Sunday, February 18, 2007

Words to Live By

Ken Pomeroy runs the most valuable resource on college basketball out there not the project of a major media outlet. Though his is site is focused exclusively on the men's college hardwood, he recently posted a list of core beliefs and philosophies, some of which are non-basketball-specific and important enough I'll list them here, and provide some of my own commentary.

1. Don't confuse achievement with performance.
The best team doesn't always win the game. This should be obvious, and people normally recognize it, but it's always so. Sometimes a better team plays a bad game. Sometimes a worse team plays a really good game. Sometimes fluky things happen.

2. One game matters little in evaluating performance.
See above. This is a big problem with Football Outsiders' rankings, particularly early in the season, as FO founder Aaron Schatz freely acknowledges. It's a little different in basketball, where you'll end up with 35 or so games from the best teams to analyze, than the NFL's 16 games. But, don't place too much reliance on any single game, or it could come back and bite you.

3. Stats don't lie.
Corollary: stats must be looked at in context. One of the keys to KenPom's stats is that he adjusts for possessions in a team's games, which can vary widely. NFL corollary: good teams face lots of pass attempts, so give up more passing yards and fewer rush yards. Looking at the NFL's yardage totals, good teams have a better pass defense and worse rush defense than you'd think, while bad teams have a worse pass offense and worse run defense, generally speaking.

4. Absolute statements are stupid.
Predictions are hard, especially about the future. Sometimes you get that one bad game. Sometimes things that aren't apparent will become obvious and important. Sometimes somebody comes up with something you haven't seen before, and the challenge is to adjust.

5. The single most important factor in determining the outcome of close games is chaos.
The FO-jargon for this is non-repeatable events. They happen. They can determine the outcome of games. This is why teams that win games by blowing out teams out are better than teams that win close games. See Guts and Stomps.

6. Statistical and visual analysis go together.
I'm a stathead. I believe in the power of statistics. Better stats are the key to better understanding. UFR, particularly as performed by me, is almost the exact opposite of statistics-it's totally my subjective evaluation of what happened. But I think I have a better understanding of the Titans because of it. Sometimes stats reinforce my subjective observations, sometimes vice versa, sometimes they lead you to opposite conclusions. But I'm better off, and believe (with some reason) I'm more knowledgeable because I have both.

7. The ultimate goal should be to find the truth.
This gets to what I was just saying. On a continuum, it looks like this.
Statistical analysis ------------------------------------------------- Truth
Visual analysis ------------------------------------------------------ Truth
----------------------Statistical and visual analysis----------------- Truth
But Ken's words on this are wise: "And if somebody says they know it, that’s when you should get suspicious." Know more today than you did yesterday, and know more tomorrow than you did today. The destination is just as likely to arrive as Godot, which is why the journey is the key.

No comments: