Pitch quantification
≈ 6 min read
Names in other languages
No names in other languages are held for this entry yet.
Article bodies are English and are not machine-translated. A name in another language is not a translation of the article.
In baseball, pitch quantification is the attempt to describe the quality of a pitch using a single numeric value based on quantifiable aspects of an individual baseball pitch. There are two main kinds of pitch quantification. The first is outcome oriented. This means that the result of a given pitch (i.e., walk, out, home run, etc.) is a component used to calculate the overall numeric value that describes the quality of the pitch. The other kind of pitch quantification does not consider the outcome of a pitch when calculating quality. Rather, it is batter independent. Its quality can be assessed without regard to what the batter does with the pitch. In 2006, PITCHf/x cameras were installed in every MLB stadium. These cameras are able to track “velocity, movement, release point, spin, and pitch location” on every pitch thrown. When this data was released to the public, many different attempts at pitch quantification began appearing. In 2010, Nick Steiner explained that pitchers have relatively very little control over the results of their pitches due to the fact that so many other factors affect a pitch, such as the batter, the umpire, the defense, and the environment. The task of baseball statistics attempting to quantify a pitch is to isolate the performance of the pitcher from the factors out of his control.…
In baseball, pitch quantification is the attempt to describe the quality of a pitch using a single numeric value based on quantifiable aspects of an individual baseball pitch. There are two main kinds of pitch quantification. The first is outcome oriented. This means that the result of a given pitch (i.e., walk, out, home run, etc.) is a component used to calculate the overall numeric value that describes the quality of the pitch. The other kind of pitch quantification does not consider the outcome of a pitch when calculating quality. Rather, it is batter independent. Its quality can be assessed without regard to what the batter does with the pitch. In 2006, PITCHf/x cameras were installed in every MLB stadium. These cameras are able to track “velocity, movement, release point, spin, and pitch location” on every pitch thrown. When this data was released to the public, many different attempts at pitch quantification began appearing. In 2010, Nick Steiner explained that pitchers have relatively very little control over the results of their pitches due to the fact that so many other factors affect a pitch, such as the batter, the umpire, the defense, and the environment. The task of baseball statistics attempting to quantify a pitch is to isolate the performance of the pitcher from the factors out of his control. Over the years, many different baseball statisticians have attempted to create a statistic that does this. For example, suppose that a batter hits the ball and runs to first base when the count is 0-2 (2 strikes). In order to calculate the linear weight of this pitch, we must compare this run with the average run. An average single is worth 0.47 batting runs. The value of a single on a 0-2 count is -0.106 , according to Table 1. The final value is calculated by subtracting the value of the count when the ball was put in play from the value of the plate appearance according to batting runs. Thus, 0.47 - (-0.106) = 0.58 . rating = -2.51rise + 1.88breakpoint – 0.47knee_dist + 0.51total break For example, suppose a single pitch has a 3” rise, 0.47’ total break, 21.5 break point, and 8” location change. The calculation of the GI would be the following: GI = (-2.51)(3)+(1.88)(21.5)-(0.47)(8)+(0.51)(0.47) After publication of the Greiner Index, it was updated to include all pitches with a 2D location parameter. By combining GI with speed, QOP was developed. The proprietary linear model for QOP is not published, and is patent-pending. The following components affect the overall QOP score in the following way: Increased rise → lower QOP (for curveballs) Increased total break → higher QOP Increased late vertical break → higher QOP Increased horizontal break → higher QOP Closeness to corners of strike zone → higher QOP Increased velocity → higher QOP Greiner and Wilson desired to develop a statistic that could be used by pitching coaches and scouts to develop and determine pitcher potential. Furthermore, in 2015 they suggested QOP has the potential for predicting (and therefore preventing) injury, as well as batter quantification. The QOP statistic is calculated by QOPBASEBALL. C = pitch count D = pitch descriptor The pitch descriptor (D) is determined by a number of select covariates: pitch location, speed, type, handedness of the pitcher, etc. To account for the complex relationship between the pitch quality and the covariates, a random forest methodology is used to obtain an estimation of the overall pitch quality. The random forest method where “ “important” covariates are determined by the splits in the tree...is attractive when we do not know in advance which variables (e.g. pitch location, pitch speed, pitch type, handedness) are predictive.” They have used this statistic to describe the change in pitch quality during a game, and also to evaluate the skills of pitchers, much like ERA.
References
Text derived from https://en.wikipedia.org/w/index.php?oldid=1335029705 under CC BY-SA 4.0; modifications have been made.