Glossary

Deception Metrics

Metric	Definition
Tunnel Spread	Measured in inches, this is the usage-weighted average distance between the pitcher's pitch types 35 feet in front of home plate (the chosen tunnel point). For every pair of pitch types, we take the straight-line distance between their average locations at 35 ft, weight it by how often that pair is thrown, and average across all pairs. A lower value means the pitch types appear in nearly the same location at the tunnel point. TLDR: the spread of the pitcher's arsenal 35 ft from home.
Plate Spread	Measured in inches, this is the same usage-weighted pairwise distance as Tunnel Spread but taken as the pitches cross home plate instead of at the tunnel point. The higher the value, the more dispersed the pitcher's arsenal is by the time it reaches the plate. TLDR: the spread of the pitcher's arsenal at home plate.
Late Divergence	Plate Spread minus Tunnel Spread: the extra separation the pitches gain between the 35-ft tunnel point and home plate. Measured in inches. A high value indicates that the pitches look alike early and then split apart late. TLDR: how much the arsenal separates after the tunnel point.
Velo Spread	The usage-weighted average velocity gap between the pitcher's pitch types. For every pair of pitch types, we take the difference in their average velocity, weight it by how often that pair is thrown, and average across all pairs. Measured in mph. Higher means more speed separation across the arsenal. TLDR: the spread of pitch speeds in a pitcher's arsenal.
Arsenal Entropy	The Shannon entropy of the pitcher's pitch-usage mix, capturing how many pitch types he throws and how evenly he mixes them. Higher means more pitch types thrown in more even proportions; lower means the pitcher leans heavily on one or two pitches. TLDR: how unpredictable a pitcher's next pitch type thrown is.

A note on the 35-foot tunnel point: The tunneling metrics above are all measured 35 feet in front of home plate rather than at the traditional “decision point” of roughly 25 feet (the spot where a batter is generally thought to commit to his swing). We measure at 35 feet because Late Divergence correlates more strongly with both Whiff% and Chase% there than it does at 25 feet. This pattern holds true for every season from 2021-2026. This suggests that deception, particularly between pitch types, may be most important shortly before the batter's decision point as opposed to at that moment.

Pitch Model Metrics

Pitching+ and Stuff+ have significant predictive power. Both forecast a pitcher's next season ERA better than his current ERA does. Shown below is the correlation of Pitching+ and ERA for predicting the following seasons ERA. This correlation only includes pitchers who had at least 75 IP in back to back seasons from 2021-2026.

Metric	Definition
Pitching+	Every pitch thrown is graded by a chain of models predicting swing probability, whiff/foul/in-play probability, called-strike probability, and the outcome of balls put in play (single, double, out, etc.), which combine into an expected run value for the pitch. Inputs to these models include the pitch's physical characteristics, its deception relative to the rest of the pitcher's arsenal, its location, and context such as the count and batter handedness. Weighted toward his most recent pitches, these run-value grades are averaged and scaled so that 100 is league average and higher is better. TLDR: a pitcher's overall pitch quality, where 100 is average. Pitching+ → ERA: r = −.44 vs ERA → ERA: r = .25
Stuff+	Stuff+ grades each pitch's expected run value through the same sequence of models as Pitching+ (swing decision, swing result, etc.), but strips out context and location to isolate the “nastiness” of the pitch itself. TLDR: the raw quality of a pitcher's pitches, regardless of location. Stuff+ → ERA: r = −.40 vs ERA → ERA: r = .25
Each metric below is derived from the same models as Pitching+. However, they focus on a single aspect of pitching as opposed to overall run value. A higher score always indicates higher pitching ability.
K+	A pitcher's ability to generate strikeouts.
BB+	A pitcher's ability to prevent walks. A higher BB+ indicates better walk prevention.
SwStr+	A pitcher's ability to generate swings and misses.
Whiff+	A pitcher's ability to make hitters miss when they swing.
CalledStrike+	A pitcher's ability to earn called strikes on pitches the hitter takes.
HardHit+	A pitcher's ability to suppress hard contact (95+ mph exit velocity).
Scorched+	A pitcher's ability to suppress scorched contact (105+ mph exit velocity).
Soft+	A pitcher's ability to induce soft contact (80 mph or less exit velocity).
Weak+	A pitcher's ability to induce weak contact (75 mph or less exit velocity).
GB+	A pitcher's ability to induce ground balls (launch angle below 10 degrees).

All grades are scaled against every qualified pitcher, starters and relievers pooled together: 100 is the average pitcher and each 10 points is one standard deviation. A grade therefore means the same thing regardless of role, though the percentile shown on the Pitcher Grades tab ranks a pitcher within his own role.

Team Grades

Metric	Definition
Team Grade	A single rating of a team's current roster, where 100 is league average and each 10 points is one standard deviation. It combines five weighted components (Batting, Starting Rotation, Bullpen, Defense, Baserunning), each built from the team's rolling player metrics weighted by recent playing time. TLDR: one number for the current roster; 100 = average, +10 = one standard deviation.
Defense+	How often a team converts balls in play into outs compared to what a model expects from each ball's physics (exit velocity, launch angle, projected landing spot) and the fielders on the field. Values above 100 mean the team recorded more outs than expected on the balls it faced. TLDR: outs above expectation on balls in play.
Catcher Framing	Called strikes above expected for the team's starting catcher: on every taken pitch, the expected strike probability from its location and context is compared to the actual call, and the differences accumulate per take. Displayed per 1,000 takes. TLDR: extra called strikes the catcher earns, per 1,000 takes.
IF / OF OAA per 100	Statcast Outs Above Average per 100 attempts for the team's projected infielders and outfielders, shown separately. Each fielder's rate blends his last three seasons with recent seasons weighted more and shrinks toward league average with fewer attempts, so it is a forward estimate that runs milder than raw single-season leaderboards. Positions are weighted by their real share of balls fielded. TLDR: the projected fielders' range, per 100 chances.
SB Run Value	Runs added or lost from stolen-base attempts, per 100 opportunities. An opportunity is a plate appearance in which the runner is on base with the next base open (on first with second open, or on second with third open). TLDR: average run value of a runner's steal attempts, per 100 chances to go.
XBT%	Extra Bases Taken percentage: how often a runner takes the extra base as the lead runner when the chance arises — first to third or home on a single, first to home on a double, or advancing on a tag-up. League average is roughly 45%. TLDR: how often a runner takes the extra base.
BR Outs%	How often a runner makes an out as the lead runner trying to advance on a batted ball — thrown out going first to third or home on a single, first to home on a double, or advancing on a tag-up, or doubled off on a line drive or fly ball. Force plays are excluded since the lead runner has no choice. Lower is better. TLDR: how often a runner runs into an out.