The model
OTIS ranks a tournament field by blending three things: baseline skill from strokes gained, fit between a player’s game and the course, and current form. Everything the model produces lives here, along with an honest account of how well it has done.
This week
- Course fitThis week's field ranked by how well each player's game matches this course — and the same ranking for previous weeks.
- Win probabilitiesCalibrated chances for this week's field, with the track record of how those probabilities have actually resolved.
- Form rankingsWho is playing well right now, from strokes-gained form rather than results alone.
Tools on the model
- Matchup pricerPrice any head-to-head, 3-ball or 4-ball in the field at the model's fair odds — this week's groupings, or players you pick.
- DFS value sheetProjected DraftKings points for the field, with ceiling and floor, an estimated salary, projected ownership and leverage for members — and your salary file priced against them.
DFS
The sheet prices the week in DraftKings points — expected points, ceiling and floor from the same simulation as the probabilities — and, for members, adds two models of what DraftKings and its crowd will do: an estimated salary for every player before the real ones post, and a projected ownership share, with leverage as the reading device between the two. Both are frozen weekly and graded against the real slate and a real contest as they arrive, the same way the points are.
Previous weeks
The ranking exactly as it stood before each event started. These are snapshots, not reconstructions — they cannot be quietly rewritten after the fact.
How it scored
Each finished event re-run as the model would have seen it beforehand, then compared with the real finishing order. Across 630 events the median rank correlation is +0.324 — where 0 is a coin flip. These are backtests, not calls we published in advance.
| Event | Field | Correlation | Grade |
|---|---|---|---|
| Baycurrent Classic2026 | 68 | +0.317 | C |
| Bank of Utah Championship2026 | 69 | +0.160 | F |
| Biltmore Championship Asheville2026 | 74 | +0.385 | C |
| TOUR Championship2026 | 29 | +0.284 | D |
| BMW Championship2026 | 50 | +0.187 | F |
| FedEx St. Jude Championship2026 | 68 | +0.329 | C |
| Wyndham Championship2026 | 80 | +0.340 | C |
| Rocket Classic2026 | 73 | +0.404 | B |
Where the model struggles
Venues whose events the model reads worst, across more than one edition. A course well below the median usually means something is wrong in its data — the wrong archetype, a wrong par, or a history split across renamed venues — rather than an unpredictable golf course.
- Congressional CC (Blue)+0.138(3 events)
- Black Desert Resort+0.175(3 events)
- PGA West (Palmer)+0.197(3 events)
- Royal Liverpool+0.204(2 events)
- Grand Reserve Golf Club+0.216(13 events)
- PGA National - Champion Course+0.235(14 events)
How the model is built · Glossary
Built and maintained by @otis_golf. Independent, not affiliated with the PGA Tour. Questions and corrections welcome.