What matters
Golf is a repeated-shot scoring game played across changing conditions. A player can gain or lose position without hitting a shot because other players are still on the course. That makes round state, holes remaining, course setup and field context essential.
For pre-tournament probability, the central question is not “who is best?” but how a golfer's distribution of possible rounds interacts with the field and the course. A 12% win probability still means the golfer loses roughly seven times out of eight. See Confidence vs certainty and Distributions.
Important data
| Input | Why it matters |
|---|---|
| Leaderboard state: position, score to par, thru | The minimum live state: where the golfer stands and how much golf remains |
| Round and hole-by-hole scorecards | Shows how the current total was built without inventing shot locations |
| Course setup: par, yardage, hole sequence | Changes the scoring environment and which skills matter |
| Field and tournament structure | Cut rules, field size and strength shape advancement and winner probability |
| Weather by tee-time window | Wind, gusts, temperature and precipitation can create asymmetric playing conditions |
| Historical tournament and major results | Adds context while keeping different years and course setups separate |
Common metrics
| Metric | What it measures | How to read it |
|---|---|---|
| Score to par | Strokes relative to course par | A live result, not a skill rating |
| Thru | Holes completed in the current round | Tells you how much opportunity remains |
| Cut line | The score threshold for advancing | Moves as the field finishes; it is not final until the round state supports it |
| Win probability | Estimated chance of winning the tournament | Read as a distribution over many possible finishes, never as certainty |
| Top-N probability | Chance of finishing inside a rank threshold | Usually less volatile than winner-only probability but still field-dependent |
| Market-implied probability | Contract price expressed as a probability | A benchmark from traders, not an input that should leak into an independent model |
Model concepts
Tournament forecasting is naturally a simulation problem. A model needs a distribution for each golfer's scoring, not one expected score, and then it must simulate the whole field through the remaining rounds. Correlation matters because golfers share weather windows, course conditions and cut mechanics.
Judge a probability model with Calibration, Brier score and log loss, and enough out-of-sample events. Winner markets are especially sparse: even an excellent 10% forecast is expected to lose nine times out of ten, so short records are weak evidence. See Sample size.
Live intelligence
Golf PBEcast is an observed-scoring cast. In live mode it follows positions, thru, scores and current-round hole cards from the approved live scoring lane, with a full-field tape and selected-golfer panel. The source does not provide shot coordinates, clubs, lies or shot paths, so PBEcast does not draw them.
That distinction matters: a birdie on the card is observed; an imagined approach-shot location is not. Live probability should update from the state that actually exists — holes remaining, leaderboard position and verified scoring — rather than from invented detail.
PropBetEdge Golf tools
PBEcast ↗
Observed live-scoring cast with leaderboard, scoring tape and selected-golfer context.
GolfLive Golf ↗
Live tournament scoring and leaderboard context.
GolfTournaments ↗
Tournament schedule, editions and results.
GolfPlayers ↗
Player profiles, form and historical context.
GolfCourses ↗
Course profiles and versioned setup context.
GolfMajors ↗
Major-championship history and intelligence.
GolfMatchups ↗
Player-vs-player matchup research.
GolfGolf Intelligence ↗
Advanced golf intelligence and model context.
GolfGolf News ↗
Evidence-backed golf reporting.
Worked example
- Illustrative only: an independent model gives a golfer a 12% chance to win a tournament.
- A YES event contract is offered at 9¢. Its price corresponds to a 9% implied probability before fees, spread and execution effects.
- Probability gap = 12% − 9% = +3 percentage points. If the 12% estimate were perfectly calibrated, expected value before trading costs would be about +3¢ per $1-settled contract.
Glossary
Related lessons
Confidence vs certainty
What does "High confidence" actually promise?
Distributions
A player is projected for 6.2 strikeouts and the line is 6.5. Is the over likely?
Sample size
How many results before a record means something?
Calibration
When a model says 70%, does it happen 70% of the time?
Brier & log loss
How do you grade a probability, not a pick?
Reading a model
I am looking at a model output. What should I check, in what order?
Game state
What exactly does a live model know that a pregame model does not?
Pregame vs live
The pregame model said 60%. Why does the live number say 35% in the second quarter?
Expected value
If my probability is right, what does this price return on average?