Darts averages miss the checkout
The three dart average is informative but incomplete, because winning a leg also depends on checkout opportunities and finishing performance.
On this page
Every darts broadcast quotes the three dart average, and it is genuinely informative. It is also the reason a lot of darts modelling is worse than it looks when the model uses that average alone, because the topline number is weighted heavily toward scoring and does not isolate checkout performance.
Scoring and finishing are different skills
An average is dominated by scoring: how often a player hits treble twenty and how consistently. Finishing is a separate ability, exercised from a different part of the board, under different pressure, with a different failure mode.
Two players with the same three dart average need not have the same chance of winning. Checkout data contains information the average does not isolate, and over a best of eleven that difference can compound in the same way a tennis serve edge compounds through the scoring system.
Checkout percentage also needs context. Research using more than 32,000 PDC checkout attempts found that success depended strongly on the number of points required, so raw percentages do not describe identical chances.
The variance is higher than the numbers suggest
A player can post the higher average and lose because legs, not aggregate points, decide the match. The PDC's own statistical analysis notes that the topline average is heavily biased toward one aspect of play and discusses alternatives that incorporate checkout performance.
Why a handicap needs more
A leg handicap cannot be recovered from the match price alone. Different leg-level models can produce the same win probability but different chances of covering a spread. The handicap therefore needs its own distribution rather than a fixed conversion from the headline price.
What this does not tell you
Context is not fully captured by either average. Research using PDC matches found performance differences between real-crowd and no-crowd conditions, but those settings changed several things at once. Treat a claimed stage effect as an estimate with confounding, not a permanent player trait.
Common questions
Is a three dart average a good predictor in darts?
It is the best single number available and it is incomplete. Scoring and checking out are related but distinct skills, and averages weight the first heavily.
Why do checkout percentages matter so much?
Because a leg is not won by scoring, it is won by finishing. Two players with identical averages and different checkout rates are not equally likely to win.
What must a darts handicap model include?
It needs a leg-level distribution that reflects scoring, checkout opportunities, finishing performance and who throws first. A match price alone does not identify that distribution.