›Navigation
Closing reconciliation

Asked whether 28 new forex feature ideas actually say anything the existing features do not. Came back with every one graded twice against production and once against each other, and a final board of 10 clean, 15 watch-listed, 3 out.

What it measured
QuantityValueMeaning
Candidates evaluated28full batch-5-10 cohort
Rank-similarity grading15 / 10 / 3 over 174 cellsclean / watch / terminal-duplicate-of-production
Curved-similarity grading13 / 12 / 0 on 25 survivors; 289 cells banded, 52 excluded at the n<80 floorChatterjee xi catches curved duplication correlation misses
Sibling duplication2 duplicate families, 10 duplicate edges among 25one family of 4, one of 6
Final board10 / 15 / 3 (was 13 / 12 / 3)3 clean candidates demoted for sibling redundancy
Closest near-clone pairrho 0.997 (groeneveld vs kelly)promote at most one of the two
How it ran
FiringWhat it did
00Fixed two launch breakages and started the overnight measurement sweep
01Graded all 28 on plain rank-similarity across 174 market scenarios
02Finished the stricter Chatterjee xi test that catches curved duplication
03Opened both review packages, one in the code repo and one for this board
04Asked the question the main run had not: are the survivors duplicates of each other
05Operator ruling applied: candidates duplicating each other may not carry a clean grade
What it produced
Carries forward

Promotion is the operator's: pick at most one representative per duplicate family. 6 pre-registered cards (#1, #3, #4, #14, #50, #71) remain untested on forex. Raw xi cells (179 MB) are bigblack-only; the replay path is documented in PR-A.

Why it is closed

'EVALUATION COMPLETE, PROMOTION YOURS.' (LOOP_LEDGER.json morning_handoff; iteration 3 metrics record loop TERMINAL next firing)

Lifecycle status is process state, not a judgement of the findings. Results are stated as numbers with their uncertainty. Generated between CLOSING-RECONCILIATION markers — regenerate rather than edit.

forex orthogonal-candidate evaluation · live

Forex Orthogonal-Candidate Evaluation (batch 5-10)

OPERATOR MORNING HANDOFF
EVALUATION COMPLETE, PROMOTION YOURS. 28 batch-5-10 candidates evaluated end-to-end on 2026-07-02: rho worst-cell 15 PASS / 10 WATCH / 3 BAN (cox_stuart_trend_z, hjorth_complexity, ramsey_rothman_bicovariance — terminal) · xi §B worst-cell on the 25 survivors: 13 PASS / 12 WATCH / 0 BAN. Review PR-A terrylica/mql5#134 (verdicts + evidence + ledger updates) and PR-B terrylica/opendeviationbar-py#573 (this board). Verdict tables: verdict.md here; per-candidate evidence: candidates_status.json. Raw xi cells live on bigblack only (~/eon/mql5/findings/audits/2026-06-03-.../eval_b5_10_xi_results/). Next: promote your picks -> merge both PRs -> dashboard:deploy. The 6 pre-registered cards #1/#3/#4/#14/#50/#71 remain untested-on-forex.
BAN3
PASS10
WATCH15

Candidates under evaluation (28) — BAN 3 · PASS 10 · WATCH 15

Gate of record: ρ worst-cell bands (PASS<0.85 · WATCH 0.85–0.95 · BAN≥0.95, ≥2-symbol) + h_norm degeneracy; ξ §B worst-cell bands per PR #564 (PASS xi_worst<0.50 · BAN xi_worst>0.95 ∧ breadth≥0.80 · else WATCH). Every verdict is shown with the empirical numbers that back it — no evidence, no verdict. SSoT: candidates_status.json.

cardcandidatesubstrateempirical evidence (Spearman ρ · h_norm · Chatterjee ξ)verdictupdated (UTC)
2l_kurtosis_tau4logret
ρ worst|ρ|=0.950 vs anderson_darling_a2 · med=0.871 · watch-cells=113/174
h_norm min=1.000 · alive
ξ xi_worst=0.778 · breadth=0.73 vs anderson_darling_a2 (XAGUSD_5_live_tail_asian)
WATCH
sibling-audited		2026-07-02T23:53:14Z
5moors_octile_kurtosislogret
ρ worst|ρ|=0.947 vs anderson_darling_a2 · med=0.726 · watch-cells=18/174
h_norm min=1.000 · alive
ξ xi_worst=0.733 · breadth=0.16 vs anderson_darling_a2 (XAGUSD_50_covid_crash_allsess)
WATCH
sibling-audited		2026-07-02T23:53:14Z
8kelly_decile_skewnesslogret
ρ worst|ρ|=0.841 vs mann_kendall_z · med=0.748 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.473 · breadth=0.00 vs mann_kendall_z (EURUSD_5_gold_parabola_allsess)
WATCH
sibling-demoted		2026-07-03T00:07:01Z
11katz_fdclose
ρ worst|ρ|=0.902 vs sevcik_fd · med=0.852 · watch-cells=95/174
h_norm min=1.000 · alive
ξ xi_worst=0.617 · breadth=0.29 vs sevcik_fd (EURUSD_2_eur_parity_allsess)
WATCH
sibling-audited		2026-07-02T23:53:14Z
12petrosian_fdclose
ρ worst|ρ|=0.966 vs sign_two_state_entropy_rate · med=0.522 · watch-cells=1/174
h_norm min=0.989 · alive
ξ xi_worst=0.840 · breadth=0.04 vs sign_two_state_entropy_rate (XAUUSD_5_gold_parabola_asian)
WATCH
sibling-audited		2026-07-02T23:53:14Z
15hjorth_complexityclose
ρ worst|ρ|=0.958 vs sevcik_fd · med=0.919 · watch-cells=141/174
h_norm min=1.000 · alive
ξ not run — excluded at rho stage (BAN)
BAN
rho-evaluated		2026-07-02T05:43:37Z
18vonneumann_ratiologret
ρ worst|ρ|=0.938 vs sign_two_state_entropy_rate · med=0.495 · watch-cells=2/174
h_norm min=1.000 · alive
ξ xi_worst=0.697 · breadth=0.02 vs sign_two_state_entropy_rate (XAUUSD_5_gold_parabola_asian)
WATCH
sibling-audited		2026-07-02T23:53:14Z
19bartels_rank_vn_ratiologret
ρ worst|ρ|=0.851 vs sign_two_state_entropy_rate · med=0.438 · watch-cells=1/174
h_norm min=1.000 · alive
ξ xi_worst=0.571 · breadth=0.01 vs sign_two_state_entropy_rate (XAUUSD_5_gold_parabola_asian)
WATCH
sibling-audited		2026-07-02T23:53:14Z
22difference_sign_test_zlogret
ρ worst|ρ|=0.929 vs mann_kendall_z · med=0.826 · watch-cells=64/174
h_norm min=0.985 · alive
ξ xi_worst=0.613 · breadth=0.16 vs mann_kendall_z (XAUUSD_25_gold_rally_allsess)
WATCH
sibling-audited		2026-07-02T23:53:14Z
23cox_stuart_trend_zclose
ρ worst|ρ|=0.978 vs mann_kendall_z · med=0.968 · watch-cells=15/174
h_norm min=0.827 · alive
ξ not run — excluded at rho stage (BAN)
BAN
rho-evaluated		2026-07-02T05:43:37Z
26ehlers_increment_asymmetrylogret
ρ worst|ρ|=0.813 vs mann_kendall_z · med=0.642 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.531 · breadth=0.00 vs mann_kendall_z (XAUUSD_50_gold_parabola_allsess)
WATCH
sibling-audited		2026-07-02T23:53:14Z
29costa_accel_decel_asymmetrylogret
ρ worst|ρ|=0.935 vs mann_kendall_z · med=0.879 · watch-cells=142/174
h_norm min=1.000 · alive
ξ xi_worst=0.651 · breadth=0.55 vs mann_kendall_z (XAGUSD_5_midcycle_range_allsess)
WATCH
sibling-audited		2026-07-02T23:53:14Z
39empirical_cumulative_entropylogret
ρ worst|ρ|=0.956 vs anderson_darling_a2 · med=0.770 · watch-cells=28/174
h_norm min=1.000 · alive
ξ xi_worst=0.743 · breadth=0.23 vs anderson_darling_a2 (XAGUSD_50_covid_crash_allsess)
WATCH
sibling-audited		2026-07-02T23:53:14Z
45heikin_ashi_open_gap_signohlc_bar
ρ worst|ρ|=0.756 vs lookback_twap_position_q1000 · med=0.459 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.451 · breadth=0.00 vs lookback_twap_position_q1000 (XAUUSD_50_gold_rally_allsess)
PASS
sibling-audited		2026-07-02T23:53:14Z
51overnight_gap_logret_sqlogret
ρ worst|ρ|=0.574 vs ma1_noise_variance_u2_bar · med=0.258 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.365 · breadth=0.00 vs bid_low (EURUSD_2_midcycle_range_londonny)
PASS
sibling-audited		2026-07-02T23:53:14Z
56hvg_tetradic_motif_profile_entropyclose
ρ worst|ρ|=0.782 vs hvg_clustering · med=0.596 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.486 · breadth=0.00 vs hvg_clustering (XAGUSD_10_gold_parabola_allsess)
WATCH
sibling-demoted		2026-07-03T00:07:01Z
58hvg_mean_degree_excessclose
ρ worst|ρ|=0.843 vs hvg_mean_shortest_path_length · med=0.651 · watch-cells=0/174
h_norm min=0.996 · alive
ξ xi_worst=0.417 · breadth=0.00 vs hvg_clustering (XAUUSD_10_gold_rally_londonny)
PASS
sibling-audited		2026-07-02T23:53:14Z
59hvg_forward_visibility_horizon_mean_barclose
ρ worst|ρ|=0.642 vs hvg_mean_shortest_path_length · med=0.344 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.325 · breadth=0.00 vs mann_kendall_z (EURUSD_2_midcycle_range_allsess)
PASS
sibling-audited		2026-07-02T23:53:14Z
60vg_time_directed_clustering_kldclose
ρ worst|ρ|=0.771 vs mann_kendall_z · med=0.258 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.260 · breadth=0.00 vs signal_noise_ratio_bar (EURUSD_2_midcycle_range_londonny)
PASS
sibling-audited		2026-07-02T23:53:14Z
66categorical_recurrence_rate_close_tiesclose
ρ worst|ρ|=0.813 vs sevcik_fd · med=0.652 · watch-cells=0/174
h_norm min=0.966 · alive
ξ xi_worst=0.549 · breadth=0.00 vs sevcik_fd (EURUSD_3_lowvol_grind_allsess)
WATCH
sibling-audited		2026-07-02T23:53:14Z
78hoeffding_phi_squared_midreturn_durationret_dur
ρ worst|ρ|=0.617 vs anderson_darling_a2 · med=0.187 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.284 · breadth=0.00 vs roll_spread_bps (XAGUSD_10_live_tail_allsess)
PASS
sibling-audited		2026-07-02T23:53:14Z
80gini_correlation_asymmetry_signedreturn_spreadret_spread
ρ worst|ρ|=0.552 vs mann_kendall_z · med=0.305 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.309 · breadth=0.00 vs signal_rv_two_scale_bar (GBPUSD_10_gold_rally_londonny)
PASS
sibling-audited		2026-07-02T23:53:14Z
98sign_markov_inout_strength_asymmetry_fluxsigns
ρ worst|ρ|=0.428 vs anderson_darling_a2 · med=0.079 · watch-cells=0/174
h_norm min=0.260 · alive
ξ xi_worst=0.365 · breadth=0.00 vs roll_spread_bps (EURUSD_10_lowvol_grind_allsess)
PASS
sibling-audited		2026-07-02T23:53:14Z
99groeneveld_meeden_b3_skewnesslogret
ρ worst|ρ|=0.831 vs mann_kendall_z · med=0.735 · watch-cells=0/174
h_norm min=1.000 · alive
ξ xi_worst=0.477 · breadth=0.00 vs mann_kendall_z (GBPUSD_10_lowvol_grind_allsess)
WATCH
sibling-demoted		2026-07-03T00:07:01Z
100hl_pseudomedian_minus_median_gaplogret
ρ worst|ρ|=0.902 vs mann_kendall_z · med=0.783 · watch-cells=14/174
h_norm min=1.000 · alive
ξ xi_worst=0.563 · breadth=0.02 vs mann_kendall_z (XAUUSD_50_gold_rally_allsess)
WATCH
sibling-audited		2026-07-02T23:53:14Z
107etc_effort_to_compress_logret_signsigns
ρ worst|ρ|=0.667 vs sign_two_state_entropy_rate · med=0.225 · watch-cells=0/174
h_norm min=0.938 · alive
ξ xi_worst=0.455 · breadth=0.00 vs sign_two_state_entropy_rate (XAUUSD_5_gold_parabola_asian)
PASS
sibling-audited		2026-07-02T23:53:14Z
108sequitur_grammar_compression_ratio_logret_signsigns
ρ worst|ρ|=0.424 vs sign_two_state_entropy_rate · med=0.102 · watch-cells=0/174
h_norm min=0.981 · alive
ξ xi_worst=0.297 · breadth=0.00 vs roll_spread_bps (EURUSD_2_midcycle_range_allsess)
PASS
sibling-audited		2026-07-02T23:53:14Z
113ramsey_rothman_bicovariance_gamma21_lag1_midlogretlogret
ρ worst|ρ|=0.995 vs pomeau_irreversibility_lag1 · med=0.191 · watch-cells=13/174
h_norm min=1.000 · alive
ξ not run — excluded at rho stage (BAN)
BAN
rho-evaluated		2026-07-02T05:43:37Z

Iteration ledger (every loop event, newest last)

#eventphaseverdictwhat happenedts (UTC)
00launchedsweepPENDINGS0 ground+gate+launch (2 repairs, both documented). (1) numpy 2.5.0 release broke unpinned uv resolution (antr2026-07-02T03:54:47Z
01rho-aggregated-xi-launchedsweepPENDINGThe first measurement pass over all 28 new forex candidate features finished cleanly overnight. We then graded2026-07-02T05:45:23Z
02xi-aggregatedsweepPASSThe second, stricter test on the 25 surviving candidates is done. This test (Chatterjee xi) checks whether any2026-07-02T06:44:09Z
03prs-openedsweepPASSBoth review packages are now open: one in the code repository with all the measurements and verdicts, and one 2026-07-02T23:08:32Z
04sibling-auditsweepWATCHA follow-up audit asked a question the main evaluation did not: are any of the surviving candidates duplicates2026-07-02T23:53:35Z
05sibling-demotionsweepWATCHThe operator ruled that candidates duplicating each other cannot be labeled clean. Three previously-clean cand2026-07-03T00:07:16Z