Simple Sabermetrics: Illustrative versus Predictive

facebooktwitterreddit

David Richard-USA TODAY Sports

There is a difference between describing something current and predicting something in the future. We know this, of course, but sometimes fantasy baseball players, prognosticators, analysts, and soothsayers can sometimes forget, which leads them to use stats in a way they weren’t meant to be used necessarily.

Let’s not beat ourselves up for being forgetful. President Ronald Reagan was famously forgetful.  (I’m going to owe the Dated Reference Library a sizable fine for checking out that reference from the 80’s and not returning it.) Instead, let’s take a moment to remind ourselves that some baseball statistics are better at describing past performance, while others are better at predicting future performance. Then let’s spit shake and promise to never confuse the two.

Flip over a baseball card. Go ahead, I know you have early 90’s Donruss Rated Rookies in a shoebox somewhere. Those numbers on the back of the card illustrate what the player did in previous seasons. Statistics like RBI illustrate how many players the batter knocked in, batting average illustrates how many hits the player got per 1000 at bats, and home runs illustrate how many times the player knocked the ball over the fence.

Statistics such as AVG, HR, RBI, and SB are illustrative statistics. They are a measure of the player’s output; they describe how a player did (or is doing currently).  These are the numbers that baseball fans have grown up memorizing on the back of a card.

But other numbers have crept into the conversation in the past 30 years, many of which are better indicators of future performance. Numbers such as batted ball numbers (LD%, GB%, FB%), K/BB ratios, and the like are more often used as predictive statistics.

If you are a fantasy baseball player and you are thinking about drafting PLAYER A, you may be tempted to look at his batting average the past two seasons, which we’re going to say was .280 in 2012 and .300 in 2103. It may be tempting to take those numbers and simply split the difference, predicting he’ll hit .290 in 2014. While you might get lucky and he indeed hits .290, you are much better off not using an illustrative stat like batting average to predict future batting average. Sound weird?

Not really. There are other stats that most affect a player’s batting average. I’m talking K%, LD%, GB%, FB%, BABIP and speed, which you won’t find on the back of a baseball card. Maybe the reason that PLAYER A saw an increase in batting average from .280 to .300 is that he matured as a hitter and cut down on his strikeouts, while simultaneously hitter the ball harder as line drives. Maybe he was simply lucky and his BABIP was unsustainably high and that was what accounted for the jump from .280 to .300. Whatever the reason, when you are projecting batting average you don’t simply “average” his last two year’s batting averages.

To begin to implement sabermetrics into your fantasy baseball research you don’t need to get very complicated. Begin with something simple like scanning the last 3 years of a player’s batted ball profile to see if one of these predictive numbers looks out of place for you. As you do this more and more you’ll start to develop a discipline an appetite for it. Also be skeptical of any fantasy baseball writer who is using only illustrative statistics to predict future results. Last I heard they are letting just about anybody on the internets nowadays.

You’ve made it to the last paragraph of what I admit is a very boring installment of Simple Sabermetrics. Congratulations, you’re a survivor. I didn’t offer an in depth look at a single statistic like I’ve done in the past as I thought it may be helpful to take a step back, look at the big picture, and simply realize that difference statistics are often just better at describing different things. If you want to take a deeper dive into an individual statistic, you can find those here: OBP and SLG, ISO, Command Ratio, Fielding Independent Pitching, and Plate Discipline,