How good are these forecasts?
Every forecast scored on seasons the model never saw while its settings were chosen.
Against the betting markets
Each market's price before the start, scored on the same finished games as the model. A negative figure means the market was more accurate.
| Market | Since | Priced | Model log loss | Market log loss | Model vs market |
|---|---|---|---|---|---|
| Kalshi | 4 Apr 2025 | 17 | 3.153 | 3.312 | 4.8% |
This is strokes gained against the field from round scores, not shot-level Strokes Gained from ShotLink: it knows how many strokes a player took, not where they came from.
Events are 72 holes with no cut from 2026 (54 before), with a fixed field of 57; the team competition isn't simulated.
Rounds on the other men's tours (PGA Tour, DP World Tour, LIV) count toward each player's skill, since the best players split their time; only this tour's events are forecast and scored.
Withdrawals during an event count as missing the rounds they didn't play.
Course fit is not in the model. LIV's short history means the ratings also lean on the prior, and players' results elsewhere (the majors, the DP World Tour) aren't included.
Questions and answers
How accurate are PodiumOracle's LIV Golf predictions?
On 40 tournaments from held-back seasons (2024 to 2026), which the model never saw while it was tuned, its favourite won 10% of the time in fields of about 55, and it gave the eventual winner 6% on average. Its log loss was 3.305, against 4.005 for treating every player as equal and 3.544 for recent form. Against the betting markets on the same finished events, its log loss was 3.153 against Kalshi's 3.312 over 17 tournaments (lower is better).