Groundballs, Closers and the Periodic Table (02/16/01)
Goundballs and Closers
At last year's Baseball HQ Baseball Symposium in Arizona, Doug Dennis of Baseball HQ wondered aloud why it was so hard to predict closers and why so many supposedly bad ones, at least sabermetrically speaking, were able to keep their ERAs artificially low even while allowing so many baserunners. I don't know if he was posing a rhetorical question or actually asking for answers, but I forwarded my theory regarding closers (060200 - Changing of the Guard) as a solution to his dilemma. There were some murmurs, but very little else was said on the subject. No big deal - I'm sorta used to people looking at me like I was crazy.
However, I guess the notion struck a chord in him because 3 months later, he posted a fairly scientific examination of my theory on the Baseball HQ website. In it, he profiled the top ground ball relievers in baseball last year to see if there was a relationship. True to the theory, 86% of them posted lower than expected ERAs. In addition, but they posted much higher than average strand rates, a skill requisite in all closers. Of all the pitchers who pitched in the majors last year (597 to be exact), only 24% were able to strand baserunners better than three quarters of the time. However, 54% of the groundball relievers in Dennis' study exceeded the 3/4 mark. One factor in their success is that they were able to keep the ball in the park better than most, with 88% of them posting HR/9 inning ratios lower than 1, which is a fairly low standard given the offensive era we are currently in. Of course, this statistical bubble does stand to reason as these guys are inducing mostly groundballs. Low ERAs, high strand rates, few home runs allowed - these are exactly the things managers look for in closers.
Being a scientific study, Dennis was oligated to caution that the small sample size - it was, after all, only one year - was not substantial enough to draw definitive conclusions. However, there is no denying that there is less left to randomness here than many of the roto-experts originally thought. There does appear to be a statistical relationship between being an effective closer in a home run era and being a groundball pitcher. Whether it is a causal relationship has yet to be determined to everyone's satisfaction. My satisfaction has come in knowing that one of my theories passed the first test of peer review with flying colors.
The Periodic Table
Spread sheet programs, more than any other development, have increased the amount of hard statistical research in the past 10 years by an incalculable amount. More people are spending their free time copying and pasting the numbers into these programs, and manipulating the number to their heart's content. From this, a number of relavent theories uncovering the mysteries of baseball have been produced, each helping us understand better what is happening on the field.
However, I'm beginning to think that we've exhausted the possibilities with the basic building blocks we're currently using. There are only so many ways one can manipulate hits, walks and outs. I believe we have come to a point where we need to break things down further. For instance, not all hits are the same. Yes, there are singles, doubles, triples and homers but even they are not equal within their ranks. 'Bloops', 'bleeders', 'seeing-eye' hits, 'line-drives', 'dying quails', 'ropes'... any of these can turn into any type of hit. Likewise not all outs are the same. Was it a strikeout looking or swinging? Was it a line-drive at someone or an easy slow roller? And depending on who's umpiring and the shape of "his" strikezone that day, even walks can be broken down into categories like "just missed" and "wounded mascot".
How much information can be derived from further qualifying these events? Well, probably a lot more than can be imagined. For instance, pitchers benefit greatly from a good defense behind them, perhaps even to the point of disguising their own real ability. Likewise, defenders probably look better than they are if the pitcher is inducing easy chances. Qualifying what kind of contact the hitters make, the locations of pitches, and the path of balls in the field might prove to be important data when analyzing players' talents and skills.
The technology is certainly out there to do this kind of monitoring. Computers can be programmed to analyze video data and catalogue every pitch. Gaming companies use scanning and video mapping to make nearly life-like video games. Why can't similar technology be applied to recording and analyzing actual games. Every professional game is video-recorded. Wouldn't it make sense for major league organizations to spend a couple million a year on getting a better understanding of what they are paying for? After all, they spend more than that on utility infielders and signing bonuses for high-school players who might not ever make the big leagues. At least this kind of spending has a very good chance of being worth it.
It took more than 4000 years of human history before the first scientific studies showed that the material universe was made up of a little more than a hundred different atomic elements. Another four hundred years later we realized those elements could be broken down into smaller particles (protons, neutrons, electrons). It wasn't until the 20th century that it was discovered that not only could those smaller particles be broken down - quarks are thought to be the building blocks of electrons - but that there are even smaller particles (muons, gluons, bosons) that comprise the building blocks of the universe. Entering the 21st century, scientists are relatively confident that even these tiniest of the tiny are not the final answer as to what makes up the fabric of our existence.
Baseball research is still at the stage of using those basic elements. Hits, walks, outs. Those are the carbon, hydrogen and oxygen of baseball. Just as we know that the universe is made up of more than the basic elements, so also should baseball researchers acknowledge that their universe is far more complex than the sabremetric periodic table. Isn't it time we move into the 21st century?