Once You Pop, the Fun Don't Stop

Two weeks ago, we discussed the early-season struggles of Jarrod Parker, and concluded that part of his difficulties stemmed from his inability so far this season to induce infield fly balls. In that article, I claimed that infield pop-ups are essentially equivalent to strikeouts in terms of their benefit for the pitcher, and there seemed to be some confusion in the comments section as to why that's the case. I'd like to spend a little time discussing the recent research into the power of infield pop-ups, and delve into what that means for the 2013 A's. To understand the importance of infield flies, we need to take a big step back, to a long, long time ago...

There's a good reason that for years ERA has been the go-to stat for evaluating pitcher performance. At its most basic baseball is about scoring more runs than the other guys, so using runs allowed as a starting point makes a whole lot of sense. It's a quick and dirty way to see just how good a pitcher is at stopping those pesky other guys from getting runs.

However, ERA can fluctuate wildly from year to year, and more importantly, ERA is affected by many things beyond the pitcher himself. A good defense can make a mediocre pitcher look like an All-Star, while a poor defense can turn a solid pitcher into a DFA. In addition, plain old luck plays a huge role- a slow ground ball might slip in between the shortstop and the third baseman, or it might go for a double play. The difference is a matter of inches.

Today most people agree that luck plays a huge role in whether batted balls go for hits, but in 1999, when defense-independent pitching was first proposed, it was far from a foregone conclusion. In fact, many baseball analysts believed that pitchers had an ability to control were hit. A screaming line drive hit right at the second baseman was a result of the pitcher "pitching to his defense". (Darth?) Voros McCracken, however, wanted a way to eliminate defense and luck from pitcher evaluation. He proposed dERA (defense-independent ERA), which was the first of a range of DIPS (defense independent pitching statistics).

Essentially, the point of DIPS is to only credit (or penalize) a pitcher for things that they are in direct control of- that defense can't account for. The defense doesn't play a role in walks or strikeouts*, so these are main components of DIPS. Home runs that clear the fence are not in play and cannot be affected by a defense, so they're included too. A hit by pitch is all on the pitcher. These three things: walks, strikeouts, and home runs have become known as the "three true outcomes" because theoretically, luck and the defense have been completely eliminated from the equation.

*This isn't entirely true... Balls and strikes aren't completely defense independent, either, but that's outside the scope of this discussion. We will discuss the recent research into pitch framing in a future article.

Today, perhaps the best known of the defense-independent pitching statistics is FIP (fielding-independent pitching), developed by Tom Tango. You'll often hear it thrown around on this site and others. The formula for FIP is:


The 3.10 is added at the end to make the stat look more like ERA, and the weights of the HR and walks are weighted for their importance to run-prevention. xFIP attempts to eliminate luck even further. Home run rates can fluctuate quite a bit from season to season- so xFIP regresses home run rates to league average. Of all pitching statistics, xFIP has the highest predictive value of future ERA.

However, the stat still wasn't perfect. FIP assumes that all pitchers have exactly equal ability once the ball is put in play. However, a few pitchers consistently have BABIPs (batting averages on balls in play) below the league average. These pitchers are well-known as exceptions: Matt Cain, Tim Wakefield, and Barry Zito are a few examples. For a long time, these pitchers were mysteries. What was it that they were doing to consistently beat their FIP and xFIP predictions?

This past March, at the Sloan Analytics conference, Dan Rosenheck presented a piece that noted that infield fly percentage can account for a large portion of this variance in BABIP. Unlike other batted balls, pop ups on the infield almost never go for hits. According to Dave Cameron at FanGraphs, during the 2012 season there were 4377 infield pop-ups, of which 13 went for hits and 28 went for errors. In other words, about 99% of infield pop-ups went for outs. Of course, this isn't 100%, but technically strikeouts are not 100% outs either- the batter can reach first base.

In other words, in terms of the likelihood of the pitcher getting someone out, an infield fly is basically equivalent to a strikeout.

Of course, these things are still dependent on defense, but with a 99% out rate, it is as close to defense-independent as you can come. If a pitcher can induce a weak ground ball, that's a good result, but there's still no guarantee that's an out. On the other hand, if they can induce an infield fly, that's essentially a guaranteed out. Indeed, the research at FanGraphs showed that the same pitchers who consistently beat their FIP and xFIP were the same pitchers who induced high percentages of infield fly balls. This explains our FIP out-performers: Matt Cain, Tim Wakefield, and Barry Zito have career IFFB rates of 12.2%, 14.7%, and 12.9% respectively, well above the league average. This would seem to imply that there is a repeatable ability to induce infield fly balls. I highly recommend checking out the original article for more detail- it's watching real baseball research unfold before our very eyes.

Whew. Now that we've gone through a history of defense-independent pitching and established the importance of infield fly balls, what does this mean for the 2013 Oakland A's?

Here's a table of the infield fly ball rates for some of our starters in 2012 versus 2013 (league average is about 10%):

Some things that immediately jump out to me:

* It is still possible to have success without infield pop-ups- look at Brett Anderson's 2012! However, his career IFFB% is 9.2%- so expect this to regress (in a good way!). An increase in IFBB% for Brett should help him a lot.

* As discussed in the Jarrod Parker piece two weeks ago, expect similar regression from him.

* Last year's success from the pitching staff may have been aided by a very high IFFB%. Every starter (with the exception of Anderson) had an IFFB% higher than league average. This year, almost all are below league average. With the exception of Bartolo, it is difficult to assess what their true IFFB talent is because they are second year players- there's just not enough data to know. I think their true talent is more likely to be somewhere in the middle. Don't expect to see IFFB rates of 16.7%, but you can probably count on rates closer to 10% going forward.