How MatchWiz golf works

A model grounded in strokes gained

Everything on MatchWiz golf starts from strokes gained — the stat that measures how a player performs relative to the field on every shot, not just whether they posted a good number. Scoring average is affected by the course, the conditions, and the competition; strokes gained adjusts for all of it and puts out one comparable skill number per player per round.

Those numbers come from DataGolf (Scratch Plus), which reconstructs strokes-gained skill ratings from historical round-level data going back years. A player averaging +2.0 SG per round is expected to out-shoot the field by two strokes in any given round, regardless of the course. That's the signal the model is built on — not recent finishes alone, not name recognition, but consistent ball-striking and short-game skill measured shot by shot across a real sample.

From SG to win probability

The raw skill number tells you who's good. The model takes it further: given this specific field, how likely is each player to win, make the cut, or finish in the top 5, 10, or 20? The answer comes from a simulation that accounts for the variance in SG performances — skilled players have higher ceilings but also have bad weeks, and the model reflects that spread. The result is a probability for every relevant finish bucket, computed fresh as the field firms up before each event.

Once the tournament starts, those pre-event projections update round by round as scores come in. The live model blends the current leaderboard position with a player's underlying SG skill — so a leader who's running hot above their usual level gets a slight haircut, while a skilled player sitting two back picks up more win probability than the raw gap suggests.

Projections meet the market

A probability is only useful next to a price. For every player DataGolf projects, we pull the best available book odds, convert them to implied probability, and measure the gap. Where the model's number and the book's number diverge by at least 1.5 probability points, that spot shows up on the Value tab — the implied edge and the best available line, posted fresh each day.

The 1.5-point threshold isn't arbitrary. At longer prices, a 1-point gap can be rounding noise in the book's line; at a meaningful market — top-10 finish, realistic odds — a 1.5-point edge is real. Anything below that stays off the board. The vig alone can eat a 1-point discrepancy entirely.

Where the edge is — and where it isn't

This is the honest part most golf sites skip: most markets are efficiently priced, and we won't pretend otherwise. We check the outright winner market and make-cut lines in every snapshot. Both almost always come back clean — the books calibrate those markets closely, so real gaps are rare once you account for the vig. You'll see them listed on the Value tab as efficient when there's nothing qualifying that week.

Derivative markets are softer. Books invest less pricing effort in top-5, top-10, and top-20 finish lines than in the winner market, and that's where DataGolf's model finds real discrepancies. Those are the plays we surface. If a week has no qualifying gaps — and plenty of weeks don't — the board says so. We'd rather post nothing than post a manufactured edge.

Group picks — round by round

Each round we post a pick for every 2-ball and 3-ball group: the player we project to post the lowest score in their pairing for that round. The pick is made before the group tees off and never revised. After the round it's graded — win, loss, or dead heat (two or three players tying the low round split the credit at a proportional rate). The running record, correct-pick rate, and units at the posted odds all live on the model report card.

The DFS optimizer

The DFS tab runs a lineup optimizer for the DraftKings Classic main slate. Projected DraftKings fantasy points come from DataGolf — the same engine as the win probabilities — so the DFS projections and the model projections are reading from the same source. You get salary, projected points, projected ownership, and wave exposure (AM vs PM first-round tee time) for every priced player.

Wave matters because round 2 and round 4 flip the tee assignment — morning starters in round 1 play afternoon in round 2, and vice versa. On days when the wind comes up in the afternoon or the greens firm up, that split can be worth a stroke or more. The optimizer lets you tilt toward one wave or spread across both, so you're building lineups with a read on the conditions, not just stacking the highest projections and hoping the AM half gets the best of it.

We grade ourselves, in public

Every pre-event projection — win probability, top-finish odds, projected rank, make-cut call — is frozen before the first tee shot. Those numbers don't move once the tournament starts. After the final round, we grade: how many of our projected top-10 players actually finished top 10? Did the make-cut calls land? Where did the winner rank in our pre-event model? The answers are graded automatically and posted to the model report card, event by event, with no cherry-picking and no quiet revisions.

Season track record

Across 164 graded events, when MatchWiz projected a player inside the top 10 before an event, 28% of those picks actually finished top 10. Make-cut accuracy — where we modeled a player as more likely to make the cut than miss it — has landed at 72% across the same sample. Every projection is frozen before the first tee shot: the model's pre-event rankings are a public snapshot, graded after the final round against the real leaderboard. Pull up any past event on the report card and read exactly what we said and where the players actually finished. The record is the product — the same ledger that tells us which inputs are pulling their weight and which ones to re-examine.

See it on the board

The model runs every week there's a PGA Tour event. Check the Golf board for the current week's leaderboard, win odds, and group picks. The Value tab shows any market gaps qualifying that week — if the board is empty, the book lines are efficient and we're not forcing one. If you play DFS, the DFS tab has the current slate's projections and the lineup optimizer. To see the full graded history — every past event, how our calls landed — it's all on the model report card.