How the rating works

USPSA classification measures how well a shooter runs a classifier, a fixed drill with a published high hit factor. It says nothing about match finishes. The rating here does the opposite. It ignores classifiers and uses only full matches, and it reads as a single number: the percent of the winner a shooter would be expected to post in a field of average strength. A rating of 100 means that shooter would be expected to win an average field. A 92 means they would be expected to finish at 92 percent of the winner.

The reason a raw average of match percentages is not enough is that fields differ. Shooting 90 percent at a Nationals where three Grand Masters are in the division is a far better result than shooting 90 percent at a club match where the winner is a C-class shooter. A raw average treats those two results as identical. The rating treats them as what they are.

Who is ranked

A shooter appears in a division's ranking when, in the last twelve months, they have shot at least four matches in that division where the division field had eight or more shooters who completed the match. Disqualifications are excluded. Shooters without a member number cannot be tracked across matches and are not rated. A division is ranked at all only when it has at least two hundred entries in the window.

Step one: percent of the winner

Each result starts as the shooter's match points divided by the division winner's match points at that match, times one hundred. This is the same number PractiScore shows and is the only input. Stage times, hit factors and penalties enter only through match points.

Step two: separating the shooter from the field

The model assumes every result is the sum of two things that cannot be observed directly: how good the shooter is, and how easy the field was to score against.

Field effect is positive for a weak field, where everyone's percentage is inflated because the winner was beatable, and negative for a strong one. Neither quantity is known in advance, so both are solved for together. The fit alternates: holding every shooter's ability fixed, each match's field effect is the average amount by which its shooters beat or fell short of their abilities; holding field effects fixed, each shooter's ability is their average result after the field effect is removed. Forty passes of this across every match and every shooter in the division converge on a consistent set of numbers, where a shooter's ability is informed by every field they have faced and every field is judged by every shooter who was in it.

Small fields say less about themselves, so a field effect is shrunk toward zero by the factor n ÷ (n + 8), where n is the number of rated shooters in that field. A field of eight keeps half of its measured effect. A field of forty keeps five sixths.

Step three: a light correction for thin records

Four matches is a short record, and four good days can produce an ability estimate that is too high. The final rating blends each shooter's fitted ability with the mean ability of all eligible shooters in the division, weighted as if that mean were two extra matches. A shooter with four matches is pulled one third of the way toward the mean; one with twenty is pulled less than a tenth. This is a deliberate choice to favour the shooter with the long record over the one with the hot streak when the two look otherwise equal.

Step four: four rankings from one rating

Shooters are ordered by rating within each division. The same order then produces a rank in the shooter's state, their USPSA Area, their country, and worldwide. Home state and country are inferred from where a shooter competes most, since neither is published. The worldwide rank is shown only for shooters who have competed outside their home country in the window; for everyone else it reads not applicable, because a rating built entirely from one country's fields has never been tested against another's.

What the rating does not do

• It does not use classifier stages, and it is not a prediction of USPSA class.
• It does not weight major matches more than club matches. A Level III match matters more only because its field is usually stronger, and the model measures that directly rather than assuming it.
• It does not weight recent matches more than older ones inside the twelve-month window. A rating is a statement about the whole year.
• It does not know about equipment, weather, squadding or sandbagging. It knows only who beat whom and by how much.
• It can rate a shooter above 100. Percent of the winner caps the winner at 100, so a shooter who wins by wide margins pushes everyone else's percent down, which the model reads as a strong field. The winner's adjusted results then sit above 100, and the rating says by how much that shooter would be expected to beat an average-field winner.

The rating is labelled provisional because it has not yet been validated against future results. The test is straightforward and the data to run it exists: fit the model on one year, then check whether it predicts the next year's finishes better than classification does. Until that is done, treat the numbers as a well-reasoned estimate rather than a measurement.

Data behind it

Results come from the match reports posted to PractiScore by match directors. Member numbers, divisions and classes are as entered at registration, so a mistyped number at sign-up becomes a separate shooter here. Some posted matches have reports PractiScore itself cannot serve; those are missing from every count on this site, and they are missing in the same way from any other analysis built on the same source.