Short-Stop Analytcs #8: Analyzing a Twitter Poll

by Matthew de Marte – August 30th, 2018

In this edition of Short-Stop analytics, we are going to expand on a question we asked our following to answer. Recently, on Twitter, SOTG ran a poll asking our following to answer: What is the gold standard for hitting and pitching? In this article, we will do a short analysis that answers this question. The following snippets are the results of this poll:

So, our following voted that the gold standard of hitting is bat to ball skills and that command is the gold standard of pitching. To begin, we have to start by answering the question, what do we use to quantify these skills? There are many different metrics that can be used to answer what all of these skills are, but for the sake of this analysis, only one metric can be chosen. Here are the metrics I decided to use in this analysis:

  • Bat to ball skills: Contact rate – Pitches in which contact was made divided by swings.
  • Pull side power and Oppo power can have the same metric to define them. ISO or isolated power (slugging-batting average) does the best job of isolating a hitter’s power output so that is what we will use in this analysis.
  • Command seems like it would be easy to quantify, but there are different ways to identify a pitcher’s command. Strike % would be an easy first choice, but is that the best way to quantify command? Strikes are not executed solely from pitchers throwing the ball in the strike zone. Zone %, the rate at which a pitcher throws the ball in the strike zone, is another quality candidate, but it does not account that pitchers try and throw the ball outside the zone ball sometimes. Pitchers often try and throw pitches outside the zone to entice hitters to swing and miss. The true definition of “command” is a pitcher’s ability to place pitches where he wants. Baseball Prospectus wrote a brilliant article on control vs. command in which the authors, after analyzing catch framing statistics, discovered pitchers with better command often miss the strike zone intentionally more often than pitchers with poor command. The metric they use to quantify command is CSAA, which is difficult to export for analysis. Due to this, I will use a combination of strike % and zone %. To equally weight each metric, I will translate a pitcher’s ranking in each category to a percentile and add the two together then divide them giving them a command percentile ranking.  
  • Velocity is the simplest metric to quantify. A pitcher’s fastball velocity will be measured here.
  • To quantify plus offspeed pitches, there a few different ways to go. Actual pitch characteristics, such as Bauer Units, would be useful, but that does not take into account armslot or movement. Movement would be good, but just because a pitch has a good shape to it and a good movement profile does not guarantee it will be successful. Pitch values is a metric thats goal is to pinpoint a number as to how valuable a pitch has been. This appears to be the best way of evaluating a plus offspeed pitch because the best pitch should be the most valuable. In this analysis, curveballs, sliders, and changeups are considered offspeed pitches.

Now that we have settled upon how each skill will be quantified, we can set up our test. This seems best to run a multiple regression model and analyze visualizations for each individual variable. For hitting, we will use wRC+ as our response variable, and for pitching ERA-. There are a wide variety of metrics that can be used for the response variable for pitching, but most people still value ERA as the best metric to evaluate a pitcher’s success, so we will use its park adjusted form that is on a different scale to better account for how much better or worse a pitcher is compared to league average. There is one last thing to sort out in this analysis. Some pitchers throw multiple offspeed pitches. Pitchers such as Max Scherzer have a slider and changeup that perform above league average. Due to this, we will only use a pitchers most valuable offspeed pitch in this regression model.

In this dataset, we will look at 2017 data. It is easier to look a full season’s worth of statistics. Hitters in this dataset must have at least 200 PA, and pitchers must have thrown at least 20 innings. The easiest way to run this test is using a multiple Linear Regression model. If you are not familiar with this test, please review it HERE if you would like.

The following table displays the coefficients and p-values for each hitting variable:

Variable Coefficient p-value
Contact % 187.147 2e ^-16
Pull ISO 107.984 2e ^-16
Oppo ISO 111.426 2e ^-16

The criteria of this test would indicate the variable with the lowest p-value would be the gold standard of hitting. According to this test, all three skills are equally as valuable a skill for a hitter to have to be successful. They all play a significant role in accounting for the production of hitters across baseball. Contact % does have the highest coefficient, but the scales of each metric make it difficult to tell if this coefficient size difference is any indication bat to ball skills are the most significant skill for a hitter to have. What this means to me is hitters across baseball utilize these skills to be successful, some specialize in one particular skill while others perform well in all three areas.

The following table displays the coefficients and p-values for each pitching variable:

Variable Coefficient p-value
Command Score -26.9802 1.62e^-7
Avg. FB Velocity -2.3152 1.85e^-6
Off-speed Pitch Value -2.5609 2e ^-16

According to this model, having a dependable off-speed pitch is the gold standard of pitching. Each variable in this model once again is significant to the model, but having a good off-speed pitch is the most valuable tool a pitcher has. Interpreting the coefficients, for every MPH a pitcher gains on his fastball, and every run an off-speed pitch saves that decreases a pitchers ERA- by 2.31 and 2.56 points respectively. A pitcher could have his command go from 0 to 1, the limit of a pitchers command score, and it will only decrease a players ERA- by ~ 27 %. A pitcher has a much better chance of improving his ERA- by throwing harder, and developing better off speed pitches.

This may not be the best model to find the gold standard of hitting and pitching, but it does provide an answer. Hitting for contact and power is equally important for a hitter, while the most significant contributor to great pitchers are having a quality off-speed pitch. This was a ton of fun to put together. Thank you to our following for interacting with our Twitter poll, allowing us to do something like this. Without your help this article would have never been written!

Leave a Reply