Advanced Stats Primer: Pitching and Defense Edition

As of this week, the official start of Spring Training is only three weeks away. Since we're getting close to the start of Baseball: 2013 Edition, I thought I would give a basic introduction to some of the advanced stats you'll be seeing on this site and around the internet during the season. There's a lot of information to cover, so I'm going to break it into three parts: last week we covered offense, and today we'll be looking at pitching and defensive stats.

DRS

The goal of defensive runs saved is to quantify how many runs above or below the league average a player has "saved."

Joe Posnanski's quote on the Fangraphs' DRS page describes it better than I ever could:

"...as I understand it, the numbers determines (using film study and computer comparisons) how many more or fewer successful plays a defensive player will make than league average. For instance, if a shortstop makes a play that only 24% of shortstops make, he will get .76 of a point (1 full point minus .24). If a shortstop BLOWS a play that 82% of shortstops make, then you subtract .82 of a point. At the end, you add it all up and get a plus/minus."

Three full years of data are needed before any significant conclusions can be made about a player's defensive abilities. DRS is scaled in the same way as UZR: 0 is average, +15 is Gold Glove, and -15 is terrible.

FIP and xFIP

Fielding independent pitching (FIP), sets out to determine what a pitcher's ERA should be by looking at the parts of pitching that he can control: strikeouts, walks, hits by pitch, and home runs. The formula for FIP is as follows:

FIP = ((13*HR) + (3*(BB + HBP)) - (2*K)) / (IP + constant)

The constant is usually around 3.20, and is used to bring FIP onto the same scale as ERA. FIP is generally more useful than ERA when it comes to predicting future performance, due in large part to the removal of the "luck factor" (pitchers don't really have control over balls in play).

xFIP (expected fielding independent pitching) uses the same formula as FIP, except that instead of using the pitcher's actual home run total, it uses the number of home runs a pitcher should have allowed. This number is found by multiplying the league average home run-to-flyball rate by the pitcher's flyball rate. This is done as an attempt to correct the inconsistencies in a pitcher's year-to-year HR/FB ratios.

FIP and xFIP are measured on the same scale as ERA: 4.00 is average, 3.00 and lower is fantastic, 5.00 and higher is awful.

Strikeout and Walk Rates

The most commonly used statistics used to measure walks and strikeouts are BB/9 and K/9. While these are useful, BB% and K% are better indicators because they take into account the fact that some pitchers face more batters per nine innings than others, and therefore have more opportunity to get both strikeouts and walks.

A pitcher should strike out approximately twice as many batters as he walks. The average strikeout rate is 18.5%, while the average walk rate is 8.5%.

BABIP

Batting average on balls in play for pitchers is based on the same premise and factors as the batters' version: defense, talent, and luck. The pitchers on a good defensive team may have lower-than-average BABIP, because their defenders make plays that those on a poor team would not be able to pull off. Talent can affect BABIP when a player has a "period of adjustment", either good or bad. If, for example, a player is dealing with a nagging injury, his mechanics may be off and his BABIP may rise (it would go back to normal when he's healthy again). And luck, well, is just luck.

As was mentioned before, pitchers have very little control over where balls in play are hit, therefore their BABIP is prone to variation. Typically, though, a pitcher's BABIP will fall between .290 and .300. Any drastic change from within that range can be expected to regress back to the mean in time.

Thanks for reading! We'll wrap this up next week with a look at win probability statistics and WAR. In the meantime, I strongly encourage you to check out FanGraphs' Sabermetrics Library for more information. It's fantastic.