In MLB handicapping, you have two distinct areas to focus on: how many runs a team is going to score and how many runs they are going to give up. Hitting statistics play a big part in predicting those numbers, but pitching plays nearly as big of a part in making your prediction. For evaluating different statistics we are going to use correlation.
Correlation is a statistical tool that measures how closely one set of numbers relates to another set. For our purposes, we are going to look at different categories and see how well they track compared to runs allowed. Correlation runs from -1 to 1. The closer you get to either extreme, the closer the two sets of numbers are tied together. The closer the number is to 0, the less likely one number will tell you anything about the other. If the number is negative it means that when one number goes up, the other goes down. If the number is positive they both rise or fall together.
The question that we wanted to know the answer to was, “which pitching stats correlate the highest with runs allowed?” We looked at all of the pitching statistics from the 2010 season in order to correlate with runs given up. Here is what we found:
Complete Games (-.38)
We take the statistics over the course of a season. Obviously, if you can predict a shutout, then there would be 0 runs allowed. If there isn’t a shutout though, this category doesn’t really tell you much about how many runs the other team will score. The more shutouts a team has though, the more outliers they will have to bring their runs allowed per game down.
Hits/9 Innings (.84)
Intentional Walks (-.18)
Hit by Pitch (.43)
This group of stats tells how often a pitcher allows a hitter to get on base. Allowing a lot of hits translates pretty well into giving up runs, but walks aren’t nearly as bad. Intentional walks look like they actually help the defense, probably because only the best hitters are given the free pass in order to get to a weaker batter.
While hits show a high correlation, the stat to concentrate on is On-Base Percentage and WHIP ((walk + hits)/innings pitched). No matter how the batter gets on base, the more base runners there are, the higher the chance one of them is going to score.
All negative correlations here, which means that the more strikeouts a pitcher has, the less likely they are to give up runs. This makes sense since it allows the pitcher to get out of jams and it prevents “lucky hits” or bad bounces from translating into runs scored.
Home Runs (.547)
The type of hits allowed seem have different levels of correlation. With doubles being able to score almost any base runner, plus putting the batter himself into scoring position, it’s not surprising that there is a high correlation with this stat and runs. The low number for triples is a little confusing, since they would also score anyone on base and leave the hitter in a position where he could easily score. Home run power doesn’t seem to be as important as slugging percentage.
The real key is finding pitchers who don’t allow a lot of base runners and don’t allow hard hit balls that turn into extra bases. Combining On Base Percentage and Slugging into OPS takes two of the most important individual stats and turns it into a real eye opener in how many runs a pitcher is going to give up.