a fast gain and does not reward a slow one, even a much larger slow one. CROX's 26.03 percent arrived in half a year, so compounded annually that pace works out to 58.83 percent. GOOG's 250.36 percent took six full years to arrive, so spread across that much longer stretch its annualized pace is 23.24 percent. Divide 250.36 percent by six years, roughly, and the number stops looking small. The IRR figure is not wrong. It is answering how fast a trade grew, not how much it grew.
- CROX: 26.03% return, 6 months held, 58.83% IRR
- GOOGL: 233.75% return, 4.0 years held, 35.16% IRR
- GOOG: 250.36% return, 6.0 years held, 23.24% IRR
- EWBC: 95.40% return, 3.25 years held, 22.89% IRR
- AAPL: 100.52% return, 5.5 years held, 13.49% IRR
- BAC: 62.13% return, 6.25 years held, 8.04% IRR
The same gap shows up in the aggregate numbers
Li Lu's published weighted return across all 14 trades is 65.49 percent. His median IRR is 7.66 percent, low enough that he does not appear on Tenachine's top ten ranking by median IRR, where the tenth spot, Guy Spier, sits at 10.4 percent. The same mechanism driving the CROX and GOOG gap drives this one. Li Lu's median holding period across his disclosed positions is 2.625 years, long enough that a strong multi-year gain like GOOGL's or GOOG's compresses into a moderate annualized figure, while his weighted return keeps counting the full size of the gain regardless of how long it took to arrive. Tenachine's guide on Li Lu's Kelly bet size already covers his win rate and payoff ratio from these same 14 trades. Neither of those numbers involves time at all, which is exactly why they land on a different answer than IRR does.
Nothing here says CROX was the better trade or that IRR is the wrong number to check. A short, fast gain and a long, large one are both real outcomes, and a reader deciding what to make of either has to know which question a figure is answering before comparing it to another figure that answers a different one. Tenachine's guide on reading a 13F filing's biggest bets makes a related point about holding period on its own telling a reader nothing about whether a position was a good one. This is a sharper version of the same warning: holding period does not just sit next to a return figure, it is baked directly into IRR's math, in a way it never touches a raw return or a weighted return at all.
Fourteen trades is also a small sample to draw firm conclusions from either way. Nine winners and five losers is enough to compute a win rate and an IRR, and it is not enough to treat either one as a stable description of how Li Lu trades going forward. Past performance does not predict future results, and a single six-month trade sitting at the top of an IRR ranking is not evidence that fast trades work better than slow ones for this investor or anyone else. It is evidence that IRR and total return measure different things, and checking which one is doing the talking is worth the extra step before repeating either number.