Mean reversion at maker prices
Crypto candles genuinely mean-revert — we measured it across 118,000 candles before building anything. The pattern clears maker breakeven and nothing else. Then 17,491 live maker fills answered the only open question: adverse selection ate it.
The hypothesisShort-horizon crypto candles mean-revert: after a streak of same-direction candles, the next one reverses slightly more than half the time. The edge is too small to pay taker costs — but a maker pays no fee, so a resting bid on the reversal side near each flagged candle's open might harvest it.
The test
Measurement came first: 60 days of exchange candles across six assets (~103k) plus 15.4k of our own recorded market outcomes, sliced by streak length and prior-move size. Every cell agreed — reversal runs 51% after one candle, 52–54% after streaks, stronger after big moves, on every asset and both timeframes. Momentum is the measurably wrong side. The live probe then sim-rests one bid on the reversal side of each streak-flagged candle, priced never-marketable (at the target or a tick under the live ask, whichever is lower), filling only by the crossed-book rule — and, crucially, it settles the unfilled flagged candles too, because the filled-vs-unfilled win-rate gap is a direct measurement of adverse selection.
The result
Decisive, and fast: the probe accrued 17,491 filled-and-settled maker entries in three weeks, and every lane lost. Five-minute candles: −17.7% (streak=2, 7,416 settled, 37% win rate) and −17.9% (streak≥3, 6,158 settled); fifteen-minute: −8.1% and −11.0%. The control group did exactly what the design predicted it would if fills were toxic: flagged candles where our bid never filled resolved in the reversal's favor 97% of the time (1,874 of 1,925 / 1,634 of 1,678 / 415 of 423 / 363 of 369). The reversal happens — and when it does, the ask never comes down to a resting reversal bid. The bid fills only when the streak keeps going.
The naive first version (fixed 49¢ rests that were secretly marketable) had flashed the same signature in its discarded shakedown run — 102 of 103 unfilled controls winning. The corrected mechanic, resting honestly at or below the live ask, reproduced it at 170× the sample.
Conclusion
The cleanest adverse-selection measurement this lab has produced. Mean reversion in the candle series is real and the maker fee waiver is real — and the order book charges for the pattern in fill selection exactly what it waives in fees, with interest. A resting reversal bid is a free option granted to every trader who sees the streak continue. This closes the series: six taker-side nulls, a maker-side grid at −31.6%, and now a maker-side signal-gated probe at −8 to −18%. The 5-minute crypto candles have no retail-reachable edge on either side of the book. Daemon retired 2026-08-21; data kept.