Evidence-quality warning
Legacy letter grades are directional/forecast context only, not trade win rates or profit. Legacy watches were graded at daily chart.close, not fills or confirmed triggers; negative excursions also make those saved grades unreliable as execution performance. Do not use them to promote or suppress agent rules. Letter-grade improvement is not proven. Forward evidence collection is paper-only with matched 1.5R/2R/3R arms; continuous archived live minute/tick data is not yet available. Unavailable data stays ungraded; unknown costs keep net expectancy null. Entry, risk and trading rules are unchanged.
Forward evidence collection
Paper-trial status: summary_available. Forward trial cumulative; not a daily/weekly sample. Primary path: missing_usable_continuous_primary_bars.
Verified saved-summary counts: {"ambiguous": 0, "cancelled": 0, "confirmations_recorded": 0, "confirmed": 0, "expired": 0, "matched_resolved": 0, "open": 0, "report_only": 38, "reports": 38, "ungradable": 0, "unsupported_watches": 0, "untriggered": 0, "unverified": 0, "watches": 0}.
Net expectancy: unavailable — costs and execution evidence not verified here. No trade win rate inferred.
📊 Weekly NQ Evidence Review
Legacy directional/forecast context for Oct 19–Oct 23, 2026 ET. Score reflects old 15m/30m/60m context outcomes only; raw GEX/options-positioning details remain internal.
Journal coverage: 963 reports seen lifetime; 39 unique weekly NQ reports evaluated; 117 legacy horizon outcomes; 0 pending windows excluded.
Horizon Performance
| Horizon | Graded | Directional Accuracy | Target Hit | Invalidation Hit | Avg MFE | Avg MAE | Assessment |
|---|---|---|---|---|---|---|---|
| 15m | 39 | 48.7% | 12.8% | 2.6% | 54.8 pts | 32.1 pts | Legacy forecast context only |
| 30m | 39 | 38.5% | 17.9% | 2.6% | 69.5 pts | 48.9 pts | Legacy forecast context only |
| 60m | 39 | 43.6% | 25.6% | 5.1% | 94.7 pts | 69.5 pts | Legacy forecast context only |
Key-Level Modules
| Module | Outcomes | Directional | Avg Score | Avg MFE | Avg MAE |
|---|---|---|---|---|---|
| ten_am_open_levels | 103 | 45.6% | 72.2 | 64.6 pts | 12.6 pts |
| primary_htf_key_levels | 210 | 45.2% | 71.8 | 61.7 pts | 50.3 pts |
| session_key_levels | 97 | 49.5% | 69.8 | 36.9 pts | 59.0 pts |
Setup Types
| Type | Outcomes | Target Hit | Avg Score |
|---|---|---|---|
| Bear/Neutral / balanced | 117 | 18.8% | 73.3 |
No-Trade Condition Quality
No-trade quality is unverified; legacy forecast grades cannot assess confirmed entry or stand-down decisions.
Missing: formal no-trade labels, confidence score, stop distance, and R-multiple fields.
Data Completeness
Available: grade, trend proxy, stretched flag, support/resistance hits, MFE/MAE, horizon moves, and forward key-level setup tags. Missing: execution fills, slippage, and true tick/100t bars. Pending windows this week: 0.
What Worked
- Historical forecast records remain available for audit, not execution-performance claims.
- Best available setup bucket: Bear/Neutral / balanced with 18.8% target-hit rate.
- Longer horizon rows highlight whether calls matured beyond first-noise reactions.
What Failed
- Formal R-multiple, stop-distance, and execution/slippage fields are still missing from the journal.
- Worst available setup bucket: Bear/Neutral / balanced with 18.8% target-hit rate.
- Repeated pending rows in the raw grade log require deduplication during dashboard aggregation.
Adjustments for Next Week
- Log explicit setup type, confidence, stop distance, target distance, and R multiple for every report.
- Collect prospective matched paper-trial evidence; do not promote or suppress agents from legacy forecast grades.
- Keep raw GEX/options details internal, but add a simple confluence tag for post-trade scoring.
- Weight 15m/30m context evidence highest; treat 60m as context rather than the default hold window.