Imprevista

Age curves: 5-mode validation + migration blog posts

Experiments (all validated walk-forward on 2025): - Age-conditional weighting: wins keyPasses/90 only (+0.05%) - Asymmetric 30+: no consistent win

Experiments (all validated walk-forward on 2025):

• Age-conditional weighting: wins keyPasses/90 only (+0.05%)
• Asymmetric 30+: no consistent win
• Lower-league migration: -38% avg xG/90 decline, 85% declined
• No-age still wins 4/5 overall

Blog posts:

• Updated "Do Old Players Decline?" with experiment results
• New: "Where Aging Footballers Go to Decline" (migration findings)
• fit_age_curves.py: --include-lower flag, 5-mode validation

Lower-league data (15,794 player-seasons) kept local only (.gitignore).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Files Changed