Harwinder Sekhon 9d7f7ff4ec research: weight-change test on the recorded replay files (pre-registered; M15 RSI x0.5 passes, not shipped) ...

First use of the --record evidence: a drift-adjusted per-bar screen ranked
regime-multiplier candidates, the top 3 per timeframe went through the full
2-year replay against same-harness controls plus the 2024-06..2025-03 OOS
window under a rule written before any run. M5: all three MACD candidates
fail (-10..-46R). M15: ema_trend x2 and divergence->0 (trending) fail;
RSI x0.5 in every regime passes every check (2-yr +117.6 -> +146.1R PF 1.43,
OOS +40.6 -> +60.7R PF 1.49, 3-yr +151.9 -> +186.6R) with the caveats recorded
(2026 -4R, 3-yr max DD 15.5 -> 19.3R, added trades mostly B-grade BUYs).
Nothing shipped; the passing candidate is an owner decision.

Claude-Session: https://claude.ai/code/session_01Nc9Ninzyi18QkWezrxeBvB

2026-09-09 17:26:21 -04:00

backtest

research: weight-change test on the recorded replay files (pre-registered; M15 RSI x0.5 passes, not shipped)

2026-09-09 17:26:21 -04:00

deploy

alerts: require a trade setup, log every send, drop stale spread-gate vetoes, daily log rotation

2026-09-08 11:13:52 -04:00

docs

chore: keep only what the system needs — drop research tooling, study scripts, diagram, plan

2026-09-06 17:35:41 -04:00

pair_signals

chore: remove test-only helpers and the duplicate confirmation map (no behaviour change)

2026-09-06 16:50:21 -04:00

providers

providers: keep real news bars (snap-back bad-print rule), Oanda retry ladder, scan_kind=failed

2026-09-07 11:07:29 -04:00

scripts

backtest: --record/--record-bars Parquet recorder + 2-year M5/M15 replay with indicator weights and calendar

2026-09-09 16:41:49 -04:00

static

chore: drop unreachable branches outside the signal core (no behaviour change)

2026-09-06 17:47:45 -04:00

.gitignore

backtest: --record/--record-bars Parquet recorder + 2-year M5/M15 replay with indicator weights and calendar

2026-09-09 16:41:49 -04:00

advisory_outcome.py

audit: fix 9 defects from the 2026-09-04 hidden-bug sweep

2026-09-04 20:19:08 -04:00

agent_panel.py

chore: repo hygiene — stale gitignore entries, finished plans, superseded research, doc drift

2026-09-06 17:01:07 -04:00

analyze_signal_log.py

providers: keep real news bars (snap-back bad-print rule), Oanda retry ladder, scan_kind=failed

2026-09-07 11:07:29 -04:00

app_config.py

news archive: keep every headline + its DeepSeek score in its own DB

2026-09-08 12:35:57 -04:00

chart_patterns.py

signal_aggregator + chart_patterns: drop dead parameters and shadowed literals (no behaviour change)

2026-09-06 16:55:26 -04:00

CLAUDE.md

research: weight-change test on the recorded replay files (pre-registered; M15 RSI x0.5 passes, not shipped)

2026-09-09 17:26:21 -04:00

config.json.example

news archive: keep every headline + its DeepSeek score in its own DB

2026-09-08 12:35:57 -04:00

config.py

config: remove the never-read PlatformConfig.pip_value (rule 8)

2026-09-06 16:57:40 -04:00

dashboard_publisher.py

chore: drop unreachable branches outside the signal core (no behaviour change)

2026-09-06 17:47:45 -04:00

engine.py

signal_aggregator + engine: drop write-only counters and unreachable branches (no behaviour change)

2026-09-06 17:46:33 -04:00

llm_client.py

chore: remove dead functions and write-only state (no behaviour change)

2026-09-06 16:48:24 -04:00

main.py

chore: drop unreachable branches outside the signal core (no behaviour change)

2026-09-06 17:47:45 -04:00

market_outlook.py

chore: remove dead functions and write-only state (no behaviour change)

2026-09-06 16:48:24 -04:00

news_archive.py

news archive: fix three bugs found reviewing 301bf90

2026-09-08 13:04:05 -04:00

news_engine.py

news archive: fix three bugs found reviewing 301bf90

2026-09-08 13:04:05 -04:00

position_sizing.py

chore: remove dead functions and write-only state (no behaviour change)

2026-09-06 16:48:24 -04:00

requirements.txt

Refocus on XAU/USD: add momentum indicators, tune TP, exhaustion block

2026-05-15 10:51:51 -04:00

signal_aggregator.py

alerts: require a trade setup, log every send, drop stale spread-gate vetoes, daily log rotation

2026-09-08 11:13:52 -04:00

signal_log.py

news archive: keep every headline + its DeepSeek score in its own DB

2026-09-08 12:35:57 -04:00

telegram_bot.py

alerts: require a trade setup, log every send, drop stale spread-gate vetoes, daily log rotation

2026-09-08 11:13:52 -04:00

test_all.py

backtest: --record/--record-bars Parquet recorder + 2-year M5/M15 replay with indicator weights and calendar

2026-09-09 16:41:49 -04:00

violent_watch.py

signal_log: weight-research data points, scan_outcomes forward labels, watch provenance

2026-09-05 22:22:59 -04:00

S

Description

No description provided

28 MiB

Languages

Python 92.4%

HTML 6.6%

PHP 1%