March 25, 2010
Ahead in the Count
Predicting BABIP, Part 3
In Part One of this series, I updated my model for projecting BABIP with new 2009 data, and in Part Two, I explained what makes BABIP Superstars and BABIP Trouble-Makers. In this final part, I will discuss some of the hitters where my Expected BABIP (E-BABIP) projections and PECOTA’s BABIP projections differ most, and discuss which number you might want to trust. PECOTA incorporates a lot of information that my model simply does not, but the batted-ball information can be particularly important for certain hitters, and those are the ones where you should place some faith in E-BABIP.
PECOTA is particularly pessimistic about Jones in 2010, but E-BABIP seems to expect a solid rebound. The problem with E-BABIP in this case is that it is not using any data other than 2009, because Jones failed to get 300 plate appearances in 2008, but had a disastrous BABIP of .229 then and a .242 BABIP the year before. However, PECOTA does not know the breakdown of Jones’ batted balls. Jones had only a .543 BABIP on line drives, well below the league average of .730, and Line-Drive BABIP is notoriously inconsistent. If Jones had a normal LD-BABIP, he would have had an overall BABIP around .260. Jones had a low, but reasonable line-drive rate of 17 percent last year. He has good power, but a poor contact rate.& He also isn’t all that slow, all of which combine to indicate he should have a nearly average BABIP. His pop-up rate is a little high (10 percent) and his ground-ball rate is a little low (34 percent), both of which indicate he has an upward plane to his swing and is prone to somewhat low BABIPs. My best guess is that Jones’ true BABIP skill level probably lies somewhere between these two projected numbers. On one hand, he has batted-ball numbers that indicate he should only be slightly below average, but several years of BABIP problems indicate a trend. I think .270 would be a good guess.
The numbers all seem to indicate that Kotchman should be about average at BABIP, but PECOTA sees him as worse than 90 percent of the league at BABIP skill. He has somewhat of a downward plane to his swing (with a ground-ball percentage ranging from 51-53% and a pop-up percentage ranging around seven) but he also isn’t especially good at hitting line drives (16-19 percent the last three years). He makes good contact but doesn’t have much power. He isn't very fast, as evidenced by his infield reach rates ranging between 6.7-8.6 percent, well below league average of 11 percent, but still gets balls to land in the outfield pretty well (.284, .193, and .250 outfield fly-ball BABIPs the previous three years). Part of the reason that his BABIP has been low is too few ground balls reaching the outfield, but that is not an especially persistent statistic. All in all, I think PECOTA is probably low here.
PECOTA knows more information than E-BABIP here, because Hairston only got over 300 PA last year (433), after several years under that cutoff. His pop-up rate was only nine percent but ranged between 14-18 percent in 2006-08. He is good at reaching on infield ground balls (17 percent last year, and 14 and 15 percent the previous two years, all above the league average of 11 percent). His BABIP was partly low last year because so few of his ground balls made it to the outfield (just 11 percent; league average is 17 percent), but primarily low because of his .589 BABIP on line drives (league average .730), which is also not all that persistent. Just 10 extra hits on line drives would make his LD-BABIP average and give him a league-average BABIP. Overall, I’d take E-BABIP here.
Overbay is able to get a lot of line drives (22-23 drives the last three years) and avoid popups (5.2, 3.2, 3.9 percent the last three years). A bad sign is his declining speed (infield reach rate just 4.2 percent last year), as is his rate of getting ground balls to reach the outfield (13 percent last year), but Overbay might look like he is struggling to make contact too because of his strikeout rate jumping (from 16 percent to 18 percent to 19 percent the last three years), but that is actually because he is swinging less. His contact rate has remained steady and high. E-BABIP is probably high here, as Overbay is getting older, but .282 seems way too low.
Pudge still has a solid downward plane to his swing, even today, generating ground-ball rates of 53, 56, and 54 percent over the last three years, all while keeping his pop-up rate low at 3.9, 3.9, and 2.7 percent. His ground-ball rate is less useful because of his virtually non-existent speed (6.5, 6.3, and 7.3 percent infield reach rate the last three years). His line-drive rate fell last year to 17 percent from 21 percent in 2008, and he is rather old. I would guess that much of his BABIP decline is due to aging, so PECOTA probably has the right idea here. However, it also probably is not quite aware of the downward plane of his swing, so I would say he’ll split somewhere in between the two estimates.
E-BABIP only has one year of data to work with to project Rasmus’ BABIP, while PECOTA also has his minor-league numbers. As he was not especially good at getting hits on balls in play in the minors, PECOTA sees a particularly low BABIP for Rasmus. His ground-ball rate was just 37 percent last year, but his pop-up rate was only 6.7 percent. He is fast, reaching on 17 percent of infield ground balls, but his outfield fly-ball BABIP was just .084 (about half the league average). He has some power, and he does hit the ball hard. Overall, perhaps E-BABIP is a little high here, but there isn’t much about Rasmus that looks like a hitter who is going to be among the bottom of the league in BABIP. He might be a little below average, but he’s not slow, he has power, and he does not seem extremely pop-up prone.
PECOTA was notoriously high on Wieters, and indeed sees him as a borderline BABIP Superstar to this day. His batted-ball rates were pretty much average as a rookie last season, but he had an extraordinary BABIP because of the rate at which his ground balls reached the outfield (23 percent) and his BABIP on outfield fly balls (.263). Wieters had high BABIPs in the minor leagues (.381 in 2008, and .358 in 2009), but is unlikely to maintain one with average batted-ball rates, a relatively high strikeout rate, and not enough power to justify a huge BABIP. E-BABIP is probably a little low, but PECOTA is very likely too high here.
Prado has fairly average batted-ball rates all around. He isn’t especially good at reaching on infield ground balls, but he has gotten quite a few ground balls through to the outfield. He also makes good contact, which is indicative that he is squaring up the ball well. Even so, he does not have much power, and without being a speedy player who hits a lot of ground balls, .324 is way too high. E-BABIP is probably closer to the truth here.
Hall has somewhat of an upward plane to his swing, with ground ball rates ranging between 37-41 percent in recent years, while his pop-up rate has ranged from 10-12 percent. He is slowing down, too, reaching on only 5.3 percent of infield ground balls last year. His contact rate is falling as well, down from 75 percent and 74 percent in 2007 and 2008 to 69 percent in 2009. His BABIP was only as high as it was this year because his line drives were hits at a .780 rate. E-BABIP is probably right here, too, and PECOTA is probably too high. There is nothing about him that indicates he is league average at BABIP skill.
Punto has a somewhat high ground-ball rate, ranging from 53 to 48 to 52 percent over the last three years. However, Punto has particularly low line-drive rates (12, 15, and 15 percent), and fewer of his ground balls have reached the outfield over the last three years (15, 18, and 12 percent). He did have a .787 BABIP on line drives last year, which certainly indicates some luck. Punto has a light bat, and that’s why he is not very good at getting hits on balls in play. Although he may be fast, it’s not enough to justify an above-average BABIP. E-BABIP seems right here, too.
PECOTA sees Fowler as an elite BABIP Superstar, but E-BABIP sees him as merely above average. PECOTA has a tendency to lump speedy players in with each other, but Fowler has an average batted-ball profile. He is above average, because of his speed; he reached on 24 percent of his infield ground balls last year, double the league average. However, he also had a BABIP of .224 on outfield fly balls, which he probably won’t maintain with his low power numbers. He also is somewhat strikeout prone, so he is not necessarily squaring the ball up too well. E-BABIP seems much more realistic here.
One year of hitter BABIP certainly is mixed with a lot of noise, but the skills involved in having a good BABIP should be pretty apparent. Some of these skills are things that PECOTA knows about, such as speed, power, and strikeout rate. However, it does not know his batted-ball rates or which batted balls he got his hits on. These are the hitters where you can take on a projection system and find an area where there are gains to be made. Certain hitters have skills that are not immediately observable by BABIP. Each year, nearly every projection system has underestimated Derek Jeter and Ichiro Suzuki, because their skills are not apparent in a box score. PECOTA certainly seems to be getting it right with Ichiro this year, but there are still plenty of hitters where using this model will help you. I have put together a google doc where you can see the list of 2010 E-BABIPs. Keep in mind that this system does not incorporate park effects or aging, nor does it have the ability to look through tens of thousands of historical players like PECOTA. However, it has some information that should be useful to anyone who wants to truly go beyond what a projection system has to say.