Methodology

Ratings — DartsOrakel. Every ranked PDC thrower (4,000+ players) with a career 3-dart average (all-time) plus three trailing form windows — 30, 90 and 365 days — each with its own ranking, so you can see who is heating up or cooling off. The 3-dart average is the standard measure of darts scoring skill. Elo, checkout% and 180s counts sit behind DartsOrakel's paywall, so we use the free 3-dart average — the single most predictive number in the sport.

Results — Flashscore. The PDC schedule plus finished results (legs won/lost) across the tours. This feed is results-level only: it does not carry per-match 3-dart averages, 180s, or checkout percentages (that data is simply not published by any free darts source), and it retains only about two weeks of results at a time. We accumulate every finished match we see into a growing archive, so a player's tracked match history and leg record deepen with every refresh. Recent-form on player pages is therefore leg-level; any 180s figure is a season-rate estimate, clearly labelled.

Honest coverage. Darts has no decades-deep granular stat history to pull — the professional PDC era is modern (1990s onward) and per-throw detail was never digitised for free. We surface everything the sources cleanly expose: full-universe averages across four time windows, and accumulating results/legs history. We do not fabricate stats the feeds don't carry.

Props — PrizePicks. Darts props (180s, legs won, checkout %, match/set legs) are event-driven: PrizePicks only posts a darts board during televised majors (World Matchplay, the Worlds, Premier League nights). Between events the board carries model win-probabilities for the current slate and no props — that is expected, not a bug.

Win probability. The difference in blended 3-dart average between two players is passed through a logistic function: P(A beats B) = 1 / (1 + e^(−k·(avgA − avgB))), with k ≈ 0.10 per point (backtested on our results archive). This is the standard, well-calibrated darts win predictor.

Form blend. Each player's average is 0.60 × (30-day form) + 0.40 × (career) when both exist, so recent play is weighted without overreacting to one hot week. Probabilities are clamped to [5%, 95%].

Distributions. 180s are modelled as a Poisson process whose per-leg rate scales with the 3-dart average; leg-count props use that same rate. These are informational projections — with no per-match stat history available, they are not a betting edge.

Betting markets are efficient and most models do not beat them. We publish calibration, not a track record: the accuracy page shows whether our stated probabilities match reality, with no profit or ROI claims. When PrizePicks posts a darts line and a no-vig market price is available, we would show edge = model probability − de-vigged market probability; until then, everything is projection/information only. We will never surface an unearned edge claim.