After the first day of the round of 32, the Madometer is still measuring well above record levels of insanity. The Wichita State upset of Gonzaga pushed it over the edge–and further cemented Mark Few’s reputation as a snakebit tourney coach. All the big-time underachieving coaches didn’t fail to disappoint: Few, Dixon, Alford, Brey. You can add John Thompson III to the list of coaches with low PASE values. Meanwhile, two coaches whose PASE was soaring have come back down to earth somewhat. But Shaka Smart and Brad Stevens failed to meet expected win totals for their seed positions.
Even if every game goes to seed expectations today, the Madometer will still sit at 22.1% madness. Something tells me that perfect higher-seed dominance is not in the cards today. I suspect will add at least one upset to the already above-average total of nine.
Yesterday’s blood-letting drove the Madometer to new heights. The Minnesota upset-that-wasn’t-an-upset was foretold by practically every bracket model. There were signs that Cal could take UNLV, and even indications that LaSalle might upset Kansas State. But hardly any numbers liked the Ole Miss upset. And then there was the Florida Gulf Coast stunning of Georgetown–which reader Alex loudly and confidently called. He’s quit his day job and bought a ticket to Vegas.
When the first round ended late last night, there we seven upsets–and the Madometer was all the way up to 25.0%. I’m going to wait for the results of round two, but if it goes any higher, I’m going to have to redesign the darn thing. Heck, even if all the higher seeds win over the weekend, the Madometer will still show 17.9% unpredictability. And, amazingly, if things go chalk throughout the rest of the dance, it will only settle to 12.0% madness. That’s already crazier than eight of the past 28 dances. Of course, the madness will continue. Some of the higher seeds looked very shaky. Gonzaga, Kansas and Marquette seem like good candidates for early exits.
As bad as it all was, the upsets didn’t do serious damage to anyone’s bracket–unless you had Georgetown, New Mexico or Wisconsin going beyond the Sweet 16. I took a quick look at the bracket models last night–and they’re mostly in the 45-85 percentile range on ESPN. But the difference between good and bad at this point is about three picks. The real shakeout is about to happen. I’ll report on the state of the models later today.
Member Mark S. pointed out that I had the wrong PASE for coach Tad Boyle. I had him at -.170, but that’s former St. Joseph’s coach Jim Boyle’s PASE. Tad’s is +.455. It wouldn’t have affected any of the brackets leveraging coaching performance, because he’s going up against John Groce, who owns a +1.290 PASE. But for the sake of accuracy, make the change in your Excel sheet.
The Harvard upset of New Mexico turned what was going to be a fairly tame first day into record-pace madness. In the Bad Coaches Region, a couple of all-time underachievers did not disappoint–by reliably disappointing. Jamie Dixon can’t drop much further on the coach ratings list (check out Ratings+; he’s third from the bottom), but his PASE will get slightly worse. And then there’s Steve Alford. I knew New Mexico was a fraud (reserving judgment on the whole MWC), but the first round? To Harvard? Yikes.
One chic Cinderella pick fell a basket short on a hellavudellavadova comeback. And another one duck-walked to victory over a rattled-looking Oklahoma State. No doubt today will bring more surprises. We’ll know tonight if the first was an aberration, of we truly are on the path to record madness.
Long-time reader Vern L. found an error in the Costanza bracket. I picked Miami to beat Indiana in the Elite Eight…but Indiana to get to the championship. How Costanzish of me. It’s supposed to be Miami in the finals. (Not that there’s anything wrong with Indiana…or those striped pants.)
In the Outcome Matching model, I’ve got Boise State beating Kansas State, because they were among the top five closest upset match-ups in KenPom efficiency differential. Boise State looks like they’re about to lose to La Salle. So what should you do with Outcome Matching if La Salle beats Boise State?
As luck would have it, La Salle’s differential with Kansas State is the next closest among the upset match-ups. So…just replace La Salle as the Cinderella, defeating the Wildcats. (I hope this doesn’t make the K-State fans in the forum too nervous. Stick to your guns.)
I’ve fielded enough emails asking for a guide to the columns in the stats sheet, that I figured I ought to get after that. A couple years back, I made an effort to put a little triangle note beside each column header in the sheet. But then I started grabbing more and more data, and I got lax. Sorry for that. Here is a glossary of the less obvious columns:
- * – Number of All-Americans, as selected by USBWA
- C Yrs – Coach’s tourney trips
- C E8 – Coach’s number of Elite Eight runs
- PASE – Coach’s performance against seed expectations
- CA – Conference affiliation, B=Big Six, M=Mid-major, S=Small
- T Yrs – Number of consecutive tourney trips for the school
- 10 – Wins in last 10 pre-tourney games, includes conference tourneys
- +/- p – Pre-tourney winning or losing streak
- EFG% – Combines two- and three-point percentage
- 3/F – Percentage of threes attempted to overall field goal attemps
- OREB % – Percentage of offensive rebounds a team gets to overall chances
- Opp OREB % – Percentage of offensive rebounds a team allows to overall chances
- Possess – Overall number of possessions, used to calculate other fields
- TO PCT – Percentage of times a team turns the ball over per possession
- Ast/ Turn – Assists-to-turnovers, a measure of ball control
- Ast/FG – Assists-to-field goals, a measure of how much a team needs passing to score
- 5PT – Number of points by top five scorers (not necessarily starters)
- 5PT % – Percentage of overall points from top five scorers
- GPT– Number of points from guards (not necessarily starters)
- Guard% – Percentage of overall points from guards
- AGE – Average age of top five scorers
- Off Eff – KenPom offensive efficiency, points scored per 100 possessions
- Off Rank– KenPom offensive efficiency ranking against 347 D-I schools
- Def Eff – KenPom defensive efficiency, points allowed per 100 possessions
- Def Rank – KenPom defensive efficiency ranking against 347 D-I schools
- Pythag – KenPom overall measure of efficiency
- Pyth Rank – Ranking by KenPom overall measure of efficiency
- COACH FACTOR – Something I was fiddling with to normalize coaching impact to KenPom. Ignore.
- PYTHAG + COACH – See above.
If you scroll all the way over to around column CV, you’ll find my calculations of the Pulse Check disqualifiers. Those should be obvious. If you want to see the values, click on the formula. Hope this helps–or at leasts
If you’re new to Bracket Science, you may not know about the Madometer. It’s a way of measuring the relative unpredictability of the tournament.
The Madometer works by measuring the seed-position differences between actual winners and perfect high-seed success or failure in all six rounds of the dance. If the higher seed advanced in all 63 games of the tourney (perfect sanity), the cumulative seed value of the winners would be 203. If the lower seed always advanced (sheer madness), their cumulative seed value would be 868. The difference between the two—665—is the predictability range. If you add up all the seed positions of the teams advancing in the dance, then subtract 203 and divide by 665, you arrive at the percentage by which the tourney deviates from perfect higher-seed dominance.
Got all that? If not, don’t worry. I’ll calculate the Madometer value for you after each night of the tourney. Then I’ll report the reading on this handy gauge that displays the range between “complete Wooden sanity” and “utter Chitwoodin’ madness”:
If you want to know more about the Madometer, and see its readings for the last 28 years, check out this blog post: http://wp.bracketscience.com/?p=186
I fully expect this year’s dance to contend with 2011′s record of 19.8% unpredictability. We shall see…
I know what I said: only five models this season, tops. And now…well…there are 15. I’m out of control! As I mentioned in an earlier post, part of the reason I was able to deliver so many this year is that member Ryan Tressler took the Sunday night/Monday morning pressure off of me and filled out Seed Match-ups and Upset/Toss-up models. I also devised a much easier way to deliver the models.
So…we now have four more–under the Tips+ section and named “Bonus Models.” I went to Nate Silver’s 538 blog and filled out a bracket using his probabilities. I also have another bracket dialing up the impact of coaching on tourney outcomes. Then there’s a “Costanza” bracket, filled out under the assumption that, “if every instinct I have is wrong, then the opposite must be right.” Finally, I’ve submitted my Keeper bracket. Fair warning: I played it pretty close to the vest this year, aiming to win a smallish pool. And you should also know: I haven’t won a pool since 2005.
Just did a quick tally of the champions picked in the 15 brackets. Here’s what we got:
- Indiana – 5
- Louisville – 4
- Florida and Duke – 2
- Kansas and Gonzaga – 1
So, all the one seeds are represented, one two seed (Duke) and one three seed (Florida). If I had to pick a team that wasn’t on this list, I’d go with Ohio State. Anyone think differently?
This is a housekeeping blog post:
- I’ll be gone tonight from 6:30 to about 10:00. I’m doing a special “Bracket Science Night” at the Wolverine Brew Pub in Ann Arbor.
- Some people are having trouble finding the models. They’re under the Tips+ section, called “2013 Models.” The link breaks across two lines.
- If you’re having any technical difficulties, email firstname.lastname@example.org
- The bracket forums are blowing up. Dive in! When I’m back, I’ll try to keep up, but there are so many comments.
- I’ll be in all day tomorrow–and I’ve got a few more models coming.
That is all.