tipper

AFL match prediction CLI combining MOV-Elo ratings with player-level PAV (Player Approximate Value). Runs as a Cloudflare Worker backed by a D1 database.

Setup

npm install -g @jackemcpherson/tipper

# Auth: uses wrangler's OAuth token automatically
wrangler login

If wrangler login isn't available, set CLOUDFLARE_API_TOKEN instead.

Usage

# Backtest the current model across historical seasons
tipper backtest

# Backtest a specific config
tipper backtest --config pavfix-blend-w06

# Predict upcoming matches
tipper predict --season 2026 --round-number 8

# Bootstrap-compare two configs
tipper compare --config-a pavfix-blend-w06 --config-b elo-only-tuned-v1b

# Manage configs
tipper config list
tipper config show pavfix-blend-w06
tipper config current
tipper config promote pavfix-blend-w06 --reason "v2: corrected PAV blend"
tipper config create new-config --from pavfix-blend-w06
tipper config diff config-a config-b

Development

bun run build        # Compile TypeScript
bun run typecheck    # Type-check without emitting
bun run test         # Run tests (vitest)
bun run check        # Lint + format (biome)
bun run format       # Auto-format

Architecture

Two parallel state machines joined by a read-only predictor:

• Elo — MOV-Elo rating system (FiveThirtyEight-style margin-of-victory multiplier). Tracks team strength over time.
• PAV — Round-by-round player approximate value using the HPN formula with a fixed pool of 100 points per team per zone per season. Captures player-level quality that Elo misses.
• Blend — Weighted combination: rating = 0.6 * elo + 0.4 * (slope * pav). The calibration slope (6.986) converts PAV into Elo-equivalent units.

The engine is pure functions with no I/O. All data access goes through the Cloudflare Worker, which reads from D1 and passes pre-fetched data to the engine.

Current model (v2)

Model:  pavfix-blend-w06
Type:   MOV-Elo + PAV (corrected defence formula)

Parameters:
  K-factor:            25
  Home advantage:      160 Elo points (11.2 scoreboard points)
  Regression to mean:  0.10
  MOV multiplier:      538_log
  Sigma:               36
  Blend weight (Elo):  0.6
  PAV cal. slope:      6.986

Performance (2021-2025, 1062 matches):
  Tip%:      66.1%
  LogLoss:   0.8607
  Brier:     0.2060

Out-of-sample (2026, 63 matches):
  Tip%:      77.8%
  LogLoss:   0.8029

See docs/ for the full tuning reports (Tasks 1-18).