Benfica·ForwardWhat the number means: if Franjo Ivanović plays 90 minutes at a neutral venue against an average Liga defence, the model expects 0.21 more goals than if that place were taken by a replacement-level player — the kind of signing any club can make for free. Goals are capped (winsorized) before the estimate, so one four-goal afternoon is not read as permanent skill.
The model puts 90% probability on the true value lying between 0.06 and 0.40. Fewer minutes, wider interval.
Goals go up and down; the skill estimate barely moves. That is by design: the model only admits a change in skill when the data demand one, and across three Liga Portugal seasons they never have.
| Season | Minutes | Goals | Skill (with margin) |
|---|---|---|---|
| 2025-26 | 885 | 6 | 0.22 |
There is only one season of this player in the model, so there is no change to measure. Even with more seasons, the measured difference is usually smaller than its own margin of error.
Match-by-match data (SofaScore). The rating is the provider's, not the model's — the model only uses goals, minutes and opponent.
Bayesian player model fitted on 29,019 individual appearances since 2023-24, with a 600-minute minimum to qualify. Club = the player's most recent club in the data (2023-24 to 2026-27); recent transfers may not be reflected.