Measures a player’s total contribution (including hitting, baserunning, defense, and pitching) to their team in a single value. It compares each player to the theoretical “replacement-level” player. The idea is to measure how much more (or less) productive the player is compared to the player that would be “readily available” to replace them.

Credits hitters for the value of their swing decisions independent of results.

The model is built on factors including game context (the count, hitter/pitcher handedness, etc.), stuff metrics (pitch velocity, movement, approach angles, etc.), and most importantly, pitch location.

An estimation of the average speed of the barrel of the bat at the point of contact.

Quantifying the “nastiness” of each pitch using only its physical characteristics. 100 represents league average for each pitch type, and thus a league average Fastball (100) is not expected to perform as well as a league average Slider (100).

Quantifying command by normalizing each pitcher’s stuff attributes and measuring how the pitch’s location affects its performance relative to all other pitch types.

Modeling overall pitch quality by combining factors of stuff, command and deception (pitching tunneling, movement unexpectedness based on release traits, count-dependent pitch usage variation, release anomaly score, etc.). Each pitch is relative to all other pitches across pitch types.