clock menu more-arrow no yes mobile

Filed under:

Talking Sabermetrics: Predictive Vs. Illustrative Statistics

This is the first in a season-long series look at different sabermetric statistics and how they apply to the Houston Astros.

What's the biggest mistake people can make with the new repository of statistical analysis out there? Confusing which statistics will tell us how a player will do in the future and how he's done in the past. That's an important distinction to make when we're talking about a player's value.

What do I mean by that? Well, let's look at two very popular statistics that get thrown around all the time to see which category they fit into and then how they apply to the Houston Astros. Two very good ones that we use all the time around here are Wins Above Replacement (WAR) and Batting Average on Balls in Play (BABiP).

For those of you unfamiliar with WAR, it's a way to basically sum up a player's overall value to his team and quantify it as how many wins better he's made the team than a replacement-level player would have at his position. There are different ways of figuring WAR, applying different formulas to reach the win figure. For our purposes, we'll use FanGraphs' WAR for a specific reason I'll get to later.

BABiP is basically batting average, but taking out any strikeouts and home runs while adding in sacrifice flies. What that does is show how successful a player has been on hits inside the foul lines. After the jump, we'll look at each one more in-depth...

There are resources out there where you can go and learn a LOT about the statistical evidence behind the statistics and what specifically they tell us. I'm not going to go into all that right now, because I'm not a statistician and you can go read smarter guys than me talk about it.

But, we will show how these two statistics can be applied to the Astros in general terms. Oh, and to get back to my headline, which ones can be used to predict performance and which ones tell us how a player has done in the past.

WAR is most certainly the latter. Just look at Carlos Lee's WAR total on FanGraphs. Do you honestly think he'll put up another season of 3+ WAR for the Astros? To do that, he'd need to duplicate his defensive numbers. To do that, he'd need to take advantage of a small sample size window in switching positions.

See, Lee had a very good half-season throwing out runners in left field in the early stages of 2011. That translated to terrific arm numbers in his defensive stats at FanGraphs. That, in turn, boosted the fielding component for his WAR. When he switched to first base, that defensive rating didn't fall as much.

The problem is most defensive metrics need more time to normalize. The half-season of data we got on Carlos at each position isn't really a great indicator of how he'll perform over an entire season at either one.

That's a big reason why we can't assume he'll put up a WAR similar to what he did last season. Even if he hits for similar numbers, if his defensive statistics aren't there, he won't post the same WAR.

So, what good is WAR, then? It's a great illustrative way to show who meant the most to the 2011 Astros team. While Carlos was very good, it meant something bad that Hunter Pence and even Michael Bourn led the team in WAR for a long time after they weren't even Astros any more.

BABiP, on the other hand, does a better job of demonstrating predictive properties. It's not an absolute predictor, but it's better than WAR. See, BABiP tends to normalize for pitchers across the league somewhere around .295. Pitchers who's BABiP is much lower than that are at a bigger risk for their performance to tail off in the next season, while a pitcher who's BABiP was much higher than that can expect better results in the future.

Obviously, there are a ton of other factors involved. A high BABiP coupled with a high line drive rate and a falling strikeout rate could be more indicative of an aging veteran who may be losing his stuff more than just bad luck. However, BABiP is a quick, easy way to predict how a player might do in the future.

For those of you just jumping into advanced metrics, it's important to understand what the stats are telling you. Next time, we'll get on what stats should be linked together to give you a clearer picture of a player's performance.