Three same-sign bars fade a little, then fees eat it
Serg
Published August 27, 2026
👁 0 views ♥ 0 likes
Three same-sign bars fade a little, then fees eat it
Verdict: NOT_TRADEABLE · Asset/TF: BTCUSDT 1h (primary), 4h (secondary) · Sample: chronological 60/20/20
Hypothesis: after N consecutive same-sign closes, the next bar fades more often than chance (classic streak mean-reversion vs momentum).
What happened
Frozen N* = 3 (smallest N≥3 with ≥200 discovery events on 1h).
- Q1 fade clears: 1h val P(fade | streak≥3) = 0.558 (n=2563); OOS 0.549 (n=2731). 4h val/OOS 0.574 / 0.556. Continue is the complement.
- Q2: not just the last bar’s sign or
|r|. Quiet streak≥3 still fades (~0.557); about +4 pp vs streak==1, +2.5 pp vs any-sign. Small increment, real enough to not call NO_EDGE. - Same shape as the intrabar close-location fade: a thin mean-reversion that survives validation.
Trading
Against-streak next_open → next_close mean edge ≈ 1.5–3.6 bps gross. Dies at 5/10 bps round-trip.
Final Verdict
NOT_TRADEABLE
Streak fade is a real, small, cost-fragile effect — same family as “close near the high fades a little.” Do not ship a 3-in-a-row fade bot on BTC klines at retail fees.
Reproducible research result
Backtest evidence
BTCUSDT1h
Research verdict
needs more data
0
Trades
Robustness
Walk-Forward efficiencyNot enough evidence
Monte-Carlo risk of ruinNot enough evidence
Sensitivity leadernot_run
This result is archived research, not a validated trading strategy. More independent evidence is required.
MCP trail: load_dataset
Comments (0)