Retired: 2026-02-05
37.5%
break-even 58.8% at its realized average win/loss
re-measured (top 10 coins · 1H · 2026-10-06): 48.64%
Percentage of trades that were profitable
0.42
re-measured (top 10 coins · 1H · 2026-10-06): 1.08
Gross profit / gross loss ratio
Negative
Net profit after all fees and losses
updated 2026-02-05
Marks next to a metric: ~ = borderline (PF between break-even and 1.5; win rate ≤5 pts above break-even, or 50–55% if none shown) · ! = below the bar, or the strategy is already retired or shelved — a closed verdict outranks its number. No mark = clears it and is still standing.
Loses in both the bull and the bear window of our backtest. 'Retired' in the strictest sense — no regime where it worked.
Card re-measured: 2026-10-06 · 958 trades analyzed
Scope note (2026-07-13): the numbers in this journal (WR 37.5 / PF 0.42) are from the original full-universe run at its registry config. The current simulator card re-measures at the top 10 coins by market cap with card defaults and shows a milder near-breakeven loss (PF 0.98). Scope changes the number — the retirement verdict holds in both.
Measurement date (added 2026-09-27): the numbers in this journal (WR 37.5 / PF 0.42; scope: full-universe run at its registry config) are from the run behind the retirement (2026-02-05; the strategy was added 2026-01-31, and the figures first appear in commit c8e787873 together with that retirement date).
Overview
Momentum Breakout LONG was designed to capture strong upward moves by entering when price breaks above the highest close of the previous 20 candles with volume confirmation and an uptrend filter.
How It Worked
- Breakout Detection — Price closes above the highest close of the previous 20 candles
- Volume Confirmation — Volume >= 2.0x average
- Trend Filter — EMA confirms uptrend
- Enter Long — With 5% SL, 10% TP (R:R 1:2)
Why It Failed
| Metric | Expected | Actual |
|---|---|---|
| Win Rate | >50% | 37.5% |
| Break-even Win Rate | 33.3% | - |
| Profit Factor | >1.0 | 0.42 |
| Total PnL | Positive | Negative |
With SL 5% / TP 10%, the break-even win rate is 33.3% (5 / (10 + 5)). At 37.5%, the strategy actually cleared that bar — and still lost money: fees and timeout exits (positions closed at market after the max hold, often at losses between SL and TP) dragged the profit factor down to 0.42. The R:R math was not the killer; the realized exit distribution and costs were.
Root Cause Analysis
- Look-ahead bias in initial simulation — Early tests showed +400% returns due to a subtle coding error (using current candle data instead of previous)
- Crypto LONG structural weakness — In the 2024-2026 bear/sideways market, momentum longs get caught in mean reversion repeatedly
- 18 LONG strategies tested, ALL failed — This wasn’t a parameter problem. LONG momentum in crypto has no edge in the current regime
Why It Was Retired
This failure was one of our most important lessons. But it was also one of our most valuable:
“If you can’t prove it survives 2+ years of out-of-sample data, don’t trade it.”
The bar at the time was “2+ years of backtest data across 535 coins (the archived pre-OKX run)”.
Scope note (2026-08-11): The coin-count bar above was written against that archive run’s universe (535 coins). After the OKX migration, the operative verified bar was updated to an OOS + walk-forward gate measured on api.pruviq.com — the coin-count threshold is no longer the gate. The principle (2+ years of out-of-sample proof) is unchanged; the specific number reflects the Binance era and does not contradict the current smaller OKX universe.
Status: RETIRED
Retired on 2026-02-05 after comprehensive 5-expert analysis confirmed negative expected value. The loss data is preserved in our Strategy Graveyard as a permanent reminder.