DeepCount xA

Method

How the BaseState Player Value Model values every player in runs, what each column measures, and the checks behind it.

The BaseState Player Value Model, the model behind Deep Count Analytics, values every plate appearance, batter faced and defensive play in one currency: runs above or below the league average. Each player's runs come from the same play-by-play the scores do, the league adds to zero, and the players on a club add up to that club's total.

This page is the short version. The technical white paper documents every element, the formulas, and the tests that chose between alternatives.

Principles

  • Context-neutral. An event is worth the same whatever the base-out state or the score. When things happened is left out, because it does not repeat.
  • Credit goes to the player responsible. Pitchers are credited for strikeouts, walks and the quality of contact they allow; fielders for what happens to the ball after that. Nothing is counted twice.
  • Average, not replacement. Zero is a league-average player in the same playing time.
  • Everything adds up. Players sum to teams, and the league sums to zero.

What each column measures

Batting
Each plate appearance valued at the league-average run value of its result in every base-out situation, so a hit counts the same with the bases empty or loaded. Outs are split by type (strikeout, grounder, fly, liner, popup, bunt), and reaching on an error counts. Adjusted for ballpark.
Situational outs
Value the hitter got from his ball-in-play outs beyond the average for that kind of out in that situation: avoiding double plays, moving runners, scoring the runner from third. Rated per plate appearance with runners on and fewer than two outs, then applied at the league-average number of those chances, so hitting behind high-OBP teammates is not rewarded.
Baserunning
Extra bases taken and outs made on the bases compared with an average runner on the same kind of ball (same result, base, outs and exit velocity band), plus steals, caught stealing and pickoffs. Rated per trip on base.
Fielding
Defensive Runs Saved without its catcher parts, centered so the league adds to zero. DRS tracked team batted-ball results best of the public metrics in 2025 and 2026. The Statcast figure is shown in each player's breakdown as a second opinion.
Catcher
Statcast framing, blocking and throwing. Those runs are taken back out of the team's pitchers in proportion to batters faced, so nothing is counted twice.
Position adj.
FanGraphs' positional adjustment. It is for comparing players across positions and is left out of team totals, because every team fields one of each position.
Strikeouts & walks
Strikeouts, walks and hit batters against an average plate appearance, with every ball in play counted at the league-average value. Minus the pitcher's share of his catchers' framing.
Contact quality
The expected value of the batted balls a pitcher allowed, from exit velocity and launch angle, against the league average. What the fielders then did with them belongs to the fielders.
Running game
Steals, caught stealing, pickoffs, wild pitches and balks against the league rate, minus the catchers' throwing and blocking share. Pitchers' own fielding is shown apart.

On this site, Value is a hitter's runs plus his position adjustment.

Left out on purpose

  • Timing: when hits and outs happened. A hit with the bases loaded is worth the same as one with them empty. Timing did not repeat from 2025 to 2026 for teams.
  • Home runs beyond what the exit velocity and launch angle predicted (ballpark and luck).
  • The share of batted-ball results that fielding metrics do not credit to a fielder.

On the Teams view these show up as the % difference: actual run differential minus the players runs total, as a share of all runs scored and allowed in the club's games.

Checks

Checks

  • Loading…