Grade-fix validation script + finding doc (before-state, blast-radius SQL, plan)

scripts/validate-grade-fix.js checks the live MLB snapshot for the three
degradation signatures (projection=0, edge=100 cluster, grade/conf disagreement)
— run after the next 14:00 UTC regrade to fingerprint the fix. The finding doc
now records root causes, fixes (commits 888d103/9fc4edf), the blast-radius SQL
(box can't reach Supabase directly), and the shared-path note for NBA/WNBA.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Kev
2026-07-17 03:18:36 -04:00
parent 9fc4edf3a9
commit 4bd770480a
2 changed files with 88 additions and 31 deletions
+60 -31
View File
@@ -1,37 +1,66 @@
# MLB Grade Pipeline Degradation — Backend Finding (2026-07-17)
# MLB Grade Pipeline Degradation — FIXED (2026-07-17)
Source: phone-audit P1-7 (broken edge board) + P2-9 (B grades at 45% confidence).
Diagnosed against LIVE `GET /api/snapshot/mlb` on 2026-07-17. This is a BACKEND
grading/feature-pipeline issue, NOT a frontend display bug. The frontend guards
(sane-edge cap, statAbbrev) are damage control, not the fix.
Diagnosed against LIVE `GET /api/snapshot/mlb` on 2026-07-17. Backend grading
bug, fixed at the source in the generic grade path (`engine1` +
`analyzeViaEngine1`), which grades EVERY sport.
## The evidence (25 live MLB grades)
- **projection == 0 for 9/25.** The model has no projection reference for a third
of props — a degraded feature path (the S58 insufficient-data refusal should
have caught these; instead they graded with projection 0).
- **edge_pct on a broken scale.** Distinct values {20, 60, 100, 140}. A real
prop-market edge is single-digit %, never past ~40. These are not a market edge
(frontend now shows them absent — slateAdapter EDGE_BOARD_SANE_MAX / EdgeCell).
- **grade <-> confidence mismatch on 25/25.** Every grade's letter disagrees with
its own surfaced `confidence` vs `grade_thresholds.json`:
- B shown at confidence 55 -> 55 is the B- band (55-59), not B (60-65).
- Systematic ~5-point (one sub-tier) gap on every prop.
The surfaced `confidence` is NOT the probability that derived the letter.
## Before (25 live MLB grades, degraded)
- **projection == 0 for 9/25** — graded on a zero projection.
- **edge_pct quantized to {20, 60, 100, 140}** — the 100s were the proj=0
degeneracy `(line - 0)/line = 100%`.
- **grade ↔ confidence mismatch** — 10/25 disagreed even at the 4-letter level
(25/25 vs the stricter 11-step bands).
## Likely root cause (to investigate, not yet fixed)
1. `confidence` field is post-`apply_data_sufficiency_modifier` (bayesian.py),
while the grade letter was assigned from the PRE-penalty prob -> the two
disagree by one sub-tier on every prop. Either surface the same value, or
re-letter from the penalized confidence. Decide which is the source of truth.
2. projection==0 for 9/25: the MLB feature/projection path is feeding 0 instead
of refusing (S58 `insufficient_data`). Trace `projectionFor` for MLB props;
projection 0 also breaks `computeEdge` (=> the 20/60/100/140 garbage).
## Root causes + fixes (commit `888d103`)
1. **projection=0 bypassed the refusal gate.** `projectionFor` returned
`l5_avg` even when 0 (finite → the `== null` gate passed it). FIX: a
non-positive reference is not a projection — `projectionFor` skips it and
falls through to the next POSITIVE reference (l5 → l20 → per_90 → xg); when
none is positive it returns null and the read REFUSES (`insufficient_data`).
The gate also gained an explicit `> 0` guard — the invariant is structural.
2. **edge_pct.** Formula was already `(model - line)/line` (the intended
semantics); the {100} cluster was purely the proj=0 degeneracy. With fix 1
those refuse. Main-line edge now reuses the VALIDATED projection so edge and
the persisted `projection` can't diverge.
3. **confidence/letter split.** `engine1.GRADE_TO_CONFIDENCE` was hand-rolled
and drifted a full sub-tier low (B → 0.55, which `grade_thresholds.json`
calls B-). FIX: confidence is now DERIVED from each grade's band MIDPOINT in
`grade_thresholds.json` — one source of truth. Applying the threshold table
to any grade's displayed confidence resolves back to the same letter (proven
for all 11 grades in `tests/unit/mlbGradeDegradation.test.js`).
## Do NOT
- Do not re-letter or shift thresholds on the frontend (would hide the bug).
- Do not "fix" edge display by rescaling 140 -> 14 (guessing the scale is its own
fabrication). The number must come out of the pipeline correct.
## Blast radius (commit `9fc4edf`) — work-order #6
The degraded grades (projection=0 → `model_value = 0`) are already settled in
the append-only `ledger_entries` and are NOT deleted. Functional marking:
`getModelAggregate` now filters `.gt('model_value', 0)` on the settled AND
pending queries — the rows stay in the ledger but leave the public model record
(their hit/miss is noise, not skill). Post-fix no new such row can be written.
## Frontend already shipped (honest guards)
- Leaderboard stat labels: `web/src/lib/statAbbrev.js` (raw snake_case -> SB/ER/TB).
- Edge board: impossible |edge|>40 rendered absent AND excluded from ranking.
**Exact count NOT queryable from the dev box** (`*.supabase.co` is unreachable
here — curl 000; only `vyndr.app`/`api.vyndr.app` resolve; no `VYNDR_INTERNAL_KEY`
locally). Proxy signal: 9/25 (36%) of the current live slate. For the precise
figure, run in Supabase SQL:
```sql
SELECT count(*) FILTER (WHERE outcome IS NOT NULL) AS settled_degraded,
count(*) AS total_degraded
FROM ledger_entries
WHERE user_id IS NULL AND model_value = 0;
```
## Other sports — work-order #5
NBA/WNBA/soccer grade through the SAME `analyzeViaEngine1``engine1` path
(`gradeSlateService` does not branch by sport; `mlbGrader.js` is dead code). So
they SHARE the disease and are fixed by the same commit. They rarely grade in
prod today (stats service offline off-season → refuse anyway). No separate fix.
## Live validation — work-order #4
The fix deploys immediately, but the SNAPSHOT only re-grades on the full cron
(UTC hours 14,19,22,1,3). Run after the next 14:00 UTC snapshot post-deploy:
```
node scripts/validate-grade-fix.js
```
PASS criteria: projection==0 count → 0; edge_pct no longer contains 100 and is
not the four-value cluster; grade↔confidence agreement 25/25. The before-state
(this file's "Before") is the diff baseline; the script's output is the
fingerprint.