500 trials with a ~3–5% expected win rate had enough variance to
occasionally land below the 0.03 threshold (~2% failure rate).
Bumping to 2000 trials reduces the std dev by 2x; lowering the
threshold to 0.02 keeps the assertion meaningful (still well above
random 0.78%) while eliminating the flakiness.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Implements a Monte Carlo simulator for men's/women's tennis seasons scored
on the qualifying_points pattern. Simulates all 4 Grand Slam majors
(Australian Open, French Open, Wimbledon, US Open) using surface-specific
Elo ratings and ATP/WTA world rankings for seeding.
New table: participant_surface_elos — one row per (participant, season)
storing worldRanking, eloHard, eloClay, eloGrass.
Key design decisions:
- Seeding uses ATP/WTA world ranking (not Elo), matching real draw procedure
- Top 32 seeded with standard slot placement (1→0, 2→64, 3-4→quarters, etc.)
- QP per round with tie-splitting pre-applied: W=20, F=14, SF=9, QF=4, R16=1.5
- Completed majors read actual qualifyingPointsAwarded from eventResults
- 10,000 Monte Carlo simulations; column sums naturally 1.0 (no normalization)
Admin UI at /admin/sports-seasons/:id/surface-elo:
- 5-column grid (Player | Rank | Hard | Clay | Grass)
- Bulk import: "Name, ranking, hardElo, clayElo, grassElo" one per line
- Fuzzy name matching (bigram Dice coefficient) with "Did you mean?" suggestions
- Inline participant creation for unmatched names via useFetcher
- Saves Elos and auto-runs simulation on submit
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
