Testing the Corrrelation Between Certain Pitching Statistics and A Team's Wins

Sorry about the gigantic gap between posts. I have been on vacation for the past two weeks and just returned on Friday. I meant to post a few more times before I left, but a few different things came up and I was held back. I will try to pick up where I left off.

This mini-investigation's goal is to deduce the importance of certain pitching statistics when it comes to a team's success (whether they win or lose). I looked at 14 different pitching statistics for this experiment, some more well-known than others. These stats were updated through August 8th. I tried to stay away from stats such as runs, for they would be a little too obvious. I attempted to select statistics that signify all different things in pitching. Some of them, like FIP for instance, I will explain as we go along. Most of them, however, will be familiar or self-explanatory.

Of the 14 statistics I looked at, WHIP had the highest r squared value (0.44). This value is not particularly high, but for the purposes of this experiment, it is sufficient. The next highest was ERA at 0.36. These two graphs are shown below.

































As you can see, teams that have allowed less walks, hits, and earned runs have been more successful in winning games. Only one team with 60 wins or more had a WHIP over 1.40. These relationships are not surprising, as these two statistics have emerged as two of the most prominent indicators of success in pitching. The next highest correlation was that of LOB% (the percentage of runners left on base). This makes sense, since as a pitcher, the less runners you allow to score, the better chance you have of winning. The graph is shown below.

















Of the other 11 statistics, only opponent batting average had an r squared value over 0.3. Here is a list of all 14 tested statistics and their respective r squared values:















Perhaps the most intriguing statistic here is FIP, or Fielding Independent Pitching. This statistic was created earlier this decade in multiple different forms. Its purpose is to evaluate a pitcher based only upon plays that he had a direct involvement in completing, or that his fielders did not participate in. It is scaled in a way to resemble ERA, so that one can determine if a pitcher is being helped or hurt by his fielding. Like ERA, a lower FIP is better. This statistic is widely referred to in the sabermentric world, though is has many limitations. The most important of these shortcomings is that it does not consider the different types of batted balls at all, so a pitcher who allows only 15% line drives, but only strikes out 2 batters for every one he walks will be disadvantaged in FIP even though he pitches to bad contact (a good thing for a pitcher). The limitations of FIP are certainly shown here, as ERA's r squared value is nearly double that of FIP.

Thanks for reading. I think I'm going to post again Wednesday or Thursday, instead of on Sunday, because I'll be visiting my family this weekend. Keep checking back for updates, though.