The Ballpark Pal pitching model evaluates every pitch during the season. It looks at how effectively the pitch is thrown (pitcher score) and how well the ball is hit (batter score against).

Pitcher Score: Based on the physical aspects of the pitch such as velocity, break, spin, location, etc. The Pitcher Score ignores actual outcomes, predicting what the outcome will be based on how the pitch travels through the air. This view tends to be more consistent and predictive in small samples than actual outcomes, which can be random.

Batter Score Against: Based on the quality of contact from the batter. This uses launch angle, exit velocity, and direction to predict what the outcome of the batted ball will be (independent of park and weather). This view tends to be more random, volatile, and less predictive in small samples.

Higher scores are BAD

Lower scores are GOOD

This table shows how effective a pitcher has been over his last 1,000 pitches, split by batter handedness and pitch type.

Pitches: The number of pitches thrown

Freq: The percentage of the total pitches

Model: The Pitcher Score on a 0-125 scale (click "what's this" in the graph above for a description)

Hit: The sum of the hit likelihood of balls put in play, divided by the number of pitches. Independent of weather and park.

HR: The sum of the home run likelihood of balls put in play, divided by the number of pitches. Independent of weather and park.

Strike: The number of strikes recorded, divided by the number of pitches. This includes foul balls when there are fewer than 2 strikes.

Ball: The number of balls recorded, divided by the number of pitches.

Red is BAD

Green is GOOD

Ballpark Pal simulations use a "matchup model" to predict the outcome of a batter/pitcher matchup. The model uses dozens of characteristics for both players to produce a probability distribution of outcomes (shown below).

PA: Number of plate appearances by the hitter this season

H: Handedness of the batter

%ile: Every hitter is lined in up in order from 1 (least favorable matchup for the pitcher) to 100 (most favorable). This corresponds to the percentile slider shown beneath the batter on the Matchup Machine page.

Adv: The Advantage from the batter's perspective. Postitive 10 indicates an extreme advantage for the batter. Negative 10 indicates an extreme advantage for the pitcher.

Hit: The model's likelihood that the batter will get a hit

K: The model's likelihood that the batter will strike out

BB: The model's likelihood that the batter will walk

HR: The model's likelihood that the batter will hit a home run

IPO: The model's likelihood that the batter will record an in-play out

2B: The model's likelihood that the batter will hit a double or triple

1B: The model's likelihood that the batter will hit a single

History (Actuals): The actual results from plate appearances during the last 3 years

History (Model): A model predicts the outcome of every ball hit in play based on how it leaves the bat (exit velocity, launch angle, direction, and batter handedness). This helps control for luck, weather, and the park the hitter was in.

Green is GOOD for the pitcher

Red is BAD for the pitcher

Ballpark Pal simulates every game during the year and logs the results. This table compares the sim results to actual outcomes, aggregated for the entire season.

Sim is HIGHER than actuals

Sim is LOWER than actuals