Wednesday, April 18, 2007

New Rankings

Both the USRowing Collegiate Poll and the cMax rankings are out. First, let's look at the coaches' poll.

Wisconsin and Princeton are tied for number one. This sounds about right to me. Both crews have been impressive - Princeton in a series of races and Wisconsin in one major regatta. After Princeton races Georgetown this weekend we should have a little better idea of their relative strengths. Stanford jumps way up to number three while Georgetown drops one place to fourth. If you look at the points, though, you'll see that these two crews are awfully close. I agree with this as well. Stanford in three is really a bit of a reward for coming out so strongly after a mediocre season last year. Georgetown didn't really lose the third slot as much as Stanford took it. They'll race in Indiana in a week and a half so shed no tears for the Hoyas now. Radcliffe and UCF look about right at 5 and 6 also. The thing about Radcliffe, though, is that they won't sit quietly at five. They have a few tricks up their sleeve and I still think the Princeton dual will be a race to watch. Radcliffe will be at home on a familiar course, and it could turn into a Princeton nightmare.

After a not-so-great week, MIT moves up a spot, largely at Bucknell's expense. Bucknell comes in at eight, which is interesting because if Bucknell puts together the light eight it had last year (they lost two), I don't think any voter thinks they'll be eighth at season's end. I think this ranking reflects some voters who felt they couldn't be ranked because they haven't raced enough, and some voters who put them higher in the list. My feeling is obvious - if they don't race, they can't be ranked. Next come Buffalo, Dayton, and Lehigh. Buffalo had a good Knecht, but Dayton and Lehigh finished behind Duquesne. Why are the Dukes way down at 15?

So, some quirks, but not a bad ranking. Again, I think it's important to note that 15 legitimate (we'll give Bucknell the benefit of the doubt) lightweight eights are ranked, with more racing. It's certainly a more robust category than the DII heavyweights, and it is showing strength.

Now let's turn to cMax. One reader already expressed frustration at the latest cMax release. A couple of points to remember - only crews that have raced two other ranked crews can be included, and the rankings improve with more qualifying races.

So, the first question is, How can Wisconsin be ranked 14 seconds faster than Princeton? Followed by How can Wisconsin be 17 seconds ahead of Georgetown when they only beat them by 11 seconds? Let's take the second question first. Now, I didn't create this system and I'm not an expert, but I like it and Chris Maxwell has explained some of it to me now and then, so let me give it a shot. I think about these time spreads as averages. In other words, given the races that have occurred to date, on an average day, we'd expect Georgetown to be 17 seconds behind Wisconsin. We rarely see and average day, however. If the two crews raced again with the same 11 second spread, no doubt cMax would be much closer to 11 seconds as well. It's still a bit early, and in Wisco's case it's very early, so take these with a grain of salt. That's not to say they're wrong, just that they'll improve with more races to feed into the black box. (Read a bit more about that black box here.)

What about Princeton? Well, I don't know the answer to that, but the formula apparently liked the quality of Wisconsin's few wins over the quality of Princeton's several. It's hard to argue against that, isn't it? Princeton beat Stanford by three seconds, a crew that, until that point, had only raced some West Coast crews that don't normally figure into the national rankings. If you're an objective observer, which the cMax engine is, everything's still a bit of a mystery. We're asking questions based on biases we hold from seasons past, but the fact is, overall, the quality of Wisconsin's victories is higher than that of Princeton's.

Stanford, Radcliffe, and UCF seem to fit, but OSU looks like an aberration to me. OSU has really only raced heavyweights at this point (qualifying races), so their ranking is bound to be squirrelly. Some amazement was also expressed at Wisco ranking 13 overall. Again, we're fairly early in the season so don't get too hung up on this. One thing that plays into my biases, however, is Wisconsin's estimate of 0.7 seconds faster than Bucknell's heavies. I've said that the IRA champ should beat the Dad Vail heavyweight champ, and this suggests that's true. Again, let's not get too hung up on this so early, but I see nothing here to lessen my faith in cMax. Remember, one of the reasons we use computer rankings is to try to take emotion out of the process. So when your emotions tell you they're wrong, you can't be surprised. Last year, before Sprints when Princeton looked like the fastest crew, cMax told us it was Wisconsin. cMax was right.


Anonymous said...

So you're saying you think Wisco is really that fast? We'll see what their own openweights have to say about it this weekend.

JW Burk said...

Well, I was actually a bit more interested in the relative lightweight rankings. I have to admit that, like you, I'm skeptical that the lights can beat their own heavies, but if Bucknell is really that much faster than the Wisco heavies, then yes, I think the lights can win. Really, I just think that there are still some aberrations here because we're still early in the season. I tend to think the relative lightweight rankings are more reliable than the overall rankings right now.

Anonymous said...

Wisco heavies are not fast right now. My money would be on the something to prove light 8.