DIGGING DATA: The Predictive Power of ISO

Our data guru Chip Bourne has been running some statistical regressions on different stats this offseason so that we can play around with the data. In our last article, we looked at whether a player’s wRC+ performance at one level is predictive of wRC+ performance at subsequent levels (it’s really not). Today, we are going to like at another statistic: Isolated Power (ISO).

ISO is a measure of a batter’s raw power. The statistic was formulated by sabermetric guru Bill James. Basically, ISO is slugging average minus batting average. It shows us how many extra bases a batter produces per at bat. It’s a descriptive stat. When a batter hits safely, what does he do with the ball? Is he a dangerous power hitter or is he a slap-hitter who sprays singles all over the place.

Obviously, for fantasy baseball purposes, we want the power hitters. Remember that in most standard roto scoring setups, HRs count 3-4x for scoring (HR, RBI, Runs, AVG or OBP). League average ISO hovers around .140. The best sluggers post ISOs over .200. Here’s a chart from Fangraphs:

For this study, we did a regression similar to the regression we did for our wRC+ study. Taking every MiLB season between 2008 and 2018 with greater than 50 ABs, we ran an R-squared regression to determine how well a hitter’s ISO at one level correlated to his ISO at later levels in his career. In general, as the r-squared regression % goes up, the linear correlation and predictive value between the two results are stronger. For our models (analog systems with human behavior as the input), anything under 10% correlation is noise. Once we get a result over 10%, we start taking notice:

Based on this study, it looks like a hitter’s prior ISO has a significant level (> 10%) of correlation with the ISO at his current level. For example, a hitter’s ISO at A-Full Season ball (Sally League, Midwest League) has a 23% r-squared correlation with his performance at A-Advanced ball (Carolina League, Florida State League, California League). Compared to our wRC+ study, this is a striking difference.

This finding is also interesting because, unlike wRC+, ISO is not league or park adjusted. The three A+ leagues really run the spectrum of run-scoring environments in the minors. The Florida State League is well-known as being extremely pitcher-friendly. The California League has some of the biggest launchpads in baseball. However, if you know a player’s propensity for producing extra-base hits at one level, you can make a reasonable guess that he will continue producing extra-base hits a similar rate higher up the ladder.

Here is a visualization of the linear regression for the ISO comparisons between Rookie and A-Short ball in our study. There are some interesting outliers. You can tell from the linear regression that, while there is a trend of correlation here, it’s not certain. You can have a batter post a .100 ISO at Rookie ball and a .425 ISO at A-Short ball (and vice-versa!).

Based on our study, you can use ISO as one of your player evaluation tools with some confidence. When looking at a prospect’s profile, know that his ISO is likely to follow him throughout his career. In other words, a batter who racks up a lot of extra base hits is likely to continue doing so, while a slappy singles hitter is also likely to stay that way as well. In general, look for prospects who have posted ISO over .200, betting that they are able to continue that great level of performance throughout their professional career.

One thought on “DIGGING DATA: The Predictive Power of ISO

Leave a Reply

Your email address will not be published. Required fields are marked *