DIME
Get the picksJoin

PROJECT · findings note

project_wnba_snapshot_players_regression

ROOT-CAUSED 2026-07-07 — never a regression. Bare /api/v1/snapshot/{id} serves a warmer-cached players:null WNBA snapshot; league route is healthy (37 players). Real victim: grading fanout.go:390 fetches the bare route → disciplined family = 0 WNBA picks all season.

← All findingssource · memory/project_wnba_snapshot_players_regression.md

ROOT-CAUSED by the 2026-07-07 full-chain audit (project_wnba_audit_2026_07). The 2026-06-11 "players block regression" was route-scoped, not a regression:

  • /api/v1/wnba/snapshot/{id} (league route) is HEALTHY — verified live 2026-07-07 on both games: 37-38 players, 26-27 with prop_diagnostic, 1,055 mispriced line_edges, team stats block, injuries hard-gate.
  • Bare /api/v1/snapshot/{id} serves a thin players:null WNBA snapshot because data-hydrator's warmer writes ALL leagues under the bare NBA-namespace cache key (warmer.go:279,300) with an incomplete assembly, while the league route uses snapshot:{league}:{id} (handlers.go:783).
  • The consumer that matters — grading-service's 18:00 UTC fanout — fetches the BARE route (fanout.go:390), so system_lineedges/both_agree/disciplined/disciplined_board have zero WNBA rows all season. That consequence part of the original note was right.

How to apply: fix = league-prefix the URL in grading fetchSnapshot (fanout.go:390) + fix the data-hydrator warmer cache key + make looksLikePlayerMap (system_lineedges.go:218-233) scan all values. Then trigger fanout and confirm WNBA rows appear. Tier-0 item #1 in the audit roadmap. Related: project_wnba_snapshot_depth_gap (per-player depth blocks still unwired — separate, real, deprioritized).