Braving New Territory: SABR Chicks Dig the ISO

Everyday Stats and Their Flaws

Stats We're Talking About: SLG

A few weeks ago, we talked a little bit about the issues with slugging percentage (SLG), but we talked about it in the context of an overall offensive indicator. It definitely isn't that. It tells us something, but when we take a look at a hitter's overall production at the plate, we're more concerned about getting on base in the first place, though we're certainly also concerned with how many bases they move once they get on. But today, we want to talk about SLG in regard to how we most often use it - an indicator of power.

One can certainly argue that it's not the intention of SLG. The idea with SLG was to tell us how many total bases the hitter would get on average when he didn't walk, etc. The issue with that argument is ... what do you do with that information? Not a whole lot to be honest. We don't really sit there wondering how many total bases this player usually gets.

Anyhoo and before we get too confused, the most common reason we use SLG is to get an indication of power. BA and OBP tell us how many times he gets on bases, and SLG tells us the damage the hitter does, right? Weellll, kind of.

What SLG really tells us is how many bases a player gets after a certain hit - 1 for a single, 2 for a double, 3 for a triple, and 4 for a home run. That's fine. It's a fact that this happens, but it's not the way we use it. We want to know how much power a player has, and is a triple really hit with more authority than a double? What about that floater down the line that turns into a double? What about the home run that barely clears the wall? A few feet can make a massive difference in the score a player receives in SLG, but if he didn't hit harder, it shouldn't really help/hurt him.

The next issue is how batting average can affect slugging percentage. Take a look at Prince Fielder and Chris Johnson. They both had SLG of .457 last season, but Johnson hit 12 HR and 34 2B while Fielder hit 25 and 36. Now who had more power? Fielder, of course. Why do they have the same SLG? Because Johnson had a .321 BA versus Fielder's .279. The singles (not really an indication of power) pushed up Johnson's BA and, as a result, his SLG.

Again, we run into the same issue we've run into this entire the series. There's a certain logic to SLG, but it doesn't quite isolate what we're looking for - power, in this case. What we're looking for is an improvement.

Nuanced Stats and Why

Stats We're Talking About: ISO, HR/FB

In order to take care of the second problem I mentioned - singles boosting SLG - we have ISO (ISOlated power). The formula for ISO is really simple - SLG minus BA. Unlike with adding OBP and SLG for OPS, subtracting BA from SLG is actually mathematically okay because they have the same denominator (AB). What this essentially does is remove the first base component from SLG. You now get no points for a single, 1 for a double, 2 for a triple, and 3 for a home runs, if you want to think about it that way.

This gets us to focus on the extra base part of SLG, and that's the part that shows us the player's power. Looking at Johnson and Fielder again, Fielder's ISO is .178 and Johnson's is .136. That seems better. Even if you don't understand the number yet (I'll get to that in a minute), Fielder's power score being bigger than Johnson's intuitively makes sense, and it makes more sense with the HR and 2B numbers we mentioned earlier. Again, we're not talking overall offensive production. We're just looking at power.

Now, let's look back at the scale of numbers. Here we go:

• .280+ is elite
• .200+ is very good
• .140-.150 is average
• .100 is below-average
• anything below that is pathetic

But that's the relative scale. The mid-1s is basically the average, and things above and below are above and below-average. I think you've got that pretty easily.

HR/FB is an interesting stat as well. So far we've only talked about it with pitchers, but it also works for hitters. In regard to pitchers, 10% is the average, and like with BABIP, hitters have more control over their numbers than pitchers do. 10% remains the average. 6% is pretty weak. 16% is pretty good. And 20% is elite. It gives us a general idea of power, but it does leave out the rest of the extra-base hits a player has, which isn't really right. XBH/PA (eXtra-Base Hits per At-Bat) might be a better indication, but it wouldn't differentiate between the types of extra-base hits.

ISO will go down as one of the easier nuanced stats to explain. It uses SLG and BA, but it does a little more to focus on extra-base hits only, which helps identify "power".

What We Have Left to Accomplish

When it comes to power, I actually think there's a lot left to do. As much as we use ISO, it's mostly because it's simply better than SLG. But it still has some issues of its own.

The first is that it's based on SLG. SLG is a flawed stat itself, so basing ISO off of that doesn't seem to be a good idea. ISO kind of removes the BA issue, but we still have home runs getting way more points than a double, triple more than doubles, etc. And that's a problem. ISO and SLG also don't really get to the value of hitting for power.

One of my ideas (and I believe it has been mentioned elsewhere, but I can't find it at the moment) is to use extra-base portion of wOBA. I call it wHOA ... get it? ... "Whoa, look how far he hit!" ... Anyway, you can essentially take away the first base portion of wOBA. Instead of ...

You get something more like ...

wHOA = (0.34*2B+0.34*3B+1.05*HR)/AB

I haven't really delved into it all that much - maybe I will for a future post - but it would do a couple things. One, it doesn't differentiate between doubles and triples. The difference there is mostly speed, anyway, and wHOA would focus more on the value of the power and not just how many bases you get. And two, it looks at the added value of that power instead of just how many more total bases this guy gets.

The issue, however, is that home runs still get a lot more value than a double, which makes sense and doesn't. A home run is certainly more valuable, but it doesn't necessarily require a lot more power. And none of these stats account for park, which is certainly an issue.

In the end, what we need is HIT f/x info that will tell us bat speeds, batted ball speeds, batted ball trajectories, and distance. That would give us all sorts of great information, and it would focus more on distances instead of whether it went just enough to get over the fence or not.

Until we get there, using the batted ball data from BIS might be helpful. Using the same info they use for defense, you could use it for hitters. This time, you would see where they normally hit the ball and adjust with the batted ball type. The issue would be how to account for home runs, but home run distance is at least tracked and can be added in somehow.

In the end, ISO has its issues, but it is better to use than SLG. Again, we're looking for "perfect", but we'll use "improved".

Central Lessons

• SLG tells us the amount of total bases that a hitter will get on average when he comes to the plate.
• BA can really mess with the overall SLG number.
• ISO literally removes BA from the SLG number with the formula of SLG - BA, and it focuses on extra-base hits, which is the indication of power.
• ISO does have some issues of its own that are not solved through simply removing BA, but it remains a significant improvement on SLG alone.