BTC ER-low Envelope Mean-Reversion
The one crypto mean-reversion edge that survives our research: fade the 4% envelope around SMA(48) on BTC 15m, but only in the ER-low (chop) regime (Kaufman ER < rolling median — the low-trend-quality state). Two-sided: long below the lower band / short above the upper band, target = revert to the MA.
- Why BTC, not ETH/SOL: the fade edge scales inversely with momentum-dominance. BTC (most mature/liquid) is the only one where it works; ETH ≈ 0 (2021-only), SOL strongly negative (pure momentum). Matches the literature — mean-reversion lives in the most efficient markets.
- Why ER-low gate: ungated the fade is a disaster (−271R at 3%). Only the chop filter flips it positive — fading in trends gets run over.
- Exit toolkit tested: initial stop (k·ATR), TP at the MA (fixed vs floating with the MA), breakeven, trailing stop. Net of fees; entry is a limit at the band.
Honest caveats: sparse (~20 trades/yr), small (~+3%/yr on 1% risk) — this is a complement sleeve, not a standalone earner. Slippage lightly modeled (fees 0.1% round-trip dominate). n≈99 over 5y → exit params NOT fine-tuned; robust pick favored over best backtest.
Equity curves — all exit variants
Summary stats
| Variant (exit config) | n | Total R | Win % | Max DD R | Ret/DD | exit mix (SL/TP/BE) |
|---|---|---|---|---|---|---|
| A base (k3, TP=MA fix, BE1R) | 97 | +14 | 59 | -7 | 1.85 | 39/42/16 |
| B TP=MA float (robust pick) | 99 | +15 | 64 | -6 | 2.72 | 31/58/10 |
| C wide stop k4 | 91 | +9 | 60 | -4 | 2.20 | 36/49/6 |
| D tight stop k2 (best BT) | 106 | +25 | 57 | -6 | 3.94 | 44/37/25 |
| E k4 no-BE (MR-pure) | 91 | +11 | 57 | -4 | 2.48 | 39/52/0 |
| F float + trail(1.5R,2ATR) | 99 | +16 | 65 | -6 | 2.80 | 31/56/12 |
Split by year (Total R)
| Variant | 2021 | 2022 | 2023 | 2024 | 2025 | 2026 | Total |
|---|---|---|---|---|---|---|---|
| A base (k3, TP=MA fix, BE1R) | +2 | +8 | +3 | +2 | -2 | +0 | +14 |
| B TP=MA float (robust pick) | +5 | +6 | +1 | +5 | -2 | +0 | +15 |
| C wide stop k4 | +2 | +7 | +1 | +1 | -1 | +0 | +9 |
| D tight stop k2 (best BT) | +13 | +9 | -0 | +6 | -3 | +0 | +25 |
| E k4 no-BE (MR-pure) | +2 | +6 | -1 | +1 | -0 | +2 | +11 |
| F float + trail(1.5R,2ATR) | +5 | +7 | +3 | +3 | -2 | +0 | +16 |
Per-variant equity (year boundaries marked)
Exit variants: A base (stop 3·ATR, TP=MA fixed, BE@1R) · B TP=MA floating (rides the
mean up — the "trailing" that adds value) · C wide stop 4·ATR · D tight stop 2·ATR (best backtest
but 2021/22-concentrated → overfit risk) · E 4·ATR no-BE (MR-pure) · F float + separate trailing
stop (adds nothing over B). Recommended: B — +15R, ret/DD 2.72, 5/6 years positive, balanced.
A separate trailing stop (F) is redundant once the TP floats with the MA. BTC 15m,
2021-04-01 → 2026-05-19, net of fees, 1% fixed-fractional sizing.