The bet: fade funding-rate extremes.
All five coins were underpowered (13-26 out-of-sample trades, just four windows each), and every trade-level t-statistic came in under 2 in magnitude. We cannot conclude either way. Not killed — untestable on the data we have.
Fade funding-rate extremes, run natively on the 4-hour funding cadence (deliberately avoiding the clock-mismatch that starved fib_trigger). There is a suggestive thread — the short side, fading crowded longs, showed high win rates (ADA 80%, XLM/XRP 62%) — but on 1-10 trades each, and shorts are not spot-deployable; the tradable long side lost (XRP -2.53%/trade, t -2.99). The binding constraint is our 180-day funding-history cap; testing this properly needs more history and large-cap coins.
underpowered — 13-26 out-of-sample trades per coin, only 4 windows, no large-cap controls. There are too few out-of-sample trades to publish a per-coin table — showing one would imply a fair test we did not get. This is inconclusive on the data we have, not a result we are hiding.
Starved by limited funding-rate history; untestable, not killed.
Every trade we place — wins and losses, from real money — is on the live track record. This page is the research that decides what we will and won't run.
See the live track record →