Files
Brent NealeandClaude Opus 4.6 c8ec99c4e4 Add 2025 forward validation report with full trade analysis
Comprehensive report comparing IS (2021-2022), OOS (2023), and 2025
performance across all 4 Phase 2 strategies. Includes per-trade logs,
period comparison tables, and root cause analysis for each strategy's
2025 performance. Key finding: only S3 survived, high Phase 2
generalization scores correlated with regime sensitivity not robustness.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-21 12:08:14 +10:00

21 KiB

2025 Forward Validation Report

Overview

This report presents the results of running the 4 Phase 2 passing strategies on 2025 data — a true forward test on data never seen during development or parameter tuning.

Strategies tested:

  • S7_Tight — Liquidity Sweep (GBP_JPY, H1)
  • S9_Filtered — London Session Breakout (GBP_AUD, H1)
  • S3 — Key Level Breakout (GBP_JPY, H1)
  • S8_OB — Order Block Retest (GBP_USD, M15)

Data: Downloaded from Dukascopy. Oct 2024 used for indicator warmup; trades counted from Jan 1 - Dec 31, 2025.

Phase 2 periods for reference:

  • In-Sample (IS): Jan 2021 - Dec 2022
  • Out-of-Sample (OOS): Jan 2023 - Aug 2023

1. Portfolio Summary

Strategy Pair TF IS PF OOS PF 2025 PF 2025 Trades 2025 WR% 2025 PnL (pips) 2025 Verdict
S7_Tight GBP_JPY H1 1.52 1.80 0.52 18 55.6% -293.6 FAIL
S9_Filtered GBP_AUD H1 1.31 2.26 0.74 18 55.6% -112.2 FAIL
S3 GBP_JPY H1 1.22 1.23 1.05 35 42.9% +52.2 PASS
S8_OB GBP_USD M15 1.39 1.59 0.84 19 31.6% -19.9 FAIL
Portfolio 1.31 1.55 0.83 90 45.6% -373.4 FAIL

2. Detailed Strategy Comparison Across Periods

2.1 S7_Tight — Liquidity Sweep (GBP_JPY H1)

IS (2021-2022) OOS (2023) 2025
Trades 23 15 18
Win Rate 69.6% 73.3% 55.6%
Profit Factor 1.52 1.80 0.52
PnL (pips) +207.2 +200.1 -293.6
Avg Win +37.8p +41.0p +31.2p
Avg Loss -56.9p -62.6p -75.8p
Reward:Risk 0.67 0.65 0.41
Biggest Win +85.2p +94.8p +58.0p
Biggest Loss -84.1p -73.7p -133.1p
TP3 Hits 9 5 3
TP1+Partial Exits 7 5 6
SL Hits 7 4 8
TIME Exits 0 1 1
LONG trades (WR) 9 (78%) 11 (82%) 14 (57%)
SHORT trades (WR) 14 (64%) 4 (50%) 4 (50%)
Avg Hold (min) 798 1120 733
Avg Confluence 3.8 3.5 3.7
Avg ATR at Entry 0.2536 0.2808 0.3216
Avg ADX at Entry 31.6 27.3 29.1

2025 Trades:

# Date Dir Entry Exit Reason PnL (pips) Hold (min) Confluence
1 2025-01-06 LONG 196.795 TP1+SL +20.5 240 3
2 2025-01-20 SHORT 190.368 SL -64.0 300 4
3 2025-01-23 LONG 192.630 TP3 +58.0 600 5
4 2025-02-18 LONG 191.401 TP1+SL +21.6 660 4
5 2025-02-26 SHORT 189.275 TP1+SL +27.6 960 4
6 2025-03-06 LONG 191.106 SL -133.1 1560 4
7 2025-03-13 LONG 192.018 SL -104.8 480 4
8 2025-03-21 LONG 193.465 SL -85.8 240 4
9 2025-04-10 SHORT 187.113 TP1+SL +48.6 1020 3
10 2025-05-07 LONG 191.644 SL -60.6 360 3
11 2025-09-25 SHORT 199.696 SL -45.3 1020 4
12 2025-10-09 LONG 204.387 SL -69.8 240 3
13 2025-11-24 LONG 205.687 TIME +8.3 2400 3
14 2025-12-03 LONG 206.666 TP3 +49.9 720 4
15 2025-12-23 LONG 210.611 TP1+SL +17.9 360 4
16 2025-12-24 LONG 210.406 TP1+SL +18.4 960 4
17 2025-12-30 LONG 210.740 SL -42.7 840 3
18 2025-12-31 LONG 210.492 TP3 +41.7 240 3

2.2 S9_Filtered — London Session Breakout (GBP_AUD H1)

IS (2021-2022) OOS (2023) 2025
Trades 42 9 18
Win Rate 59.5% 66.7% 55.6%
Profit Factor 1.31 2.26 0.74
PnL (pips) +299.5 +217.8 -112.2
Avg Win +50.3p +65.2p +32.7p
Avg Loss -56.3p -57.7p -54.9p
Reward:Risk 0.89 1.13 0.60
Biggest Win +144.8p +128.3p +63.5p
Biggest Loss -73.9p -77.5p -79.1p
TP3 Hits 8 3 3
TP1+Partial Exits 17 3 7
SL Hits 16 3 8
TIME Exits 1 0 0
LONG trades (WR) 19 (58%) 6 (67%) 8 (50%)
SHORT trades (WR) 23 (61%) 3 (67%) 10 (60%)
Avg Hold (min) 937 893 600
Avg Confluence 4.8 4.8 4.9
Avg ATR at Entry 0.0028 0.0030 0.0023
Avg ADX at Entry 33.8 38.1 35.2

2025 Trades:

# Date Dir Entry Exit Reason PnL (pips) Hold (min) Confluence
1 2025-01-14 SHORT 1.96987 TP1+SL +20.5 540 5
2 2025-01-15 SHORT 1.96823 TP1+SL +21.7 1080 5
3 2025-02-05 SHORT 1.99311 TP1+SL +22.5 1080 5
4 2025-02-11 SHORT 1.96452 SL -52.8 120 5
5 2025-02-12 LONG 1.98255 TP3 +63.5 1320 4
6 2025-02-13 LONG 1.99095 SL -55.8 1140 5
7 2025-02-25 LONG 1.99364 TP3 +51.0 360 5
8 2025-03-19 LONG 2.05049 SL -58.7 660 5
9 2025-05-27 LONG 2.09807 TP1+SL +23.8 480 5
10 2025-08-21 LONG 2.09756 SL -60.1 240 5
11 2025-08-28 SHORT 2.07018 TP1+SL +12.8 300 5
12 2025-09-02 SHORT 2.05877 TP1+SL +19.6 240 5
13 2025-09-11 SHORT 2.04354 SL -39.0 300 5
14 2025-09-25 SHORT 2.03833 SL -43.0 300 5
15 2025-10-08 LONG 2.04341 SL -50.3 300 5
16 2025-11-12 SHORT 2.00723 TP1+SL +28.4 1620 5
17 2025-11-25 LONG 2.03384 TP3 +63.0 300 5
18 2025-12-17 SHORT 2.01252 SL -79.1 420 5

2.3 S3 — Key Level Breakout (GBP_JPY H1)

IS (2021-2022) OOS (2023) 2025
Trades 84 32 35
Win Rate 53.6% 59.4% 42.9%
Profit Factor 1.22 1.23 1.05
PnL (pips) +359.2 +135.5 +52.2
Avg Win +43.6p +37.8p +68.8p
Avg Loss -41.1p -44.8p -49.0p
Reward:Risk 1.06 0.84 1.40
Biggest Win +219.3p +94.4p +138.3p
Biggest Loss -136.4p -59.4p -75.5p
TP3 Hits 13 3 8
TP1+Partial Exits 32 16 7
SL Hits 39 13 20
TIME Exits 0 0 0
LONG trades (WR) 51 (51%) 20 (65%) 18 (44%)
SHORT trades (WR) 33 (58%) 12 (50%) 17 (41%)
Avg Hold (min) 1145 818 1106
Avg Confluence 4.3 4.3 4.5
Avg ATR at Entry 0.2846 0.3391 0.3467
Avg ADX at Entry 34.6 35.7 32.7

2025 Trades:

# Date Dir Entry Exit Reason PnL (pips) Hold (min) Confluence
1 2025-01-02 SHORT 194.497 SL -48.7 180 5
2 2025-01-10 SHORT 193.326 TP3 +105.7 3660 5
3 2025-01-15 SHORT 191.684 TP3 +102.2 720 5
4 2025-01-16 SHORT 190.121 SL -36.2 60 5
5 2025-01-24 LONG 194.435 SL -69.4 3480 4
6 2025-02-05 SHORT 190.408 TP3 +106.0 1260 5
7 2025-02-21 SHORT 189.529 TP1+TP2+SL +72.3 3420 4
8 2025-03-12 LONG 193.040 SL -73.4 60 4
9 2025-03-17 LONG 193.069 TP3 +86.8 1140 5
10 2025-04-01 SHORT 192.294 SL -53.6 120 4
11 2025-04-03 SHORT 191.928 TP3 +138.3 1320 4
12 2025-04-08 SHORT 187.910 TP1+TP2+SL +125.8 660 4
13 2025-05-13 LONG 196.010 SL -39.1 60 4
14 2025-05-13 LONG 196.058 SL -45.5 840 4
15 2025-05-20 SHORT 193.229 SL -37.6 120 4
16 2025-05-23 SHORT 192.468 SL -38.2 180 4
17 2025-06-05 LONG 195.007 TP1+SL +18.8 180 4
18 2025-06-06 LONG 195.787 TP1+SL +16.4 5340 5
19 2025-06-13 LONG 195.670 SL -51.3 240 4
20 2025-06-20 LONG 196.536 TP3 +61.7 3960 4
21 2025-06-30 LONG 197.968 SL -46.7 420 4
22 2025-08-01 SHORT 197.420 TP3 +76.6 120 5
23 2025-08-14 LONG 199.554 TP1+SL +16.5 720 5
24 2025-08-29 SHORT 198.084 SL -28.7 60 5
25 2025-09-03 LONG 199.683 SL -50.8 120 4
26 2025-09-04 LONG 199.530 SL -37.2 540 5
27 2025-09-26 LONG 200.176 TP1+SL +9.8 3420 5
28 2025-09-30 SHORT 198.775 SL -39.8 960 4
29 2025-10-02 SHORT 197.699 SL -66.8 720 5
30 2025-10-31 SHORT 201.986 SL -38.0 180 5
31 2025-11-06 SHORT 200.678 SL -75.5 1440 5
32 2025-11-13 LONG 203.814 SL -40.4 120 4
33 2025-11-17 LONG 204.529 SL -62.6 420 4
34 2025-12-08 LONG 207.684 TP3 +48.0 1080 5
35 2025-12-17 LONG 208.258 TP1+TP2+SL +46.9 1380 5

2.4 S8_OB — Order Block Retest (GBP_USD M15)

IS (2021-2022) OOS (2023) 2025
Trades 67 17 19
Win Rate 37.3% 52.9% 31.6%
Profit Factor 1.39 1.59 0.84
PnL (pips) +223.4 +89.6 -19.9
Avg Win +31.7p +26.9p +17.1p
Avg Loss -13.6p -19.0p -9.4p
Reward:Risk 2.34 1.41 1.82
Biggest Win +84.7p +68.7p +35.4p
Biggest Loss -39.1p -37.5p -13.4p
TP3 Hits 16 4 3
TP1+Partial Exits 9 4 3
SL Hits 41 7 13
TIME Exits 1 2 0
LONG trades (WR) 25 (28%) 9 (56%) 12 (25%)
SHORT trades (WR) 42 (43%) 8 (50%) 7 (43%)
Avg Hold (min) 203 247 91
Avg Confluence 4.4 4.8 4.9
Avg ATR at Entry 0.0014 0.0015 0.0010
Avg ADX at Entry 29.7 38.2 29.2

2025 Trades:

# Date Dir Entry Exit Reason PnL (pips) Hold (min) Confluence
1 2025-01-17 SHORT 1.22070 TP3 +22.3 105 5
2 2025-01-17 SHORT 1.22051 TP3 +35.4 390 5
3 2025-02-13 LONG 1.24897 SL -5.7 135 4
4 2025-02-13 LONG 1.25058 TP1+SL +12.0 210 5
5 2025-02-18 LONG 1.26054 SL -11.5 75 5
6 2025-02-18 LONG 1.25983 SL -4.3 15 5
7 2025-02-18 LONG 1.26051 TP1+SL +7.3 165 5
8 2025-03-28 LONG 1.29491 SL -10.1 90 5
9 2025-03-28 LONG 1.29473 SL -9.3 45 5
10 2025-04-29 LONG 1.34259 SL -8.6 15 5
11 2025-09-23 SHORT 1.35107 SL -12.8 105 5
12 2025-09-23 SHORT 1.35144 SL -8.9 45 5
13 2025-09-23 SHORT 1.35120 SL -11.5 90 5
14 2025-11-07 SHORT 1.31134 SL -13.4 30 5
15 2025-11-07 SHORT 1.31222 TP1+SL +5.8 45 5
16 2025-11-18 LONG 1.31568 SL -9.3 90 5
17 2025-11-26 LONG 1.31552 SL -7.6 15 5
18 2025-12-04 LONG 1.33390 SL -9.7 15 5
19 2025-12-04 LONG 1.33364 TP3 +20.1 45 5

3. Key Findings

3.1 S7_Tight — Structural Failure

What changed: The reward:risk ratio degraded from 0.67 (IS) to 0.41 (2025). Average wins shrank from +37.8p to +31.2p while average losses expanded from -56.9p to -75.8p. ATR at entry increased from 0.2536 to 0.3216, meaning wider stop-loss distances and larger losses when hit.

Root cause: The strategy depends on a high win rate (70%+) to compensate for a structurally poor R:R. In 2025, win rate dropped to 55.6% — not enough to offset losses that are 2.4x the size of wins. TP3 full-target hits dropped from 9 (IS) to just 3 (2025), while the strategy mostly captured small partial profits (6 TP1+SL exits) before reversing into SL.

Notable: March 2025 was catastrophic — three consecutive SL hits totaling -324 pips (trades #6-8). GBP_JPY exhibited wider ranges in 2025 that exceeded the strategy's stop placement.

3.2 S9_Filtered — Compressed Winners

What changed: Average win size collapsed from +50.3p (IS) / +65.2p (OOS) to +32.7p in 2025. The biggest single win shrank from +144.8p / +128.3p to just +63.5p. Meanwhile, losses remained similar (-54.9p vs -56.3p IS).

Root cause: ATR on GBP_AUD decreased from 0.0028 (IS) to 0.0023 (2025), indicating tighter ranges. In lower-volatility conditions, the strategy still enters at the same confidence level but the take-profit targets are harder to reach. Most wins exit at TP1 (7 of 10 wins were TP1+SL partials) rather than running to TP3.

Notable: The OOS period (9 trades, PF 2.26) was likely a small-sample anomaly rather than representative performance. The IS period (42 trades, PF 1.31) was closer to reality.

3.3 S3 — Surviving Through Bigger Winners

What changed: Win rate dropped from 53.6% (IS) to 42.9% (2025), but the R:R improved from 1.06 to 1.40. Average win grew from +43.6p to +68.8p. TP3 full-target hits increased from 15.5% of trades (13/84 IS) to 22.9% (8/35 in 2025).

Why it survived: Key level breakout is based on support/resistance — a structural market feature that persists across regimes. When breakouts occur, they tend to produce genuinely large moves. The strategy lost more often in 2025, but its winners ran further, keeping it marginally profitable.

Concern: PF declined from 1.22 (IS) to 1.05 (2025), and the strategy is heavily dependent on a few large winners. Removing the top 3 trades (Apr 3: +138p, Jan 10: +106p, Feb 5: +106p) would make 2025 negative. This concentration risk warrants caution.

3.4 S8_OB — Win Rate Reversion

What changed: Win rate in 2025 (31.6%) reverted to IS levels (37.3%). The OOS win rate of 52.9% was the anomaly. Average win also compressed from +31.7p (IS) to +17.1p (2025), and biggest win shrank from +84.7p to +35.4p.

Root cause: ATR on GBP_USD dropped from 0.0014 (IS) to 0.0010 (2025) — lower volatility compressed the profit potential. The strategy's order block patterns still generate signals at the same rate, but the displacement moves after retests are smaller. Long trades were especially poor (25% WR in 2025 vs 28% IS).

Notable: Multiple instances of rapid re-entry after SL (e.g., Feb 18: three trades in one day, Sep 23: three trades in one day), suggesting the strategy doesn't adapt well to choppy conditions.

3.5 Portfolio-Level Observations

Correlation risk: S7_Tight and S3 both trade GBP_JPY H1. In months where GBP_JPY was unfavorable (e.g., March 2025), both strategies suffered simultaneously, amplifying portfolio drawdowns.

Regime dependency: Three of four strategies showed material degradation in 2025. The edges identified during 2021-2023 development do not appear to be persistent. Only the most fundamentally grounded pattern (key level breakout) retained marginal profitability.

Generalization score paradox: The strategies with the highest Phase 2 generalization scores (S9: 1.712, S8: 1.397, S7: 1.272) all failed in 2025. S3, with the lowest score (1.068), was the sole survivor. High generalization scores during a favorable OOS period may indicate regime sensitivity rather than robustness.


4. Conclusions

  1. Only S3 (Key Level Breakout) demonstrated forward viability, and even then with marginal profitability (PF 1.05) and high concentration risk.

  2. S7_Tight and S9_Filtered failed due to poor risk:reward structure — both strategies cut winners short (TP1 partials) while allowing full SL losses. This only works with very high win rates, which proved to be regime-dependent.

  3. S8_OB failed due to lower volatility — the GBP_USD ATR decreased ~30% from IS levels, compressing both winning and losing magnitudes but disproportionately affecting win rate.

  4. The portfolio as a whole lost 373 pips in 2025 with a PF of 0.83, demonstrating that Phase 2 validation alone is insufficient for live deployment.

  5. Recommendation: The Phase 2 portfolio is not ready for live trading. Future strategy development should prioritize (a) structural edge patterns over statistical edge, (b) robust R:R profiles that don't depend on high win rates, and (c) multi-year forward testing before deployment.