Get inside March Madness with SI.com's Luke Winn in the Tourney Blog, a daily journal of college basketball commentary, on-site reporting and reader-driven discussions.
3/13/2007 02:13:00 PM
Feel, Feel, Feel Their Heat
Aaron Brooks and the third-seeded Ducks could be peaking at the right time.
Jonathan Ferrey/Getty Images
In the NCAA selection committee's bunker, the "hotness" of tourney squads is usually measured by the results of their last 10 games. There's a entire, pull-out section for "Last 10" on the committee's team sheets -- you can view it on the UCLA example Seth Davisposted last week -- that appears below the breakdown of a team's wins and losses against different tiers of the RPI. Regular fans, I think, use more arbitrary methods to take a team's temperature, such as gauging conference tournament performance -- or even more unscientifically, how good the team looked the last time you caught them on TV.
The Blog, in its ongoing obsession with efficiency-based analysis, is here to provide you with a new -- and improved -- way to identify the hottest teams heading into the NCAA tournament. I've enlisted the services of esteemed amateur stat-head David Hess, whose full previews can be found this week on HackTheBracket.com. Hess has been graphing Ken Pomeroy's game-by-game efficiency numbers for NCAA tourney teams, adjusting the data for home/road/neutral location and level of competition, and then applying what he calls "trendlines" -- curves that allow us to see whether a team is becoming more or less efficient leading into the dance.
Now for the juicy information: I asked Hess to pinpoint the hottest team in each region, based on the trendlines, and he delivered exclusive graphs to the Blog that appear below along with my commentary. (Note that an optimal team graph would have the top curve going UP, representing increased offensive improvement, and the lower curve going DOWN, representing increased defensive stinginess.)
The Midwest Region's hottest team:
The Ducks are even hotter than Florida, according to Hess' figures. Their offense has steadily improved over the course of the season -- a very promising sign -- and is at its peak level following a run through the Pac-10 tourney. Defense is not Oregon's strong point, but it has dipped down to a sub-90.0 rating (meaning it's giving up fewer than 0.9 points per possession) for the first time in months.
The West's hottest team:
This one should come as no surprise. The Jayhawks were consistently one of the nation's best defensive teams for the entire season, but have blossomed on offense over the past month. The only worry -- even though the figures are adjusted for competition -- is whether or not the Jayhawks' offense peaked against the weakest part of the Big 12 schedule, and is in for a reality check in the NCAA tournament.
The East's hottest team:
North Carolina and Georgetown, watch out. I put Mr. Durant and the 'Horns in the Final Four of my SI.com Writer Bracket for a reason. The most important thing in this graph is not the quality of Texas' offense, but the fact that its defense -- which has always been its weakness -- is finally hitting a respectable, sub-90.0 level. Finally, the South's hottest team:
Ohio State has been a high-quality team for months, but the school whose efficiency margin is peaking here is Louisville. The Cards' offense, even though it bottomed out in its Big East tournament loss to Pitt, has been hitting incredible levels down the stretch. I put Texas A&M in my title game, but this information, needless to say, has me worried about the Aggies' probable second-round matchup with the 'Ville in Lexington.
Hmm... lets see here... according to the "experts" at SI and ESPN, TAMUCC is a "team of destiny", UNLV and GT are "underseeded/extremely athletic/sleepers", Oregon is the "most dangerous team/peaking/could give Florids a real go", and Florida "will be the champs/blow them off the court in a laugher". I think that covers it.
If one of you "experts" could please call Bo Ryan and all of the Badger fans and tell them to not show up in Chicago, that would be great... might as well save the gas money since they don't have a prayer of winning any games in the tourney... let along reaching the Final Four/Championship Game... especially since they are "slumping" now.
Geesh... you would think playing MSU and OSU five of the last six games would have accounted for *some* of that "slumping"... but I guess the "experts" still think the Bennetts are coaching the Badgers (although they are "sexy" now). I've heard "grind it out", "defensively minded", "athletically challenged" and other such quotes from the gurus... I guess pounding FSU (scoring 81 pts), beating a cinderella in Winthrop (82), crushing Pitt (89!), and averaging 70 points a game during the season (Ohio State was 73.6) in a conference that actually plays defense doesn't matter.
Also remember that Wisconsin gave NC the best game, by far, the year that they won the NCAA tourney... despite not "having a prayer" in that game either.
David Hess here. That link to HackTheBracket in the post isn't a bad link, it's just that we're having server issues. We should get those straightened out tonight, and the site will be up and running all through the tournament. So if you're interested, please check back late tonight or tomorrow.
Well, anonymous seems pretty incredulous about his Badgers; perhaps he/she should have left a name. Wisconsin was actually my favorite to win it all this year...until Brian Butch was ruled out for the year with a really unfortunate elbow injury. Alando Tucker is just as good as Durrant is (granted, he's 3 years older) and Wisco role-players such as Kamron Taylor are second-fiddle guys any coach would love to have. However, Butch gave them a weapon few teams can counter in a solid post player with significant size who can also stroke 3's and knock down free throws. Bo Ryan will keep his team around longer than people think but sadly that injury took Bucky out of the Final Four.
to the badger fan who posted first: not quite right on wisconsin giving unc the best game in '05. if my memory serves me correctly, unc beat nova in the sweet 16, 67-66, and needed a phantom traveling call on allan ray to get by jay wright's boys.
Yup... you are right daddydaycare... my response was probably over the top a bit... o.k. a lot. It is just frustrating with all of the experts not recognizing that the Big Ten does play tough defense, and that the Badgers are more than just Tucker. Losing Butch does hurt a lot... no question... but playing MSU and OSU five of the last six games had more to do with how the Badger's played than losing Butch.
Oden played a large part in that... he is an incredible player, and not having Butch for those games hurt.
I'm not going to say that UW does not get any respect... they have received a lot of love this year. I guess I just wanted to remind people that they are still a very good team and not to jump off the bandwagon just yet.
I must say I'm not all that impressed with the analysis at HackTheBracket. The reason is simple, the trend lines offer no adjustment for two key stats: whether games were at home or on the road and the quality of opposition.
Let's take a few examples:
Oregon - on a tear ... through 6 games at home or on neutral courts against opponents ranked outside the top 25 (except the excruciatingly inconsistent Arizona, at 23).
Texas - roaring ... except for two losses to Kansas, and a whopping 1-5 except top 20 teams. Yes, they're 8-2 in their last ten, but seven of those wins came against teams ranked #38 and lower.
Kansas - legitimately on a winning streak, but hasn't beaten a top 15 team since November.
Louisville - also legitimately on a winning streak (minus its tournament loss to Pitt), but has lost 5 of its last 6 to top 25 teams.
Georgia Tech - who they say is playing like a #4 or #6 seed (seeded #10), but who is 1-8 on the road, who played 6 of its last 9 at home (just when its play seemed to get better), and who lost 2 of the 3 on the road during that final sequence.
TAMU: Split 1-1 against top 10. Split 1-1 against Texas (#16). Next highest ranked opponent? #38 Oklahoma (to whom they lost in the conference tournament).
Florida - what were the NCAA crowd thinking? Split 1-1 with top 10. Beat Kentucky (#18) twice. Has not played anyone else in the top 25. Swept #45, #60, and #30 in the SEC tournament. Hardly an overwhelming performance. Wisconsin beat #13 and #24 and lost to #4. Ohio State beat #49, #22, and #7. I'd say those were a little more impressive.
UNC - Doing better than many above, but still has 4 losses since Feb. 1 and won its conference tournament against #37, #42, and #66.
A few more tidbits:
On Big 10 defense: Despite the lack of respect the conference has gotten this year, the Big 10 still has 6 teams in the top 25 on adjusted defensive efficiency. The ACC has 4 and the SEC has 3. It may be worth noting, vis-a-vis prior posts in this list, that Wisconsin's last 6 games have come against the nation's #13, #10, #13, #13, #3, and #10 defenses. They're 4-2 for the sequence. The overall rankings of these opponents (kenpom): #13, #4, #13, #13, #24, #4. No team played a more difficult last 6 games.
It does say that, which I missed, but in fact their method of adjustment (if you go look at their website) is quite problematic for looking at trend lines.
You have to be very careful reading "adjusted" data. The adjustments are not actual adjustments but rather national average adjustments. They are, in other words, what the average team playing in an average game would have experienced. They bear essentially no relationship to what any individual team experienced on any individual night.
So, for example, the average team plays 1.014 times better at home than on the road, according to Ken Pomeroy. But some teams do much better at home than that and much worse on the road. Two of the biggest offenders this year happen to be Georgia Tech and Michigan State. The selection committee recognized that and seeded GT at #10 and MSU at #9, even though their kenpom.com ratings are #14 and #13, respectively. In both cases, adjustments for home games significantly underestimate the advantage the team gets at home, while road adjustments underestimate how badly they play away from their home court.
Even more importantly, the adjustments as applied are meant to apply statistically over a whole season. This is how Ken Pomeroy uses them. He has an adjusted efficiency rating for offense and defense for the whole season. However, these adjustments are meaningless on a game by game basis, which is what the trendlines are based on. This is why Ken only provides unadjusted efficiencies on his Game Plan page. Of course, the adjusted efficiences can be used in a predictive sense, in trying to predict who will win a future match up. But they create new uncertainties when applied retroactively to adjust actual game data for intergame variation. On a given night, we simply have no way of knowing whether a team got an average boost from being at home, a bigger boost then normal, or a smaller one.
The trendlines themselves also add uncertainty, as they are fits to infinite length curves, but of course the data sets in question are finite. The result is a magnification of small changes at the end of the sequence.
So, in conclusion, the "adjusted" trendline data for Oregon may indeed indicate their offensive efficiency is going up. I note, however, that it has done so against very weak opponents in relatively friendly settings. Given the uncertainties associated with adjusting data on a game-by-game basis, I wouldn't want to count too much on the conclusion that Oregon was too hot myself. Especially when its best offensive performances in the closing stretch came against the nation's #34, #85, and #109 defenses. It won't meet too many teams who have defenses that bad in the tournament. Ditto my other comments.
Clark, that's a really good point about the home field advantage not being equal for each team. We've got a lot of ideas for things we can do in the offseason to make the site better - we put this whole thing together in about a week, so it's obviously not a finished product yet. One of the things we'll definitely take a look at is how to improve the adjustments. If you have any ideas, feel free to comment here, or email us at "hackthebracket ATSYMBOL gmail PERIOD com" (gotta avoid the spambots)
"Especially when its best offensive performances in the closing stretch came against the nation's #34, #85, and #109 defenses. It won't meet too many teams who have defenses that bad in the tournament."