- Remove unused _gate_eligible_levels filtering logic and its tests (research-only)
- Add prominent RESEARCH/DIAGNOSTIC markers and docs to clear-air/ATR fallback helpers
- Document production vs research BACKTEST_* environment variables in backtest_service
- Minor cleanups: update legacy report text, improve outdated function docstring
Three follow-ups to the gate probability floor (8f41143):
- Signals table shows the starred primary target (shared primaryTarget
helper) instead of an independently computed max-probability best,
so Overview, Signals and ticker details agree by construction.
- Targets pinned at the 3% probability clamp floor collapse to the
nearest one (enhance_trade_setup + backtest candidates in parity):
floor-pinned levels are indistinguishable to the model, so farther
ones were duplicate 3% rows inviting lottery headlines.
- get_trade_setups only returns setups re-emitted within
LIVE_SETUP_MAX_AGE_DAYS (3): an older latest row means the daily
scan no longer confirms the setup, and such rows otherwise surface
forever on Overview/Signals/ticker/alerts. History endpoints keep
full history.
Backtest on the Jul-3 snapshot is metric-identical to the gate-floor
run on all qualified stats (1089 qualified, Sharpe 2.02, CAGR +49.6%,
DD -15.8%): the prune only removes noise the gate already rejected.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A setup's primary target could carry a ~3% probability: the picker
chose the most likely target among those with R:R >= 1.5, and after a
run-up that pool can contain only far "lottery" levels (the near,
likely levels fail the R:R floor). The lottery target's inflated R:R
then became the setup's headline and passed the activation gate's
min_rr floor - the gate's probability check only requires a value to
exist.
Fix, no new tuning knobs: the primary must clear BOTH floors
(R:R >= 1.5 AND probability >= 20%). When nothing does, fall back to
the most likely target overall, so the headline carries an honest low
R:R and the gate rejects the setup on real numbers instead of being
gamed by an unreachable target.
Deliberately NOT pure EV-maximization (p*RR): the probability model
adds strength/alignment bonuses as flat percentage points, so EV
arithmetic would scale those bonuses by (RR+1) and systematically
favor far targets on aligned setups - the same lottery bias through
the back door.
Shared by production (enhance_trade_setup) and the backtest simulator,
so backtest comparisons stay apples-to-apples. Unit tests pin the
degenerate case; full unit suite green (497 passed).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Backtest (32k setups) showed the touch-only probability model was ~2x
over-confident — predicted 70% hit 39%, predicted 88% hit 46% — because it
ignored the competing stop. estimate_probability now multiplies the reach
probability (touch within horizon) by the two-barrier gambler's-ruin ratio
1/(R:R+1) = P(target before stop). A 3:1 setup now reads ~25% base, not ~70%,
which lines up with realized rates. Strength/alignment modulation unchanged.
Recalibrates every probability and the EV ranking; the min_target_probability
gate threshold now means roughly what it says. Re-run the backtest to confirm
the calibration table flattens toward the diagonal.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Trade-setup targets now pre-merge near-duplicate S/R levels into zone
representatives (same 2% clusterer as chart + alerts) before generate_targets
runs. A clustered wall (e.g. 183 + 185) becomes one target carrying the zone's
COMBINED strength (capped 100) instead of two near-identical targets that each
undervalue the wall — which also feeds a more honest reach-probability via the
S/R-strength magnet. Representative price is the zone's near edge; the strongest
constituent's id is retained. Singleton levels pass through unchanged, so the
downstream band-spreading / probability / primary-selection pipeline and its
tests are untouched.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
MKC showed 5 targets all far/Aggressive: target selection was top-5 by
quality (0.35*R:R + ...), and R:R grows with distance, so far levels
crowded out every nearby one.
generate_targets now selects for spread: always include the nearest
level, plus the best-quality representative from each distance band
(Conservative <=2.9 ATR, Moderate <=4.6 ATR, Aggressive beyond), then
fill remaining slots by quality. Restores a Conservative/Moderate/
Aggressive mix with the nearest target always present.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes nonsensical "Conservative @ 90%" on far targets (AJG: +39% target
labelled Conservative/90%).
- Probability = chance of touching the level within the outcome horizon
under a driftless random walk: 2*(1 - Phi(d / (ATR*sqrt(T)))), T=30d to
match the outcome evaluator. Distance (in ATR) dominates; strength and
alignment modulate modestly. Far targets are now correctly low-prob.
- Classification derived from that probability (>=60 Conservative,
40-60 Moderate, <40 Aggressive) instead of distance-rank.
- Combined with the most-likely-worthwhile primary pick, the nearest
strong resistance becomes the Conservative primary; far levels demote
to low-probability stretch targets.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- JWT now carries a username claim; sidebar shows "Signed in as <name>"
instead of the bare user id (sub). Re-login required for the new claim.
- Signals: Min R:R / Min Confidence inputs reflect the effective filter —
auto-filled from the activation gate when "Qualified only" is on, reset
to 0 when off (no more misleading 0 while the gate is active).
- Signals layout: Run Scanner moved to its own action row (it's a job
trigger, not a filter); qualified toggle grouped with the refinement
filters under one Filters panel.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Triggered by CNC showing "LONG (High Confidence)" with SHORT reasoning
and no long setup.
- A: recommendation action + reasoning are ticker-level and identical
on both setups; reasoning always matches the shown action
- B: recommended_action only picks a direction with a tradeable setup;
strong bias with no setup (e.g. price at ATH) → NEUTRAL with an
explanatory reason instead of a fake LONG_HIGH
- C: confidence is a directional-agreement model — opposing signals push
it below 50 (SHORT on a 92-technical/99-momentum stock ~0%, not 55%)
- D: fundamental score requires >=2 real metrics (market-cap-only no
longer yields a high score)
- E: RSI score peaks at healthy momentum (~60) and penalizes
overbought/oversold extremes instead of treating RSI 90 as maximal
- F: fundamentals chain merges fields across providers (FMP market cap
+ Finnhub P/E) instead of stopping at the first with any field
- NEUTRAL label: "No Clear Setup" (covers untradeable-bias case)
Scores recompute on next scan/scoring run; C and E shift score
distributions intentionally.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
dennisthiessen
