CAT researchOverviewEURUSDGBPUSDXAUUSDGBPCADEURCADAUDJPYGBPJPYResearch notes

entry5m v6 (v3 divergence rule, EURUSD + GBPUSD, half off at 1R + breakeven, rest at 3R) pre-registered screen — run record

Pre-registration: entry5m/docs/2026-09-06-entry5m-v6-usd-scaleout-prereg.md at commit af55f80, committed before this run; no code change (HEAD 24aa2de at run time, prereg commit af55f80 is docs only). Parents: v3 (research/2026-09-06-entry5m-v3-fixed3r-screen.md, FAIL 1/3 with EURUSD passing alone), v5 (scale-out mechanics).

Primary cell (unmodified): v3's cell plus scale_out_r = 1.0, scale_out_frac = 0.5, breakeven_after_scale = True; EURUSD and GBPUSD, same data windows as the v1 record.

Judged line (prereg section 2: both symbols must clear all five sub-criteria): FAIL — neither does.

  • EURUSD FAIL — n 474 (ok; v3 415), PF 1.11 / ex-outlier PF 1.08 (fail; v3 1.21 / 1.17), expectancy 0.050 R (fail; v3 +0.130), best month share 27.5% (ok), halves 20.7 / 3.0 R (ok). Win rate 55.1% (v3 39.8%), net R 23.7 (v3 +53.9), max drawdown 20.9 R, 207 of 474 trades took the 1R partial, 669 of 1608 orders had the 3R target beyond the zero-touch level, average hold 119.4 bars.
  • GBPUSD FAIL — n 450 (ok; v3 388), PF 1.00 / ex-outlier PF 0.97 (fail; v3 1.08 / 1.04), expectancy -0.001 R (fail; v3 +0.051), best month share n/a, net R negative (fail), halves 6.5 / -7.1 R (fail). Win rate 52.4% (v3 35.1%), net R -0.6 (v3 +19.6), max drawdown 19.0 R, 198 of 450 trades took the 1R partial, 620 of 1574 orders had the 3R target beyond the zero-touch level, average hold 124.5 bars.

The variant as specified is dead; no parameter search was run. The pre-declared confirmation on XAUUSD/EURJPY is not triggered.

Reading against v3 (same bias, same stop, same 3R target, no scale-out). The scale-out did what it did on gold: win rate up about 15 points, drawdown roughly halved, and net R down — on EURUSD from +53.9 to +23.7 R, on GBPUSD from +19.6 to −0.6 R. On this rule the edge sat in the minority of trades that ran the full 3R (and, for the 40% of orders whose 3R target lay beyond the zero-touch level, in the bias-end exits beyond 1R); halving those and stopping the remainder at breakeven gave back more than the partial saved on the losers. The 15m secondary run (2026-09-06-entry5m-v6-usd-scaleout-screen-15m.md): EURUSD 278 trades net +5.7 R PF 1.05, GBPUSD 262 trades net −20.9 R PF 0.84, FAIL.

The sensitivity block below is reported as required and is NOT judged.

entry5m screen tf=5m stop_mode=h1_extreme target_mode=fixed_r target_r=3.0 bias_source=divergence scale_out_r=1.0 scale_out_frac=0.5 breakeven=True symbols=EURUSD,GBPUSD

  symbol      n    win%      netR     expR      PF    exPF      ddR     month   share%       h1R       h2R    bars  verdict
  EURUSD    474    55.1      23.7    0.050    1.11    1.08     20.9   2024-06     27.5      20.7       3.0   119.4     FAIL
  GBPUSD    450    52.4      -0.6   -0.001    1.00    0.97     19.0   2025-06        -       6.5      -7.1   124.5     FAIL

bias / pullback funnel:
  EURUSD   biases=235  biases_RDIV=225  biases_HDIV=10  biases_ARM=0  bias_end_target=120  bias_end_contradiction=50  bias_end_timeout=28  bias_end_replaced=37  bias_end_range_end=0  bias_no_target=0  pullbacks_too_short=2113  pullbacks_qualifying=6485  rr_rejected=958  already_broken=348  skipped_in_position=3571  orders_placed=1608  cancel_extreme_taken=699  cancel_expired=2  cancel_bias_end=5  cancel_replaced=428  filled=474  target_beyond_zero_touch=669  partial_exits=207
  GBPUSD   biases=185  biases_RDIV=179  biases_HDIV=6  biases_ARM=0  bias_end_target=87  bias_end_contradiction=35  bias_end_timeout=33  bias_end_replaced=30  bias_end_range_end=0  bias_no_target=0  pullbacks_too_short=2124  pullbacks_qualifying=6535  rr_rejected=1066  already_broken=320  skipped_in_position=3575  orders_placed=1574  cancel_extreme_taken=748  cancel_expired=3  cancel_bias_end=5  cancel_replaced=368  filled=450  target_beyond_zero_touch=620  partial_exits=198

  FAIL  group usd (EURUSD, GBPUSD)
  FAIL  group metal (XAUUSD)
  FAIL  group jpy (EURJPY)
  FAIL  overall: 0 of 3 groups (need 2)
sensitivity (NOT judged): net R / PF / n per cell
cell                        EURUSD                  GBPUSD
N=2 rr=1.0     13.7   1.06    518       0.1   1.00    489
N=2 rr=1.5     13.7   1.06    518       0.1   1.00    489
N=3 rr=1.0     23.7   1.11    474      -0.6   1.00    450
N=3 rr=1.5     23.7   1.11    474      -0.6   1.00    450
N=5 rr=1.0     21.0   1.11    403       0.9   1.01    375
N=5 rr=1.5     21.0   1.11    403       0.9   1.01    375