Level Equivalence

We can't watch the same run scored at two levels at once, so the model is fitted on athletes who competed at both levels in the same season — their ability is fixed, so the gap between their scores is the level gap. Each link is an ordinary regression; sparse jumps (a COMP skier almost never enters a World Cup) are reached by chaining through the levels they do overlap (COMP → Nor-Am → World Cup).

Because a strong regional result only weakly predicts an elite one, the translation regresses toward the elite field and carries a wide ± band — it's an estimate of where you'd stand, not a promised score. FIS tiers differ too (a World Cup is ~17 points stricter than an entry-level FIS race), so you can pick the target. Validated against the athletes who really did make the jump (bias +4, typical error ±14 points).