Mike Fought the Law, and the Law Won

Mike Bordick had an April to remember in 2000. After nine years of being seen as a good-field/no-hit shortstop, Bordick exploded for an April AVG/OBP/SLG of .352/.365/.682, hitting seven homers in 88 at bats. Theories abounded about the newfound hitting prowess of Bordick. Was it off-season conditioning? A new batting stance? Was it due to being moved up in the lineup? Whatever the reason, it was clear that Bordick had found a new “level” of performance.

Or was it. Here are Bordick’s stats in 2000 from May until October, as well as his stats in 1999 and 1998:

2000 (post-April):               .273/.336/.400
1999:                            .277/.334/.403
1998:                            .260/.328/.411

So what happened. Did he start eating cheeseburgers and watching soap operas? Did he forget to use his new stance?

What probably happened was that people read far too much into far too little. In other words, Mike fell prey to what I call, Voros’ Law:

Any major league hitter can hit just about anything in 60 at bats.

Whether some people admit or not, most baseball fans like baseball statistics. Moreso than any other sport, baseball is heavily interconnected with its statistics, as even the most casual fans know the basics and what they mean. As such, after six months of the same old statistics never changing, when the new ones start compiling, most of us start to get a little giddy.

To wit, in the last few weeks I’ve heard people theorizing the death knell of Jim Thome’s career, the possibility of an A-Rod flop and David Justice’s rapid aging. We’ve seen a team in Tampa bench its starting second baseman, bench its starting third baseman, fire its manager and then bench the replacement for that third baseman, all before they were halfway through April.

You see at around sixty at bats, things like batting average start to look real. No one’s hitting .715 anymore so they must be okay now, right?

So is Voros’ Law true? If so, why? It, of course, isn’t literally true. Rey Ordonez isn’t about to bust out and hit 40 Homers in 60 at bats. It is more of a warning not to read too much into a handful of at bats, especially when we have perfectly good information based on much larger samples from previous years. Basically, the point is that if Mark Grudzielanek hits five homers in fifty at bats, one should assume the version of Grudzielanek you’re familiar with is capable of doing so without fundamentally changing as a player.

Why? Well there are a bunch of reasons, some involve statistical theory, but others involve logic. Here is a short but not all-inclusive list of reasons:

1. Random Chance – A lot of people understand this fairly well. Baseball statistics are such that the differences in players, even what we would normally consider large differences, aren’t easy to detect at small number of at bats. This isn’t to say that the results on the field are randomly determined, just that if you had a player with an infinite number of at bats at a certain level, there’s no reason to expect him to hit exactly at that level for smaller stretches of time. However, this often isn’t near enough to explain some April deviations in performance. For example, the little homer binge by Grudzielanek mentioned above wouldn’t normally happen by chance, except…
2. Multiple Endpoints – Ah yes, we’ve been down this road before. This is a clear example of how multiple endpoints can trip us up when we draw conclusions about the likelihood and explanations for an event, after it has already happened. Though it may be that Grudzielanek is unlikely to hit five homers in fifty at bats by chance, if it had happened to Deivi Cruz, Eric Young or Mike Lansing instead, we’d then be talking about one of them. You see it’s not just the chances of Grudzielanek having such a stretch of at bats, but instead the chances of just one player like Grudzielanek having such a stretch. Suddenly, the odds of it happening become, much much better.
3. Unrepresentative Sample – What this means is that in 60 at bats, the circumstances under which the stats were achieved are often (usually) much different than the average set of circumstances a hitter would normally face over a season or career. Each team has only played a few other teams, in only a few other parks, under April weather conditions, often not facing fifth starters. For example, after 64 at bats Ben Grieve had faced Pedro Martinez eight times, or in one-eighth of his at bats. If Grieve faces Martinez one-eighth of the time over an entire season, chances are he’d be a little unhappy. Unless you look really hard, it will be difficult to determine what kind of conditions each player achieved his stats under. Even if you could, because of the above, the results you’d normally get would probably not be worth it.

Now before anybody gets upset with me and calls me the Grinch Who Stole Shane Spencer, I’m not here to rain on anybody’s parade. Baseball statistics can be great fun, even 20 at bats worth. Do with these April statistics whatever will bring you the most pleasure. Far be it from me to deny anyone that. But understand that when people start drawing conclusions from these statistics, they really are seriously rushing to judgment. If 60 at bats were really that meaningful, Alex Rodriguez and Manny Ramirez would probably be in a different line of work, and Chris Stynes would be Ted Williams.

Just remember that using baseball statistics as a tool, means using the tool in appropriate situations, for appropriate jobs and at appropriate times.

Now, using statistics for fun is a whole separate story. Like any other tool, lots of fun can be had using it incorrectly. Who here can’t dig up a fond memory like chasing a friend or sibling around the house with their dads’ power saw? Boy, you should have seen the look on Half-Eared Pete’s face. Man, that takes me back…