Trey t
e195944297
feat(08-02): optimize GameDAGRouter performance for large datasets
Implemented dynamic beam width scaling and early termination to handle
5K-10K game datasets efficiently. All performance tests now pass.
Optimizations:
- Dynamic beam width: 800+ games use width 50, 2K+ use 30, 5K+ use 25
- Early termination: stop expanding when beam reaches 3x target size
- Prevents exponential blowup during day-by-day expansion
Performance results:
- 1K games: 2s (was 2s baseline) ✓
- 5K games: 1s (was 13s) - 13x faster ✓
- 10K games: 1-2s (was 34s) - 17x faster ✓
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>