Tuesday, March 8, 2011

The problem with relying solely on sabermetrics

Much to the chagrin of baseball purists, sabermetrics are here to stay.

Many baseball fans are well-aware of the new brand of statistics, but even the casual fan has started to be introduced to the emerging new style of keeping track of how well a player performs.

Sabermetrics are a very useful tool. They dig far deeper into the actual game than traditional stats do. For instance, one of the stats that has long-determined how well a pitcher performs is his win total. Sabermatricians rightfully argue that a win does not necessarily tell the whole story of how well a pitcher pitched in a particular game.

For instance, a pitcher could conceivably throw a complete-game, one-hitter, and lose 1-0, picking up the loss. On his very next outing, he could give up five runs on 10 hits in five innings, and pick up the win because his team scored 10 runs in the first three innings.

Clearly, a win doesn't tell the whole story about whether or not a pitcher performed well on a given day.

However, the problem with sabermetrics is that while they attempt to dig deeper than traditional stats, they often simply trade one flawed stat for another.

An example of this is BABIP (Batting Average for Balls In Play). What this statistic measures is what a hitters batting average is for everything that was not a home run, a strike out, or a foul out. It essentially attempts to measure how lucky a hitter gets. Statistics would say that regardless of how good or bad a player is, 30% of all balls put into play should fall for a base hit.

So when a player has a .212 BABIP, as Chris Iannetta had in 2010, it is easy for fans to suggest that he simply was a victim of bad luck.

On the other hand, Jonathan Herrera, a guy firmly in the mix for the second base job, logged a .330 BABIP in 2010. According to sabermetrics, Herrera was lucky, and Iannetta was extremely unlucky.

The problem with this stat? It assumes that all balls put into play are equal. It doesn't take into account the pitches that a player chooses to swing at, or any flaws in his swing. For instance, Iannetta, who struggles with a loopy swing, found himself consistently popping up. Instead of squaring balls up, he was constantly working underneath pitches, lifting them high in the air, easy for the defense to catch.

Herrera, on the other hand, is a hitter who has a plan at the plate. When he doesn't get the pitch that he wants, he simply fouls off the pitch and waits until he gets the one that he likes.

While Herrera is by no means the definition of a great Major League hitter, his .330 BABIP should not simply be excused as luck. Same goes for Iannetta, his low BABIP was not simply line drives finding gloves, it was pop flies waiting for a defender to camp underneath the ball.

Another beef I have with sabermetrics is their favoritism towards power. One of the statistics that has widely been accepted by even the purists is OPS. That statistic measures on base percentage, plus slugging percentage. Anyone with a .900 OPS or above is an amazing player, but average is around .750 or so. Anything lower than .700 is bad.

On Monday, respected Rockies blogger Andrew Fisher of Purple Row tweeted "I wonder how many Rockies fans know that Chris Iannetta had a better OPS in 2010 than Jonathan Herrera. Perception is reality."

While I respect Fisher and his writing, I strongly disagree with the correlation. OPS strongly favors a player who can hit for power. It is a great statistic, but when a power-hitting catcher and a singles hitting second baseman are put side-by-side, the stat favors the guy who can hit the ball out of the yard.

Obviously home runs are an important part of baseball, but suggesting that in Iannetta's worst season, he was still better than Herrera in his best season is not doing justice to either of the two players. Herrera has an instant obstacle in his way on the slugging percentage side because he isn't really a power hitter. But just because he doesn't trot around the bases often does that mean he's not as good of a hitter as Iannetta?

Take a step back and look at the roles that both players are supposed to fill. The Rockies don't need Herrera, presumably a two-hole hitter, to be hitting doubles, triples and home runs. He is supposed to be a table-setter, a guy who can get on base and be on the pitcher's mind as he deals with the middle of the lineup.

Iannetta, however, is expected to hit for a few more home runs than others. He is not a guy who is going to foul off a bunch of pitches and steal the occasional base. While he does walk quite a bit, he is going to have an advantage in the OPS category because of his power.

Of course sabermetrics help. They dive deeper than the traditional stats. However, when they are relied upon as gospel, the fan has simply traded one flawed statistic for another. They should be used to find out certain statistics, but not used as a complete replacement for the stats that have been relied upon for over a century.

Disagree? Tell me why.

This article is also featured on INDenverTimes.com

For more on the Rockies visit RockiesReview.com

7 comments:

  1. You hit the nail on the head. While sabermetrics reveal more statistically about the game than traditional statistics, relying on them as absolute fact is a mistake. Another case in point about the flaws of BABIP; Carlos Gonzalez. After he signed his contract every sabr site commented on his high BABIP and how unsustainable it was. What they don't take into account is the skill involved in hitting hard line drives. Cargo's bat stays in the zone longer than most players, allowing him to barrel the ball up more than most, producing line drives that are likely to find gaps/holes. To chalk that up to luck alone is specious thinking at best. Sabermetrics are good, but should'nt be the be all end all of evaluating players.

    ReplyDelete
  2. Good read. I agree, to an extent. I think all the statistics need to be used to come to a conclusion. You cannot just look at OBP or OPS or wOBA or even a players batting average. Put them all together and they start to paint a picture.

    And you are right, comparing a catcher to a second basemen generally isn't a good idea. Two different body types (and generally skill sets at the plate) make up those two positions. All to often in the past center fielders have been compared to first basemen who have been compared to shortstops. People are starting to realize that it just isnt the same.

    This is where WAR comes in handy because it takes into account the players position.

    But, again, it is not the end all be all argument winner.

    ReplyDelete
  3. Iannetta's BABIP wasn't because of Popups. He had a lower popup rate last year than his amazing 2008, and a lower popup rate than Tulo, Cargo Olivo, or Stewart. Also, wOBA corrects for the slugging discrepancy, yet Iannetta still did better in that regard.

    ReplyDelete
  4. Mr Anonymous just made my point.

    According to strictly SABR stats, I am supposed to believe that Iannetta would have had just as good of a season as Tulowitzki and Gonzalez, if only his luck was better.

    Sorry, I watched the games. That simply isn't anywhere close to the truth.

    ReplyDelete
  5. Travis-
    I agree, I actually like WAR. I still think that it has it's flaws, but it helps to really show the impact acquiring a certain player will have. I think that SABR fans quit relying on what they see, and rely solely on what some sort of statistic that really doesn't tell the whole story says.

    ReplyDelete
  6. Dan-
    Thanks for the comment. Like I said to Travis, I don't think that SABR stats are all bad, I just think that they are thought of as gospel. They should be taken the same way as traditional stats, as something that helps paint the picture, but doesn't make everything perfectly clear.

    ReplyDelete
  7. Many SABR people take a nuanced approach to these statistics and realize that no one stat is the be-all-end-all; they all have their limitations. If one isn't careful, though, you are certainly correct: it's easy to start throwing stats around inappropriately to make a fallacious argument.

    BABIP is affected by a number of factors, including how a batter puts balls in play. On average, 3/10 grounders go for hits, while 4/10 fly balls & 7/10 line drives do. One rule of thumb I've heard is that a good approximation for what a player's BABIP should be is .12 + line drive rate.

    Per Fangraphs, Herrera's LD% was 18.4%, which would lead to an expected BABIP of .304; Iannetta's LD% of 13.5% leads to an expected BABIP of .255. Despite the former getting lucky and the latter being unlucky, Herrera still comes out clearly on top.

    In 2008, Iannetta managed a LD% of 21.4%, which goes a long way towards explaining his success that year.

    More about BABIP:
    http://sports.yahoo.com/mlb/blog/big_league_stew/post/Everything-you-always-wanted-to-know-about-BABI?urn=mlb-203710

    As far as OPS goes, it's generally regarded as being an imperfect stat in that it does, indeed, considerably overvalue power. The problem is that slugging percentage would tell you that doubles are worth twice as much as singles, and home runs twice as much as doubles. This doesn't really square with reality.

    More about OPS:
    http://www.fangraphs.com/library/index.php/offense/ops/

    Presently, the favorite offensive stat of SABRs (SABRites? SABRheads?) is wOBA, which attempts to value walks, singles, doubles, triples, and home runs appropriately.

    wOBA:
    http://www.fangraphs.com/library/index.php/offense/woba/

    I'm pretty new to SABR stats, but it's been fun (and informative) just picking up a bit of a familiarity with them. For anyone interested in learning more, the Fangraphs library makes for a pretty good primer.

    ReplyDelete