Why I Have an ISO Issue

The stat ISO, short for Isolated Power (ISO = slugging percentage – batting average), according to Fangraphs is “is a measure of a hitter’s raw power.” The idea is that by subtracting out batting average (AVG) you are “isolating” the power component of slugging, and ISO is now frequently cited in statistical baseball conversations. I’ve never understood why it has become so popular, because I find it to be a bit clunky.

The problem is that while ISO is a step in the right direction, I think it has a lot of the same problems as Slugging (SLG). SLG tells you how many bases a player gets per at-bat, no matter how the player gets them. For a long time we used SLG as a measure of power, but we all knew that this came with some problems. For example, a .450 SLG is usually indicative of a hitter with solid power, but if a player was batting .400 then the .450 SLG would actually mean the player had almost no power.

The same type of issue comes up with ISO. Fangraphs actually agrees, noting in their glossary that “you need to be careful because not all .200 ISO are created equally.” ISO tells you the number of extra bases that a player gets per at-bat (instead of just total bases like SLG), but again, it doesn’t matter how the player gets them. Let’s look Votto (.194 ISO) and Bruce (.195 ISO) as examples, since their ISOs are basically identical.

  • Votto: 41 hits, 69 total bases, 1.7 bases per hit.
  • Bruce: 26 hits, 51 total bases, 2.0 bases per hit.

So, Bruce hits for more power when he gets a hit, but Votto has gotten more extra-base hits and extra base hits overall. ISO counts them the same because it doesn’t effectively separate the frequency of extra-base hits from the magnitude of extra base hits. A guy with a low batting average but a lot of HRs can have the same ISO as a guy with a higher batting average and more doubles.

Does This Matter?

Thinking about this brought back a question that I’ve thought about for a long time and never really looked into: is it better to get your team’s power in the form of one homerun or two doubles? Is it better to have extra-base magnitude or extra-base frequency, or does it not matter? ISO would say that it doesn’t matter, that Jay Bruce and Votto have the same amount of “power.”

To look at this I pulled together team data for runs, walks, SLG, and ISO for the seasons 1980 through 2014. I then created a new stat Bases per Hit (BPH), which is just what it sounds like: total bases divided by hits (or SLG/AVG). BPH is a true measure of a player’s power in isolation of how many hits they get. I also created a stat to measure the frequency of extra base hits (FXB), like a batting average but only for doubles, triples, and homeruns, expressed as a percentage. FXB = (2B + 3B + HR)/AB.

Here are the averages of each of those stats for the sample:

Runs per team per year: 720
Walks per team per year: 522
Team SLG: .406
Team BPH: 1.55 (average hit goes for 1.55 bases)
Team FXB: .083 (8.3% of teams’ hits go for extra bases)

I did a linear regression to look at the correlation of BPH and FXB with runs scored, and I included walks to control for teams that have more base runners. The results were really interesting.*

  • A percent of FXB was correlated with a change of 80 runs, so for each 10th of a percentage point above 8.3 you would add 8 runs to 720 if you were trying to predict a team’s runs scored at the end of the year.
  • A point of BPH was correlated with a -419 change in runs, so for each 100th of a base per hit above 1.55 you would subtract 4 runs from 720.

To unpack that, the analysis suggests that if you hold walks constant, you want to maximize the frequency of extra-base hits, while minimizing your team’s bases per hit. Obviously those two things go in opposite directions, because the more extra base hits, the higher your bases per hit. We don’t want to take this analysis into an absurd place where singles somehow are the gold standard. Clearly homeruns are great, and in any given at-bat, they are the best possible outcome.

But what this analysis suggests is that while you definitely want to hit for power, hitting more extra-base hits is a better way to get that power than getting more bases per hit (essentially relying on the long ball). Or put another way, since not all ISOs are created equal, Votto’s ISO is a more helpful kind of power for scoring runs than Bruce’s.

Which Brings Us to the Reds

There has been something bothering me about the Reds offense that I couldn’t quite put my finger on until we started playing the Royals: the 2015 Reds never hit doubles. As of this writing the Royals are tied for the most doubles with 87, and the Reds are dead last with only 40, which is 11 less than the next worst team. The Reds are an interesting case because while they’re dead last in doubles, they are 4th in total team homeruns (with 48, which also makes them the only team to have more HRs than doubles), and 10th in team ISO (.153).

But what my analysis shows is that while by most measures of power (SLG, ISO, HRs) the Reds are an average or above power hitting team, they are going about it the wrong way. The Reds FXB (percent of at-bats that result in an extra-base hit) is 7.3% a full percent less than average, and 24th in the league. Their BPH (bases per hit) is 1.64, .09 higher than average, and 7th in the league. So while they are getting lots of bases per hit because of the long balls, they aren’t getting enough extra-base hits per AB, and their team ISO may be a little deceptive because of this. By the number that seems to mean the most (frequency of extra base hits), the Reds are a substantially below average power hitting team.

* (For the nerds, all of the coefficients were significant at the .0001 level and the R2 was .83).