Owner question: can score and confidence be combined into one number that is
easier to understand and for the system to decide on?
Research (pre-registered, nothing ships to the signal path):
- signals.grading_mode: "pair" (default, the shipped two-floor rule, byte-
identical: all five reference MD5s reproduce on both scripts), "min"
(min(|score|, 0.6*conf)), "product" (|score|*conf/100); thresholds fixed from
the control's corner points, no tuning. Helpers _final_signal_and_grade /
_transitioning_boost / conviction / CONVICTION_THRESHOLDS in signal_aggregator;
mirrored in K2 (GRADING_MODE_CODES, PF_conv_* packed from the live module);
twin refuses other values; M15 OOS MD5 equal on both scripts for both modes.
- Verdict: FAILED on M5 and M15. K-MIN moves 43-54 trades for -1.2R / -2.9R
(clauses a,b,e / a,b,d); K-PROD -29R on M15 (PF 1.29, W3 DD 28.9R), neutral
on M5. H1/H4 product +11.9R / +17.0R on W3 from extra low-score trades
(H1 PF 1.98 -> 1.75, DD doubles): hypothesis only.
- Code review (high) before the runs: 6 findings fixed, 1 recorded
(engine_config_fingerprint omits the switch under "pair", so live rows keep
their fingerprint).
Delivery (books untouched): alerts lead with the grade and print the two raw
numbers in words (main.plain_words: "Direction: strong buy · Reliability: good
· timeframes agree", cut-offs = the grade table's own) on Telegram, /status,
the CLI and the dashboard card (plain_words in the snapshot); "MTF" label
replaced by "Other timeframes"; guides updated. Suite 566 passed.
Evidence: backtest/research/conviction_campaign_2026-09-27.md
Live signals.db review (2026-09-04 → 09-25, 100 unique setups replayed on the
scans' stored candles): the stop-hunt claim is refuted (71% of stop-outs ran
≥ 1R past the stop, the same as the books' 70–72%); −25.1R of the −37.8R raw
live book were consecutive-bar re-entries the backtester's 5-bar cooldown never
takes; the cooldown view sits at the 3rd–5th percentile of the books' three-week
windows.
Delivery (book-neutral):
- telegram_bot: 5-bar alert cooldown per timeframe (_ALERT_COOLDOWN_BARS =
backtest.cooldown_bars); setup block prints the trail first as the tested
exit, TP1/TP2 as guides with their R, "Planned TP1 RR", and one validated-book
line per setup class (agent_panel_base_rates.json); the informational 100-pip
TP line is removed (owner request).
- AI block rewritten for a new trader: "AI check: the odds look better than /
worse than / about usual", the chance "in 100" against the usual odds, one
reason for and one against, no verdict word or confidence figure; rendered
whenever the verdict carries a probability. agent_panel: persona prompts ban
trading jargon (PANEL_PROMPT_VERSION v2.1-2026-09-26); the macro-source
verdict now carries its persona's p_win (every live v2 row had
panel_probability NULL). Dashboard and guides mirror the wording.
- telegram_bot: level-breach rescans skip the settlement-pause window (the
20:45 M15 bar was re-logged at 22:01/22:06/22:11 every evening).
Signal (pre-registered rule-1 campaign, CUDA twin paired with gpu_backtest.py):
- M15 grade-A SELLs blocked (sell_min_quality ["A+","B"], M15 override 6):
W2 813/+154.15R PF 1.53 vs 931/+146.14R PF 1.43, OOS +68.5 vs +60.67,
W3 max DD 20.9 vs 19.3R. PASSED a–f.
- Printed target 1 floored at 1R in every generator (signals.tp_print_floor_rr),
applied AFTER the min-RR and spread gates so no trade changes; mirrored in
cuda_signal_kernels.py K2. Label-only re-pin of all four timeframes (only
tp1/rr_planned columns move). A floor applied before the gate cost −24.6R and
was rejected.
- FAILED: M15 min_rr_ratio 1.0 (trade-identical, −2.0R). Owner's call: M5
tp_floor_rr 1.0 (+48.9R W2, max DD 10.1 → 15.9R), not shipped.
New pins: M5 W2 e599c8a8…, M15 W2 5b65ecf6…, M15 OOS 09b8a8e7…, H1 W3
a44fd19b…, H4 W3 f3587bae…; twin and reference MD5s equal on every window.
Base-rate table rebuilt from the new pins. Suite 558 passed.
Evidence: backtest/research/rr_and_sell_campaign_2026-09-26.md (plus the
2026-09-18 loss-anatomy and wrong-direction reports it builds on).
Agent panel v2: personas return p_win anchored to the book base rates with a
code-computed decision context; verdict = equal-weight mean (verdict_source
"mean"); only the macro persona runs live (agent_panel.personas ["macro"]);
prompt version and exact payloads logged; offline replay harness
scripts/panel_replay.py; base-rate table agent_panel_base_rates.json
(scripts/build_panel_base_rates.py). Backtester `strategy` column now carries
the generator's own stamp (label-only; five reference books re-pinned, all
62 other columns identical).
News corpus (after evaluating the owner's ML-NEWS scikit model against
DeepSeek: golden set 39/78 vs 78/78, out-of-sample agreement 64% against a
90% self-agreement ceiling; verdict: fallback only, not a replacement):
- NewsSentiment.xau_score + archive column xau_score keep DeepSeek's gold
score for every headline (930 of the first 1,546 DeepSeek rows had none);
an existing newsdata.db gains the column in place on first open
- headlines naming no tracked currency are scored and archived
(currencies_json []) but never returned to the platform
- _apply_llm_sentiment sends every new headline in its cycle, 25 per call;
a failed call ends the cycle's sending
Dashboard publisher never writes a database: main() calls
app_config.set_db_writes_for_process(False), honoured by signal_log_enabled()
and news_archive_enabled() (owner decision: the bot is the single writer).
Test suite: the hermetic app config now points both writers at a
suite-private temp dir — the RSS fixture tests were writing fixture
headlines into the real logs/newsdata.db — with a Module 32 test pinning it.
529 tests.
Owner decision 2026-09-15. Gone: market_outlook.py, the post-H1-scan Telegram
outlook DM, main.py --outlook, the outlooks table DDL/migration/writer in
signal_log.py, the market_outlook config block + MARKET_OUTLOOK_ENABLED, and
test Module 26. Last commit containing the feature: 2836024.
- app_config: "market_outlook" stays in _KNOWN_TOP as a TOLERATED, ignored
legacy key (oanda.account_id precedent) with one warning, so a live
config.json that still carries the block keeps booting.
- signal_log: existing DBs keep their orphan outlooks table and rows (no
DROP TABLE); fresh DBs never create it.
- tests: Module 26's 4-thread concurrent legacy-DB migration test (the
suite's only one) moved to Module 31; new test for the tolerated block.
519 -> 509 tests, 34 -> 33 modules, 0 failed.
- docs: CLAUDE.md usage/config/file-map lines and a dated state note.
Verification: W2 books (2024-09-05 -> 2026-09-05, $0.65, NumPy CPU, 0 cache
fallbacks) byte-identical before/after — M5 29120a1ee32979431d062f210d41a9d1
(781 / +146.07R / PF 1.52), M15 d185f099482a07f29502cb5a371a9bbf
(931 / +146.14R / PF 1.43).
- providers: _to_oanda_instrument / _to_tiingo_ticker / _to_td_symbol no longer
guess a wire symbol for pairs that are not registered — the registries are
the only source, so an unknown symbol raises the provider's DataIntegrityError
instead of silently fetching something else. Every scanned symbol is validated
against PAIR_INFO at import, so nothing reachable changes.
- main / telegram_bot _fmt_price: the sub-$10 five-decimal branch could never
fire for gold.
- news_engine._parse_pair: BASE/QUOTE only; the Yahoo "=X" fallback was
unreachable (its test goes with it).
- dashboard_publisher.event_to_dict: drop minutes_until (frozen at publish
time, misleading) and is_high_impact — dashboard.js never read either.
- static/dashboard.html: unused id="cal-table".
Tried and reverted: trimming the LogRow fields nothing in analyze_signal_log
reads — the tests read them to verify the loader round-trips those DB columns,
so they stay.
Suite: 460 passed, 0 failed.
Claude-Session: https://claude.ai/code/session_0156YhogLytzj9P2iH1skGbL
main.py kept a private confirmation table and passed it into the pair factory,
which honours any non-None list. For M5 that table said [M1, M15] while the
factory, telegram_bot.TF_CONFIRMATIONS and dashboard_publisher.CONFIRMATIONS
all use the validated [M15, H1] - so `main.py --timeframe M5` scanned with an
unvalidated pairing (and an extra M1 fetch) that no other surface used.
- build_config now returns PAIR_PRODUCERS[pair](tf, None): the factory map is
the single source of truth. The unreachable bare-PlatformConfig fallback is
gone here and in telegram_bot._run_signal_sync (every symbol that reaches
either is registered).
- --timeframe offers the four validated primaries (M5, M15, H1, H4). M30 and
D1 were never validated and silently fell through to the generic [M15, H4]
map; they are removed from the CLI.
- The "all valid timeframes" test now asserts each CLI timeframe's
confirmation list equals pair_signals.xauusd._CONFIRMATIONS - the check
that would have caught this.
CLI behaviour change: M5 confirmations [M1, M15] -> [M15, H1]; M30/D1 rejected.
Bot and dashboard scans are unchanged.
Suite: 461 passed, 0 failed.
Claude-Session: https://claude.ai/code/session_0156YhogLytzj9P2iH1skGbL
New violent_watch.py assesses each scan from the cached Forex Factory week events, the
UTC clock and the report's own indicator states, and stamps SignalReport.watch_line
post-hoc (telegram alerts, CLI, dashboard). A standalone heads-up goes to the owner on
the fastest scanned timeframe when the level is "hot", once per UTC hour per trigger
set, capped per day. Config block `violent_watch` (enabled False, alert_line, headsup,
max_headsup_per_day, headsup_min_level); env VIOLENT_WATCH_ENABLED.
Rule chosen from a 3-year bar-by-bar sweep (backtest/research/watch_rule_eval.py,
recorded in sizing_campaign_2026-09.md): hot = high-impact USD release within 30 min
(before or after) OR a hot clock slot (weekday 12-14 UTC, 01 UTC, Sunday 22-23 UTC)
with at least one price-extreme reading (outside Bollinger bands, RSI <30/>=70, outside
Keltner); ~8.6% of bars, next-bar odds 3-5x normal, ~4 hot hours/day. watch = clock
slot alone or two price-extreme readings (~2x). Context readings (ADX rising, extreme
volatility, transitioning, MACD acceleration) are phrases only - letting them raise the
level flagged 27% of bars at ~2x and would have sent ~13 heads-ups/day (rejected).
Direction sentence: DOWN when bearish readings dominate (70-78% on both TFs), UP on M5
only. Never a gate or exit: every open-trade protection tested lost money.
Test-locked out of analyze_from_dataframes / engine / backtester. Module 30 (18 tests).
Suite: 453 passed. Guide PDFs not rebuilt.
Claude-Session: https://claude.ai/code/session_01RWV2yLt3YXHHGVYHSQU2EE
agent_panel.verdict_source defaults to "macro": the macro take's stance is the
verdict, the synthesizer call is skipped (4 LLM calls per panel instead of 5);
"chair" restores the four-voice consensus + OPPOSE guard. Verdict source is
cached per mode and logged (panel_verdict_source). 406 tests.
Claude-Session: https://claude.ai/code/session_01FC4s6VCURFte4dub4xKxYL
New signal_log.py logs each SignalReport to logs/signals.db via a
fail-safe hook in SignalAggregator.analyze_pair — the live-only entry
point used by the bot, daemon, CLI and dashboard. Backtester calls
analyze_from_dataframes directly, so live==backtest parity is preserved
(test-locked). Rows carry signal/grade/price/spread, MTF + trade-setup +
reasons/warnings JSON, and news-at-scan-time (calendar events + scored
headlines JSON via NewsEngine class-level caches).
Config: signal_log block (enabled, path; env SIGNAL_LOG_ENABLED/PATH),
source tagged per consumer (telegram/oanda_daemon/cli/dashboard).
381 tests pass (24 modules).
main.py: --panel runs the agent panel on demand (force bypasses
enabled/surface/quality toggles; daily budget + LLM cap still apply),
prints the verdict + persona takes, and report_to_dict carries the four
panel keys — which auto-propagates them into the dashboard's
snapshot.json via report_to_webapp_dict. dashboard_publisher attaches
in scan_tf gated by agent_panel.dashboard (default off); dashboard.html
renders an "AI panel (advisory)" section in the TF detail dialog.
Claude-Session: https://claude.ai/code/session_01SisdfkxnXNL9iZsXxN3T15
Three display/warning-only upgrades to the news engine, all behind the
existing keyword path as fallback — signals are never gated on LLM
output (event_risk_caps_signals / sentiment_affects_signals stay False
per the 2026-07 audit; backtester use_news=False untouched):
- Headline sentiment: one batched call per 600s cache window (capped at
25 new headlines, sha1-cached so repeats are never re-sent) overrides
the keyword scores per currency with the gold-inverse doctrine encoded
in the prompt.
- Release surprises: get_event_risk_level's description now says which
way the print pulls GOLD ("— gold bearish: strong jobs lift yields"),
fixing the documented numeric-only gap; risk level and windows
unchanged.
- Market summary: 1-2 sentence macro blurb on SignalReport.news_summary,
rendered in Telegram alerts (📰) and the CLI (🧠).
New llm_client.py: requests-only OpenAI-compatible client (no SDK), one
attempt per call, UTC-day request cap, singleton so the cap survives
per-scan NewsEngine reconstruction. Config: deepseek block in
config.json + DEEPSEEK_API_KEY env; absent key = fully inert.
Live-measured API behavior baked in: V4-Flash is a hybrid thinking model
— requests send thinking:disabled (reasoning otherwise blew the timeout
and could exhaust max_tokens before the answer); prompts are static
module constants so DeepSeek's prefix cache prices repeat calls at the
cache-hit rate; scores use compact [s, xau] pairs; timeout 12s.
Tests 273 → 286 (new Module 19 + deepseek config tests).
Claude-Session: https://claude.ai/code/session_01Nsd8HcrV67GbCZXxZEn1Ma
Five-agent audit of the signal path; all fixes are delivery/measurement
side. Regression-proven behavior-neutral on validated signals: the 5yr
M15 replay reproduces the pinned book exactly (+210.65R PF 1.35
WR 57.8% MaxDD 33.90R). Tests 215 -> 221.
- TradeSetup carries trail_arm_r/trail_dist_r stamped from the producing
config; Telegram trail plan reads them (was showing default 0.6/0.75
instead of the validated M15 arm 0.70 / M5 dist 0.85)
- Alert dedup: fingerprint = (signal, quality, direction); entry movement
compared against last-ALERTED entry/risk (old SL-dist bucket churned on
ATR drift -> near-duplicate alerts every candle)
- Fresh fetches enforce the closed-bar boundary check with one 3s grace
retry, then warn (feed lag no longer silently analyzes the prior bar)
- Backtest max DD includes the zero baseline (opening losing streaks
were undercounted; pinned M15 DD unchanged)
- Replay loads the regime-filter TF when the H4-ADX gate is on (was a
silent no-op for M5/M15 primaries)
- clear_cache under _cache_lock; dead ROC zero-cross branches removed
from both engines (provably unreachable); RSS dates via
parsedate_to_datetime (named zones no longer stamped UTC); event risk
grades from the soonest event; .env inline comments stripped;
auto-scan skips NEUTRAL like the breach path; report_to_dict honors
the weight-0 skip rule and emits strategy/trail params; PB funnel
counts spread_gate/accepted; session-filter fail-open paths log loudly
- Docs corrected to match validated code (H&S neckline = lower peak,
exhaustion anchors farthest tracked swing) with re-validation warnings
Claude-Session: https://claude.ai/code/session_01R5soRUeRBi8kHMB99csrJy
- Delete test_ws_dual.py (standalone probe, unused) and a stale tracked log
- config.py: drop MarketSession enum, is_jpy_pair, candle_buffer_seconds,
and the never-read legacy SignalConfig/XauStrategyConfig fields
- data_provider.py: drop uncalled get_latest_price/get_spread_estimate
- Remove unused imports (engine, telegram_bot, main, news_engine)
- test_all.py: drop/trim the corresponding defaults assertions (199 tests)
- CLAUDE.md updated to match
Zero behavior change: no production code read any removed field
(verified reference-by-reference); suite green at 199/199.
Claude-Session: https://claude.ai/code/session_01VPNF37MP9Kuv62B1VBrZ1b
- telegram_bot: SCAN_PAIRS/_WS_TO_PAIR/SCAN_BUTTONS all derive from
PlatformConfig.scan_pairs (the single pair on/off switch); buttons and
keyboards are generated per pair; _run_signal_sync builds per-pair
configs via pair_signals factories (was: XAU defaults for any symbol);
scan-all works per pair; PB badge matches any *_PB tag; daily-rotating
file logs in logs/telegram.log (30-day retention)
- main.py: --pair argument (XAUUSD | EURUSD + aliases) routed through
pair_signals factories; price display is precision-aware (2dp gold,
5dp FX)
scan_pairs default stays [XAUUSD] until the EUR M5 re-tune validates.
Improvements from a live-verified audit of the news/calendar pipeline:
- Currency tagging: keyword matching now uses letter-boundary regexes
and drops ambiguous "us "/"eu "/"uk " tokens — "versus" no longer
tags headlines as USD (confirmed live false positive polluting the
XAU/USD sentiment average).
- Fetch caching: calendar events and headlines are cached class-level
(10 min TTL, 60s retry after empty/failed fetch, lock-guarded for
concurrent scans). Stops per-scan refetching that was tripping the
Forex Factory rate limit; TTL stays under the 30-min pre-event
buffer so the danger window can't go stale.
- Surprise detection: EconomicEvent.surprise is computed from parsed
actual vs forecast (handles %, K/M/B/T suffixes, negatives). A
surprise print widens the post-release caution window to 45 min;
inline releases keep the original 15 min.
- Gold-aware sentiment: headlines are scored per currency, with
monetary-policy keywords (hawkish/rate hike/tightening vs dovish/
rate cut/easing/stimulus) inverted for XAU — "hawkish Fed" headlines
scored +1.0 for gold before, -1.0 now (verified on live feed).
- New sentiment sources: Google News gold-focused query and CNBC
Economy RSS (verified live; 40 → 74 unique headlines). Added
cross-feed title dedup so syndicated stories count once.
- events_today now buckets by UTC date (feed timestamps carry US-
Eastern offsets); --json output includes news_description.
- Removed dead NewsConfig.min_impact and unused HIGH_IMPACT_EVENTS.
Tests: 142 passed (7 new covering boundary matching, surprise
classification, post-event windows, cache reuse, gold inversion).
Remove backtest.py, backtest_gpu.py, metrics.py, local CSV data,
Dukascopy loader, and all backtest CLI modes from main.py. Rebalance
engine weights: leading indicators (divergence, RSI, stochastic, BB)
now carry 66% of signal weight vs 34% for lagging (EMA, MACD). Add
ws_twelvedata.py for live tick streaming. Drop pyarrow and tqdm deps.
Session filter restricts trades to a UTC hour window (default 07-16,
London + NY) — opt-in via --session-filter, with --session-start /
--session-end overrides. Wired into both the backtester (cheap hour
check next to the regime gate) and live mode (downgrades signal to
NEUTRAL/F when current UTC hour is outside the window).
Auto-toggle (--auto-session-filter) tracks rolling mean pnl_pct over
the last N closed trades; when it falls below threshold, the session
filter activates, releasing once expectancy recovers. Default lookback
10 trades, threshold -0.3%.
Trade dataclass extended with diagnostic fields captured at entry:
entry_hour, entry_dow, ADX, ATR%, market_regime, volatility_regime,
mtf_agreement. These let --journal-csv exports surface loss-pattern
clusters when a window underperforms.
Validated on 2024 trend / 2023 chop / 2026 recent windows. Auto-toggle
preserves ~50% of trend edge while still cutting ~25% of recent-regime
losses — pure regime filter is still best on cumulative, so both
session flags ship default-off.
Restructure print_signal_report so decision (signal · grade · score ·
confidence · MTF) sits at the top, followed by trade setup, context line,
MTF breakdown, top supporting/opposing reasons in two columns, warnings,
and an optional detail block under --detail. Adds _compact_reason helper
that strips known indicator name prefixes via _INDICATOR_PREFIXES so
reason lines stay short.
Framework:
- TechnicalConfig.disabled_indicators skips named indicators in engine.analyze
- Optimizer supports disable_indicator override for parallel ablation sweeps
- main.py --ablate runs leave-one-out across all 10 directional indicators
Cleanup applied (validated across 2024 trend / 2023 chop / 2026 recent):
- Drop pivot by default (cumulative +9.5% to disable)
- Drop candlestick by default (+3.4% cumulative)
- Regime-gate ichimoku: 1.3x in trending, 0.2x in ranging (was a chop villain)
- Trim RSI weight 1.0 -> 0.3 (cumulative noise across regimes)
Result: +10.60% on 2024 trend (+33.24% -> +43.84%, Sharpe 2.17 -> 2.77),
+15.71% on 2023 chop (-18.27% -> -2.56%). Recent regime still adverse.
Harwinder Sekhon
harry