Граф коммитов

213b3aac15 fix(chart): the reconciliation never ran - it was hooked to a sweep deployed charts do not do main AnimateDread 2026-08-27 16:16:40 -04:00
4e2cdd3aff fix(chart): reconcile the cooldown over the CHART, not over one producer's record AnimateDread 2026-08-27 16:12:49 -04:00
844aac653a fix(train): the OOS final pass ran a full epoch at an undecayed rate AnimateDread 2026-08-27 13:46:48 -04:00
0dda31534e fix(chart): purge stale member arrows even when there is no sidecar to restore AnimateDread 2026-08-27 12:09:04 -04:00
cc45451a22 fix(signal): the source override was applied before the minutes input, which could defeat it AnimateDread 2026-08-27 12:06:10 -04:00
c5b9a1ad10 fix(signal): a changed input default cannot reach an already-attached EA AnimateDread 2026-08-27 12:04:03 -04:00
09af7d5ee9 feat(signal): default the cooldown to 30 bars - 10 thinned almost nothing AnimateDread 2026-08-27 11:59:17 -04:00
3218db4a38 feat(train): ONE pass over the held-out slice at deploy, on the restored checkpoint AnimateDread 2026-08-27 11:05:43 -04:00
322c052a65 fix(chart): the persisted vote-arrow restore put back every arrow the cooldown removed AnimateDread 2026-08-27 10:58:54 -04:00
2ca32e933f fix(signal): the overlay cooldown prune ran backwards and left ONE arrow per chart AnimateDread 2026-08-27 10:34:46 -04:00
110080dc86 fix(signal): the cooldown belonged at the VOTE layer, as a filter - not per member AnimateDread 2026-08-27 10:29:12 -04:00
6308a19f27 feat(signal): make the signal cooldown tunable, and add a hard any-direction gate AnimateDread 2026-08-27 09:48:28 -04:00
8083a31754 diag(gate): move the conviction curve to the horizon that has value, and add mean-d per rung AnimateDread 2026-08-27 09:12:28 -04:00
0d9320cc87 diag(gate): the ORACLE - what a perfect caller of this label would earn AnimateDread 2026-08-27 08:21:59 -04:00
1a9b56e3b0 diag(label): expose bars-to-pivot - the confound the payoff test was missing AnimateDread 2026-08-27 08:16:47 -04:00
3372b82dfa diag(gate): the conviction curve - does payoff rise with vote magnitude? AnimateDread 2026-08-27 08:10:47 -04:00
feaadd80a2 diag(gate): split the payoff by side at the horizon that can actually resolve it AnimateDread 2026-08-27 07:48:53 -04:00
ce4f74fe2c diag(ensemble): measure how much the four members actually disagree AnimateDread 2026-08-27 07:41:42 -04:00
7500e08e17 feat(gate): the payoff number needed a zero-skill book and a by-side split AnimateDread 2026-08-27 07:36:25 -04:00
98f485b901 fix(gate): the payoff horizon ended before the pivot it was measuring AnimateDread 2026-08-27 07:19:40 -04:00
43c1b27654 feat(gate): measure what a call was WORTH, not only how often it was right AnimateDread 2026-08-27 07:14:15 -04:00
0e1e952b96 feat(topology): cap the input window at 6 bars for capacity - 588 inputs -> 294 AnimateDread 2026-08-27 06:16:36 -04:00
1e6d00e602 fix(vote): a resumed converged model passed the eligibility test and then abstained on every bar AnimateDread 2026-08-27 00:52:59 -04:00
bbe26a09fb fix(diag): the lag profile's first-ever run produced a spectacular false positive AnimateDread 2026-08-26 22:02:14 -04:00
4113afd293 fix(diag): the linear lag profile never ran once - it started at the newest bar AnimateDread 2026-08-26 21:50:52 -04:00
6c2959dc32 feat(topology): re-derive capacity once when the training pool appears AnimateDread 2026-08-26 21:42:35 -04:00
00699e1af8 fix(chart): stale combined-vote arrows survived every wipe, because two files lived outside Warrior_EA\ AnimateDread 2026-08-26 21:07:52 -04:00
afe1038d11 fix(topology): stop a training-alone size becoming permanent, and stop the keep-screen latching underpowered AnimateDread 2026-08-26 19:27:48 -04:00
a970405042 feat(pool,mi): one feature layout fleet-wide, and the keep-screen stops self-disabling on a cold start AnimateDread 2026-08-26 17:41:14 -04:00
d9092a2408 fix(vote): persist the member's skill verdict - a converged model was ruled no-skill on every restart AnimateDread 2026-08-26 16:53:16 -04:00
15b028450b fix(vote): follow the derived rung until a checkpoint exists, pin thereafter AnimateDread 2026-08-26 16:36:20 -04:00
326e314b3e diag(features): emit the keep-set as a comparable hex mask AnimateDread 2026-08-26 16:09:47 -04:00
8c1266db0b diag(mi): name which BuildMiSample exit abandoned the sample AnimateDread 2026-08-26 15:57:41 -04:00
a9e941d7ee feat(features): per-column MI keep-screen (report only) AnimateDread 2026-08-26 15:43:59 -04:00
32eb5c5f58 feat(vote): edge-over-chance currency, no-skill exclusion, checkpoint burn-in AnimateDread 2026-08-26 15:38:26 -04:00
d059780c22 fix(io): stage atomic writes to a PER-CHART temp, not a shared one AnimateDread 2026-08-26 13:48:43 -04:00
869cd1b40c fix(pool): defer the atomic promotion to the timer instead of spinning on the tick AnimateDread 2026-08-26 13:41:35 -04:00
72dba892cc fix(chart): configure the vote-arrow layer with the PINNED threshold AnimateDread 2026-08-26 12:09:36 -04:00
b1c3a898aa fix(persist): adopt the pinned threshold on load; trim the accuracy label AnimateDread 2026-08-26 11:59:25 -04:00
ad4ae58814 feat(vote): exit-on-reversal boolean, pin the threshold, retry the atomic rename AnimateDread 2026-08-26 09:53:56 -04:00
34f1e09372 feat(magic): assign the magic number once, then remember it AnimateDread 2026-08-26 09:35:58 -04:00
17270ab308 feat(trade): two books per symbol, and delete the vote exit AnimateDread 2026-08-26 09:20:35 -04:00
c6eb9085d5 feat(vote): derive the threshold instead of configuring it AnimateDread 2026-08-26 09:02:58 -04:00
51620fe9e0 fix(vote): Signal_ThresholdOpen 25 -> 15, from the sweep's own numbers AnimateDread 2026-08-26 04:24:37 -04:00
b9da557e4e diag(gate): report what the vote would score at every threshold rung AnimateDread 2026-08-26 04:17:27 -04:00
1533365a85 diag(gate): the coverage refusal contradicted itself AnimateDread 2026-08-26 01:47:03 -04:00
0f756faf2a diag(gate): name the deployability condition that actually failed AnimateDread 2026-08-26 01:14:51 -04:00
2e22e714c6 fix(pool): length-prefix the fingerprint - the cross-instrument pool was inert AnimateDread 2026-08-26 00:41:41 -04:00
994fe3899c feat(label): pivot-EVENT target replaces direction-to-next-pivot AnimateDread 2026-08-26 00:16:16 -04:00
6adb710a79 fix(binomial): correct tail calculation in BinomialUpperTailP and add tests for accuracy AnimateDread 2026-08-25 23:37:22 -04:00
0fddaeea12 fix: correct edge floor percentage calculation and logging for model training AnimateDread 2026-08-25 23:16:05 -04:00
b2784b5a4d Enhance Feature and Topology Interfaces with Bulk Operations and Cache Management AnimateDread 2026-08-25 22:51:50 -04:00
ccfe5563e3 perf(tester,bn): no sub-second timer in the tester + BN kernels on the DLL tier AnimateDread 2026-08-25 20:44:18 -04:00
bd46374954 perf(train): DLL-side mini-batch apply + 300ms slice - the era bottleneck AnimateDread 2026-08-25 19:32:15 -04:00
781ae3a702 perf(deinit): I/O-free chart cleanup, dead-panel purge, skip clean weight saves AnimateDread 2026-08-25 18:46:51 -04:00
4e4bff51d4 feat(vote): backfill the ensemble win-rate record from the overlay sweep AnimateDread 2026-08-25 18:22:12 -04:00
b13a68e411 chore(panel): trim the live vote line to direction, magnitude, voters, verdict AnimateDread 2026-08-25 15:43:22 -04:00
26fc9a1217 fix(replay): resolve labels inline - the prebuilt cache's window never overlapped the rescan AnimateDread 2026-08-25 15:22:01 -04:00
5aec69fe4c feat(vote): replay pass rebuilds a deployed model's ladder without retraining AnimateDread 2026-08-25 15:01:24 -04:00
cb1d86e477 fix(vote): persist the tier ladder - a converged model was mute after every restart AnimateDread 2026-08-25 14:26:33 -04:00
b92e233b88 fix(chart): a deployed model rescans history to rebuild its vote arrows AnimateDread 2026-08-25 13:11:36 -04:00
484a9d8b0f fix(panel,arrows): one deploy predicate, a deployed-only readout, and persist the vote arrows AnimateDread 2026-08-25 13:00:15 -04:00
ee0dc78382 chore(repo): move research/ and references/ out to ..\Warrior_Research AnimateDread 2026-08-25 11:00:24 -04:00
34180b8c9c fix(exits): never let a declined tick leave an open position unchecked AnimateDread 2026-08-25 10:52:07 -04:00
5f7bbd5d6d docs(overview): trade-management enums are placement-only, and pin ordinals AnimateDread 2026-08-25 10:10:46 -04:00
15827a6b77 refactor(trade-mgmt): remove all confidence-scaled trade management AnimateDread 2026-08-25 10:10:20 -04:00
6d48fdb4cd perf(tester): stop agents doing chart work on deinit; enforce Expert_EveryTick AnimateDread 2026-08-25 09:50:27 -04:00
1baa13c5b4 refactor(meta): remove meta-labeling entirely - RETRAIN-NEUTRAL AnimateDread 2026-08-25 09:44:52 -04:00
ad5c2542ec perf(tester): skip the signal DB in tester/optimizer, drop ExportFeaturesOnly AnimateDread 2026-08-25 09:32:44 -04:00
e035a3076d feat(pool): default cross-instrument training pooling on AnimateDread 2026-08-24 22:38:38 -04:00
b5d34a82b4 feat(panel): one live vote line, no stale era count, no per-model HUD AnimateDread 2026-08-24 22:27:41 -04:00
12d9871650 chore(comments): drop two references to the deleted drift verdict AnimateDread 2026-08-24 22:11:50 -04:00
cdc8b2d1ec fix(enumerations): remove duplicate MARKET entry and maintain trailing strategy consistency AnimateDread 2026-08-24 21:34:22 -04:00
4ad079aaed fix(topology): size the network against observations, not bars AnimateDread 2026-08-24 21:21:05 -04:00
ec1692f348 feat(mi): the screen is an alarm, not a gate AnimateDread 2026-08-24 21:01:08 -04:00
8f2164698b feat(target): delete the barrier/geometry stack - the label is the verdict AnimateDread 2026-08-24 20:42:31 -04:00
8c945bf752 feat(target): swing is the default, and tau is measured, not chosen AnimateDread 2026-08-24 19:33:37 -04:00
95503c2101 fix(target): pivot finality is an event, not a waiting period AnimateDread 2026-08-24 18:40:06 -04:00
2abca1298c fix(labels): a closed candle shifts the cache, it does not invalidate it AnimateDread 2026-08-24 18:36:16 -04:00
579e8b45ef feat(target): swing-pivot direction label, and drop the ADZigZag name AnimateDread 2026-08-24 18:26:25 -04:00
0cd20a5749 diag(geometry): the sweep must not present "least negative" as a recommendation AnimateDread 2026-08-24 17:59:49 -04:00
1882f87451 feat(geometry): price every stop/target pair on the trades the model actually called AnimateDread 2026-08-24 17:35:12 -04:00
d12b742a40 fix(deploy): print the selection score in the unit it is actually in AnimateDread 2026-08-24 17:21:57 -04:00
9883b209c7 feat(deploy): ship on positive EXPECTANCY, and let the chart draw before convergence AnimateDread 2026-08-24 16:15:06 -04:00
7075747f4a fix(training): a new best must beat the noise; the blank-chart census must name its cause AnimateDread 2026-08-24 15:58:34 -04:00
c14ffc84e2 fix(training): a yielded pass is not a finished pass - Train() must return AnimateDread 2026-08-24 15:23:20 -04:00
cbd077c679 diag(training): report the window an era ACTUALLY trains on AnimateDread 2026-08-24 15:12:05 -04:00
9c31625aae fix(training-pool): say why a peer was rejected instead of adopting nothing in silence AnimateDread 2026-08-24 14:43:12 -04:00
906c60e227 feat(training): wire TrainingPool into pass 2 - peer rows contribute gradient only AnimateDread 2026-08-24 14:00:21 -04:00
e7b4442999 feat(training): TrainingPool - cross-instrument training rows, compile-verified AnimateDread 2026-08-24 13:46:58 -04:00
4c7c063e77 chore(research): drop the scratch placebo driver, now a sqxrepl subcommand AnimateDread 2026-08-24 13:31:07 -04:00
5ac948595b research(metafilter): shuffled-outcome control, and correct the "data adds nothing" verdict AnimateDread 2026-08-24 13:16:40 -04:00
a86019c46a fix(edge): save_rows walked only the top level, and my first fix was the reason AnimateDread 2026-08-24 13:12:21 -04:00
1c2030bcb8 research(sqxrepl): measure the hidden long-beta, as a beta subcommand AnimateDread 2026-08-24 13:06:19 -04:00
5b33c8396b research(sqxrepl): three-arm placebo, and a placebo subcommand to run it AnimateDread 2026-08-24 12:54:44 -04:00
cc68a768b1 fix(sqxrepl): the "buy and hold" benchmark shared the strategy's own exit bar AnimateDread 2026-08-24 12:42:41 -04:00
37764981ea fix(sqxrepl): the weekend clock was a day early and keyed on a bar most feeds never trade AnimateDread 2026-08-24 12:35:02 -04:00
061128a88c research(metafilter): correct the claim that the geometry control carries no information AnimateDread 2026-08-24 12:31:04 -04:00
22e75741b0 research(metafilter): the geometry control, which turns the finding inside out AnimateDread 2026-08-24 12:27:04 -04:00
5c3bd493ed research(metafilter): ask whether the data can RANK trades, not call direction AnimateDread 2026-08-24 12:15:28 -04:00