This repo now holds only EA (MQL5) sources. The Python research scripts and the
third-party MQL5/PDF reference material live in a sibling workspace folder,
..\Warrior_Research\, with their own git repo (initial commit ec2214a there).
Nothing in the EA depends on either folder at build or run time, and the research
scripts address ..\Market Data\ and the MetaTrader Common\Files directory by
absolute path, so the relocation breaks no path. EA comments that cite scripts by
name (research/edge.py, research/altdata/export.py, research/test_spread.py, ...)
stay accurate - only the parent folder moved.
.gitignore drops the two rules that only existed for the moved trees
(references/*.pdf, research/edge_rows.npy); they were carried over to the new repo.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
CFirstPassageLadder (RungFor/StoreBar/FirstTouch/OutcomeR/WinShare),
CTripleBarrier+CLabelOverlap (ApplyMinStopWidening/ComputeLevels/SnapToLadder,
the label-overlap effective-sample-size correction), CMetaFamilies (the
classic-pattern taxonomy + table-naming rule), SGeometryScan::Reset() (guards
against 7452bd1's partial-reset shape recurring) and System/BinomialStats.mqh
(every deploy-gate/edge-floor formula this codebase shares). Each is a small
.mq5 Expert Advisor under Tests\ printing PASS/FAIL per assertion via
Print(), sharing Tests\TestHarness.mqh. All 5 self-compile-verified 0
errors/0 warnings. FirstPassageLadder.mqh/TripleBarrier.mqh expect
BARRIER_LADDER_COUNT/BARRIER_LADDER/BARRIER_HORIZON_LADDER_COUNT predefined
by their includer (normally ExpertSignalAIBase.mqh); the test EAs define
copies matching production values rather than including the whole AIBase
chain. SGeometryScan is reproduced verbatim from ExpertSignalAIBase.mqh for
the same reason, flagged in-file as needing to stay byte-identical.
Also adds Tests\convert_sample_data.py, which runs research/sqxbars.py's
decoder against a COPY of SP500_the5ers_H1.dat (never the SQX install
itself) so the operator has real sample data to point a manual tester run
at. Decodes structurally (52,542 H1 bars, monotonic, 0 high<low violations)
without calibrating a price scale - sqxbars.load()/sqx.calibrate_decimals()
both require a validated reference series to do that safely, which this
self-contained script does not have. Tests\sample_data\ (the raw copy +
decoded .npz) is gitignored, same policy as the existing Market Data/ rule.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
40 MB of third-party copyrighted books sit in references/ so the research
scripts can read them. Untracked is not the same as safe: one broad
`git add -A` puts them in history permanently and on a public remote.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A 2.1 GB tick .dat was committed in c7e9777 by a broad 'git add -A' and removed again in
6bb3386 - but a delete does not remove the blob from history, so the 2096 MB object is
still reachable and still gets pushed. That is what made syncing hang.
Ignoring the paths only prevents a recurrence; clearing the existing blob needs a history
rewrite, which is the user's call since it rewrites pushed commits.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
- Introduce per-bar feature cache to avoid redundant recomputation of input vectors during training.
- Rename EnsureLabelCacheCapacity to EnsureBarCachesCapacity to reflect management of both label and feature caches.
- Fix oversampling logic to maintain balanced representation among minority classes, replacing independent 5x caps that caused relative bias.
