Boyd's World-> Breadcrumbs Back to Omaha-> Interesting Rest Areas of Southern Kentucky for $400, Alex

About the author, Boyd Nation

Interesting Rest Areas of Southern Kentucky for $400, Alex

I was working on a long-term project I've got going earlier this week and ran into a minor partial result that really interests me. I also needed a short topic to balance some general notes stuff that's been building up, so stick with me through a little statistical tapdancing here at the top and see if I'm right about this being interesting, then we'll get to the news and notes.

The project as a whole involves trying to get some measure of prediction as to how a team will do in the postseason. While a lot of folks, me included, try to do this with the ISR's, that's not really what they're for, and I won't ever optimize them to predict the postseason at the expense of how they measure the quality of the season. For the potential independent measure I'm trying to develop, though, I'm looking at things like how "hot" a team is through various types of sequences of wins and losses, distribution throughout the season of opponent quality, and that sort of thing.

Anyway, one thing that I've found is the following little tidbit. Let's make up two measures of run scoring for each game for each team. Just to make them easier to follow, we'll call them Offensive Quality (OQ) and Defensive Quality (DQ). All they are is the ratio of how the team did against how most teams do against their opponents:

               Runs Scored
OQ = --------------------------------
     Average Runs Allowed by Opponent

              Runs Allowed
DQ = -------------------------------
     Average Runs Scored by Opponent

There are all sort of adjustments you could make to that based on park factors and the like, and I may do some of that going forward, but you get some interesting stuff from just taking the average of those scores for all of a team's games (call 'em OQA and DQA, just for simplicity), or at least interesting in the sense that the result defies traditional wisdom, which is that strong pitching is a requirement to advance in the postseason.

The best measure I've been able to come up with so far for postseason success is number of tournament wins. This isn't perfect (three wins may or may not mean a regional win, for example, which is not what most folks think of as equal success), but most alternatives are too subjective. Since we need a measure to be able to correlate with, we'll go with this.

Now, if you correlate OQA and DQA as I was talking about earlier with the number of postseason wins, you get a measure of how much each one -- offense and defense -- gives you a sense of predictability. I imagined going in that the correlation for DQA would be much higher -- after all, the cliche is that hitting pulls in the fans, but pitching wins championships. It turns out, though, that the two are almost equal -- .48 for DQA and .46 for OQA. In other words (and in conclusion) there's not an advantage to putting more of your resources on one side or the other; just recruit and play the best players you can regardless of which side of the equation they're on.

• It's kind of gotten lost with all the attention on the Olympics, but the U. S. National Team for this summer captured the gold medal at the FISU World University Baseball Championships in Taiwan. This group of college All-Stars had no real connection with the group that failed to qualify for the Olympics, and their accomplishments should be noted.

• I've gotten a couple of copies of the official RPI report for this year (thanks to those who forwarded them), and I've been able to verify that the formula I'm using is still correct. There are two sources for discrepancies, and they're worth a quick discussion. The first is that the NCAA is doing some intermediate rounding in their calculations. There's no real reason to consider what they're doing or what I'm doing right or wrong (other than that there's is what counts), but it's worth noting that that tiny change (something like .0004, if I remember right) was enough to switch Arizona State and LSU in the final rankings this year. That should be remembered whenever you're looking at teams next to each other who are very close in the actual RPI number. The other discrepancies come in because the question of what is a home game and what is a neutral-site game is a quite fuzzy one at times -- some early-season tournament games are home and some are neutral, and, as far as I can tell, it's up to the teams which are which.

• Speaking of the RPI, long ago I promised an explanation of exactly what was wrong with the RPI mathematically. I never got around to it (darn archives), and frankly there are enough different things wrong that no one explanation will ever quite suffice, but Paul Kislanko has written a very good discussion of the math behind the problems; head on over and take a look, and see what else is on the site while you're there.

Boyd's World-> Breadcrumbs Back to Omaha-> Interesting Rest Areas of Southern Kentucky for $400, Alex

About the author, Boyd Nation