It's Complicated: A Look at James Shields and BABIP

James Shields is a polarizing figure, to say the least. With his poor overall performance in 2010 and Joe Maddon's decision to start him in the ALDS, Shields became a lightning rod for debate between the mainstream and the stats-savvy. On the one hand, people watched Shields pitch and they thought he was garbage. Batters were ripping balls off of him, he was letting up homeruns every start, and he looked like a disaster on the mound. But on the other hand, the stats-savvy pointed to Shields's unsustainable rates of homeruns allowed and hits on balls in play, suggesting that Shields was due for regression. 

With the 2011 season inching closer and closer, Shields's status is becoming an important issue yet again. If Shields can return to his pre-2010 form, the Rays get a solid ace back in their rotation and may have more leeway to trade a pitcher. Shields is not a pitcher that will make grown men swoon (like David Price can), but he's normally a solid, dependable starter. He'll go late into games, limit walks, keep the score close, and finish the year with 200 innings pitched and an ERA just under or above four. In fact, when you look at all starting pitchers from 2007 to 2009, Shields had the fourth most innings pitched and the 25th best ERA. Durability and run prevention - what else can you ask for in a pitcher?

Many people, though, don't think Shields will be able to rebound and want to move him to the bullpen. The argument is that he's a "hittable" pitcher,  letting up a large number of homeruns through his career and posting a career Batting Average on Balls in Play (BABIP) that is slightly above the league-average (.316 vs. .302). Detractors point out a three-year trend in Shields's numbers: more hits, more homeruns, fewer swinging strikes, worse ERA, and worse Fielding Independent Pitching (FIP). If you look at these numbers in a vacuum, he does appear to be spiraling downward.

So what's the answer? Is Shields a worthless lump, or should we expect him to rebound? We've discussed this issue many times before, but I've recently been doing a lot of thinking about BABIP and what it truly means. It's...well, complicated.

Batting Average on Balls in Play (BABIP) is simply a measurement of how many balls in play against a pitcher fall for hits. The league-average rate is around 30% (AKA, .300 BABIP), and saberists argue that pitchers don't have much control over their BABIP rates. Once a ball is struck, it's up to the defense to get the batter out, and there is little a pitcher can do about it. BABIP rates fluctuate wildly from year to year, and rates that stray too far from the average (like a .230 or .360 BABIP) will likely return close to normal the following year. We normally refer to low or high BABIP rates as "lucky" or "unlucky", since the pitcher is apparently benefiting (or hurting) from the performance of the defense behind him.

I used to believe this was all 100% true, but the more I think about it these days, I've realized it's way too simplistic an explanation. Here's my most recent explanation of BABIP, published over at Beyond the Boxscore: 

Why do players over- or under-perform on single-season BABIP rates? Why does BABIP fluctuate so much? We tend to simplify and call this fluctuation "luck" or "random variation", but I see it as a combination of three influences:

a) Defense - If a pitcher has a collection of stiffs fielding behind him, then he should let up more hits on balls in play than a team with a superior defense. Also, defensive shifts can help or hurt players. As an exaggerated example, a batter that consistently hits into a shift may have a lower BABIP than a typical player.

b) Minute Changes in Talent Level - Over the course of a season, players can go through periods of adjustment (AKA "slumps"). Maybe pitchers adjust to a weakness that a batter has, and the batter starts making less solid contact and getting fewer hits. Maybe a pitcher is off with their mechanics or batters have learned to adjust to certain pitches, and the pitcher leaves more hittable pitches over the plate than normal. The batters make more solid contact, resulting in a higher BABIP for the pitcher.

c) Luck - Sometimes, even with a great defense, bloop hits can fall in. A batter can turn a nasty pitch into a dribbler that just sneaks past the first baseman. Hits can fall in despite the best pitches and the best defenses - that's just the game.

d) Remember: regardless of the reason why a player's BABIP is fluky, they are still likely to regress the following season. Team defenses change, players constantly make adjustments and improve their skill levels, and luck / random variation balances out.      

So maybe, just maybe, Shields truly was more hittable last season. His problem seemed to be with his fastball location, with the occasional fastball catching too much of the plate and being hit hard. It's tough to tell precisely what was troubling Shields - it's likely a combination of a couple factors - but if batters were hitting the ball harder against him, more hits would fall in and more fly balls would turn into homeruns.

However, even if Shields was more hittable last season, he's still just as likely to regress next season. All pitchers go through slumps and BABIP fluctuations, but BABIP rates still always revert close to league-average. Maybe this is a result of the pitcher making adjustments, maybe it's the result of luck and defense evening out - most likely, it's a bit of both. The only pitchers that would conceivably put up consistent high BABIP rates would be minor-league-caliber pitchers that don't belong in the majors. Unless you think Shields no longer belongs in the majors, you have to admit that he's likely to be better in 2011 than he was in 2010.

When you watch every game, it's tough to remember how small the margin of error is for major-league players. One extra bad pitch in a 100 pitch start can be the difference between allowing no runs and three runs - one pitch, one hit, many possible consequences. 22 hits and 6 homeruns - that's all Shields needed to prevent to make his 2010 BABIP and homerun rate match his career numbers. If he had allowed one fewer hit per game and one fewer homerun every five starts, his season would have looked totally different.

With this frame of reference, is it that hard to imagine that Shields could turn things around in 2011? I think not. A few adjustments, a bit of luck, and one fewer mistake a game - that's all it will take for Shields to revert to his durable, underrated form.