Grand Canyon Baseball

Defense App

Outs Above Average from TrackMan fielder positioning

Model primer

Every ball in play has an expected out probability.

OAA credits a fielder with 1 − expected out probability when they convert an out and debits the expected probability when they don't. Air balls use a catch probability model (distance to the landing spot, hang time, route direction); ground balls use an out-conversion model (distance to the ball path, exit velocity, spray angle). Models are trained on the full 2025 college season and scored on 2026.

Air OAACatch probability · nearest fielder to landing spot
Ground OAAOut conversion · nearest infielder to ball path
Success Added(Outs − Expected) per chance
Percentilevs D1 fielders at the position

Alternate OAA lens

WAR-style shortstop method

The referenced shortstop notebook builds the same OAA ledger from a different feature set: estimate out probability from the fielder's starting spot, batted-ball location, movement speed, hang time, launch metrics, and a throw/handling-time estimate. The scoring is unchanged: made play = 1 − xOut; missed play = −xOut.

1Model xOutCalibrated probability of out credit on each fielded ball.
2Score the playActual out minus expected out probability.
3Roll up valueSum OAA by player, position, team, and DRS inputs.

2026 GCU roster

Fielding Leaderboard

Purple = lower percentile · white = midpoint · gold = higher percentile. Click a player for the position breakdown and their biggest plays. Sorted by OAA; click headers to re-sort.

National context

D1 Leaderboards

Team defense

D1 Team OAA

Sum of player OAA on chances while the team was in the field.

Team defense

Team Defensive Runs Saved

DRS = rPM + rGDP + rSB, in runs saved vs the level average. rPM converts outs above average to runs using run values from this season's own run-expectancy (RE24) deltas by position and batted-ball type. rGDP is double plays turned vs expected on fielded ground balls with a runner on first and fewer than two outs. rSB is the run impact of tagged stolen-base attempts (success vs caught) against the level-average attempt; steal tagging is sparse in TrackMan, so teams with no tagged attempts sit at 0. Base-out states are reconstructed from TrackMan pitch sequences.