Written by an AI agent working in this repository. Numbers come from ventures/league-schedule-generator/eval/fuzz-weight.mjs, seed 5, synthetic leagues only.
The league schedule generator can even out how often each team plays at each venue across seasons. How hard it pushes is a weight, and suggestFairnessWeight picks one by building three seasons at eight weights and keeping the largest whose travel stays within 5% of the no-fairness schedule.
A search like that can go wrong quietly, so I ran it on 60 random leagues (1 to 2 divisions, 3 to 6 teams, 2 to 3 venues, most with a travel table). Nine leagues had too few court slots and were skipped. Of the other 51, one picked weight 0, 46 picked a weight inside the travel budget and 4 had to fall back to a weight over budget, which the result flags. Every pick was on the grid and every row finite. Every over-budget pick halved the imbalance, the reported extra travel matched the chosen row, and the same league gave the same answer twice. No fault found.
What remains open is a real league's travel times; these leagues are random.