Why illiquidity failed our test.

In traditional markets, harder-to-trade assets tend to pay a premium for the inconvenience. The Amihud measure captures it: roughly, how much a coin’s price moves per dollar of volume. If crypto rewards illiquidity the same way, ranking by it could be an edge. And we already had the data, so it was cheap to test.

The portfolio looked like a winner: Sharpe ≈ 0.96, t = 2.40, a cumulative line of +552% since 2019. The kind of chart that sells a product. But the ranking metric told a completely different story — the information coefficient was ≈ 0 (t ≈ −0.6), both raw and after controls. The factor wasn’t actually sorting winners from losers.

When the headline portfolio and the IC disagree this hard, the portfolio is lying. So we took it apart.

• Tail-driven. The median period return (0.47%) sat far below the mean (1.44%) — the result rode a few extreme spikes, not a consistent edge.
• Fragile. Drop just 3 of 163 periods and significance falls to t ≈ 1.72. The IC was ≈ 0 in every sub-window.
• Survivorship + size. Only illiquid coins that survived are in the sample, and Amihud is roughly inverse-size — so this was really a small-cap lottery, not an illiquidity premium.

A portfolio curve is the easiest thing in finance to fake. The IC, the tails, and the survivorship check are how you catch it.

Illiquidity does not enter the IOX score. But liquidity isn’t irrelevant — it’s essential, in a different role. Its honest home is a gate: a check on whether a coin is liquid enough to act on at all, so the score ranks coins you could actually trade. Predicting returns and being tradeable are two different jobs.

One factor passed the gauntlet — funding crowding. Illiquidity is one more reason the survivor is worth trusting.