Friday, July 20, 2007

What Do Baseball Stats and Temperature Scales Have in Common?

If you're like me, you have a hyperactive mind, one that makes unnatural analogies and connections. That may not make me a hit at parties, but it does provide me with a unique perspective on some things.

For example, the average student thinks about absolute zero and the Kelvin temperature scale long enough to pass a test, then files them away in the back of his mind. I took a different tack, asking why we don't measure everyday temperatures in Kelvin. The answer is fairly intuitive once you think about it: the Fahrenheit (and to a lesser extent, Celsius) scale makes it far easier to interpret a weather forecast.

100 degrees Fahrenheit is considered hot weather for most of us, and 0 degrees very cold. We understand this because 100 is a relatively high number and 0 a low one. Now imagine that instead of 100 and 0, the forecasts read 310 K and 255 K. Those are both fairly high numbers, and the difference between 310 and 255 sounds a lot less significant than the gap from 100 to 0. Day-to-day temperatures, when expressed in Kelvin, will all look closely grouped together, minimizing the appearance of change. Meanwhile, all the Kelvin temperatures between 0 and 200 are completely useless because we will never encounter them. All this makes the Kelvin scale a very poor choice for comparing weather conditions.

By now, you're surely asking what all this has to do with baseball. The answer is that we have many Kelvin scales in baseball stats, where zero (absolutely no production) is a useless baseline because no one ever approaches it. These include batting average, runs, and even OPS. We practically never encounter a batting average below .200 or an OPS below .600. Why confuse the issue by utilizing a scale with so much wasted space, and where most of the players are closely grouped together?

This is especially relevant because we already have a baseball equivalent of the Fahrenheit scale. It's called VORP, and it expresses each player's value in terms of how many runs he has contributed to his team. In measuring on-field production, runs are probably the easiest possible baseline to understand and interpret. Instead of needing a graphing calculator to make sense of Barry Bonds' 1.421 OPS or .362 batting average in 2004, we can say he was worth 132 runs. No more, no less.

Are sportswriters happy that they have this catch-all (except defense) metric at their disposal?
No, they bitch and moan about how much better the world was when they were younger, before technology came along and ruined everything. I suggest that for the next year, Murray Chass and his followers use the Kelvin scale when deciding how warmly to dress for a day at the ballpark, and see if it improves their experience.

1 comment:

Kevin said...

Great post. Maybe I'm just a freaky numbers geek, but the definition of VORP and the thought process behind it makes perfect sense to me. Then again, the first Bill James Baseball Abstract I purchased was the last year it was self-published (1981), so this stuff is in my blood.

I have to laugh at Chass's notion that understanding how baseball really works will somehow detract from fans' enjoyment of the game.