[Mine-PA] Liquidity sweep reversal with trend
public
@djgambhirBTCUSDT · 4h · 2021-12-31 → 2026-08-31 · 10,228 bars in 9.7ms
Underperformed the market. Holding BTCUSDT returned +63.72% over the same period, against this strategy’s -15.04%.
| Segment | Period | Return | Max drawdown | Trades | Win rate |
|---|---|---|---|---|---|
| Training | 2022-01-01 → 2024-10-19 | -12.67% | -17.44% | 118 | 34% |
| Validation | 2024-10-19 → 2025-09-25 | -2.70% | -5.34% | 44 | 30% |
| Out-of-sample | 2025-09-25 → 2026-08-31 | -0.02% | -5.40% | 45 | 31% |
Risk-adjusted
- Sharpe
- -0.52
- Sortino
- -0.73
- Calmar
- -0.20
- Return ÷ max drawdown
- -0.86
- CAGR
- -3.43%
- Exposure
- 30.2%
Per trade
- Expectancy
- −$7.27 · -0.30%
- Average win / loss
- $72.72 / −$45.54
- Payoff ratio
- 1.60
- Largest win / loss
- $371.24 / −$124.28
- Longest losing streak(winning: 4)
- 10 trades
Drawdown over time
- Max drawdown
- -17.45%
- Longest time under water(still going at the end)
- 1684 days
- Worst drawdown recovered in
- not recovered
Costs and mechanics
- Fees paid
- −$454.15
- Funding
- −$134.59
- Liquidations
- 0
- Warmup bars(loaded before the start date)
- 199
- Benchmark(buy and hold)
- +63.72%
Monte Carlo
1,000 alternative histories, each 207 trades drawn at random from this run’s own. The bands hold the middle 50% and 90% of them; the solid line is what actually happened.
| Unlucky (5%) | Typical | Lucky (95%) | |
|---|---|---|---|
| Return | -30.85% | -15.33% | +2.85% |
| Max drawdown | -32.93% | -20.48% | -10.28% |
91% of histories lost money, and 66% had a deeper drawdown than this run’s -17.45%. Assumes trades are independent; real losing streaks cluster, so the true spread is usually wider.
Settings
- ll20 length
- 20
- hh20 length
- 20
- ema200 length
- 200
- atr14 length
- 14
- Stop loss (× atr14)
- 1.5
- Trailing stop (× atr14)
- 3
- Leverage (×)
- 1
- Position size (% of equity)
- 25
BTCUSDT with every trade
Trades
207 trades · 201–207 shown
| # | Side | Entry | Exit | Held | P&L | Return | Closed by |
|---|---|---|---|---|---|---|---|
| 201 | short | 2026-05-26 16:00 @ 76,462.90 | 2026-06-07 20:00 @ 63,887.26 | 12d 4h | $346.97 | +16.53% | trailing stop |
| 202 | short | 2026-06-12 12:00 @ 63,716.65 | 2026-06-14 20:00 @ 65,269.72 | 2d 8h | −$55.62 | -2.54% | trailing stop |
| 203 | short | 2026-06-22 04:00 @ 63,922.31 | 2026-06-22 12:00 @ 65,024.19 | 8h | −$39.60 | -1.82% | trailing stop |
| 204 | short | 2026-07-02 00:00 @ 59,987.60 | 2026-07-02 12:00 @ 61,494.35 | 12h | −$56.21 | -2.60% | trailing stop |
| 205 | short | 2026-07-03 16:00 @ 61,882.82 | 2026-07-04 16:00 @ 63,144.88 | 1d | −$45.26 | -2.11% | trailing stop |
| 206 | short | 2026-08-03 20:00 @ 63,824.13 | 2026-08-05 12:00 @ 64,737.41 | 1d 16h | −$32.31 | -1.51% | trailing stop |
| 207 | short | 2026-08-17 08:00 @ 63,480.80 | 2026-08-17 12:00 @ 64,004.08 | 4h | −$19.53 | -0.92% | trailing stop |
How this was run
Orders fill at the next bar’s open, never at the close that produced the signal. When one bar’s range covered both the stop and the target, the stop was taken, because the data cannot say which came first. Fees, slippage and funding are all charged.
Warmup history was loaded from before 2022-01-01 so the requested window is fully tradable.