Scientific proof that preseason predictions do mean something
Posted: Thursday December 11, 2003 1:25PM; Updated: Thursday December 11, 2003 4:42PM
Proudly presenting ... the 'Bag's first scientific paper! (Don't worry, it's actually pretty brilliant -- a word we don't often associate with the 'Bag.)
EVEN A BLIND HOG FINDS AN ACORN NOW AND AGAIN: A STATISTICAL ANALYSIS OF SPORTS ILLUSTRATED'S PRESEASON COLLEGE BASKETBALL PREDICTIONS
Trip J. Meine MD, Svati H. Shah MD and Manesh R. Patel MD Cardiology Fellows Duke University Medical Center
We here in the cardiovascular division of the Duke University Clinical Research Institute enjoy very little more than performing complex medical research. We do, however, enjoy taking off time from our busy schedules to peruse Sports Illustrated on a weekly basis, especially the annual College Basketball Preview issue.
And that got us thinking. Aren't we, at one of the top research (and basketball) institutions in the nation, uniquely qualified to evaluate SI's forecasting ability? So we pulled our statisticians away from their drudgery evaluating life-saving medications to have them crunch the really important numbers.
We took SI's preseason predictions for the last 10 years and compared them to the end-of-the-year consensus polls. We employed lots of nifty statistical models and methods, most of which we didn't even understand. To analyze the data, we considered using the statistical Operation of Kendall Tau, because we think he was a gritty Loyola Marymount point guard in the late '80s, but instead we settled on Spearman's Correlation Coefficient, mainly since it sounded pretty cool.
Here's what we learned: The average end-of-year rank for SI's preseason No. 1's was 8.5. The end-of-season No. 1's, on average, began the season as No. 10.5 in the SI preseason rankings. Of SI's preseason Top 10 teams (100 total over the 10-year span), 46 -- almost half -- finished outside the Top 10 and 22 finished out of the Top 25 at year's end. In fact, 11 of SI's 100 teams finished the season receiving no Top 25 votes at all.
Oh, you've had your Nostradamus years (in 1994-95 your preseason No. 1 and No. 2 finished in that order), but you've also had a few Dionne Warwick-Psychic Friends Network years. (Your 1996-97 preseason top three, Cincinnati, Kansas and Wake Forest, found themselves Nos. 16, 5 and 9 at the end of March Madness.)
Using the Spearman Correlation Coefficient, in which a value of 1 is perfect prognostication and a value of -1 is the equivalent of letting your little sister pick the winners by the color of their uniforms, you've done passably well. You had a statistically significant Spearman Coefficient of 0.55 for all 10 years combined, which elevates you just above Jimmy the Greek status.
So, strictly speaking, there is only a loose correlation between your picks and the end-of-the-year position of college basketball teams. However, we are sure your staff will be happy to know that an informal poll at our offices turned up the opinion that during these last 10 years you have never, ever screwed up the Swimsuit issue..
By the Numbers
SI's Preseason No.
End of Year No. 1 (SI Preseason Rank in Parentheses)
Michigan State (3)
Major props to the "Lethal Weapon 3" medical team down in Durham for their analysis.
"Just above Jimmy the Greek status?" Is that a compliment? (Wait, don't answer that.) Then again, SI now has the leeway to blare this coverline on next year's preview: SI'S STATISTICALLY SIGNIFICANT TOP 20 (SCIENTIFICALLY PROVEN TO BE MORE ACCURATE THAN PICKING AT RANDOM!) That'll sell a few more copies for sure.
I'd be curious to see how SI stacks up against other publications over the same period -- and how our college hoops guys stack up against SI's prognosticators in other sports. (Any takers to run a Spearman Coefficient on Dr. Z's NFL picks?)
It confirms last week's argument that preseason rankings do mean something -- and that Syracuse's run from preseason obscurity to the national title is in fact a historical aberration. Only twice in the last 10 years has a team ranked below No. 7 in SI's preview won the national title (the other was Arizona in 1996-97). And you wonder why my Magic Eight (a January prediction, remember) is usually a slam dunk? Let's say it together, kids: Who will be "this year's Syracuse"? Likely answer: Nobody.
The mastermind behind the golden years of 1994-95 (during which SI tabbed both the national champ, UCLA, and runner-up, Arkansas) and 2000-01 (with the preseason No. 1, Arizona, and No. 2, Duke, meeting the final) was Greg Kelly, SI's college hoops editor emeritus -- and the only non-beer-drinking Irishman I've ever met.
The 'Bag knows the identity of the SI editor who pushed for Auburn as our preseason No. 1 four years ago (yikes!) but will withhold it for obvious reasons. No lie: he was the same guy who lobbied for Oregon State as our preseason No. 1 in football a couple years ago. (Though the 'Bag, being a standup guy, will admit that he was the flake most responsible for SI's disgraceful NCAA tournament picks last spring. Yet another aberration! Check my previous years' picks and you'll see.)
Got your own take on the Dookies' stats? Submit your thoughts via the Mailbag form below.
Opening the 'Bag ...
Vanderbilt's Matt Freije has already proven that he's capable of performing. Is this something he can keep up all year and does Vanderbilt stand a chance of being one of the 64 teams in the tournament in March?-- Brad Cellio, Atlanta
As any Vandy fan knows, the Commodores have flatlined in February the last four years. But I really think changes may be afoot in Nashville. Freije (20.2 ppg) is playing like the SEC player of the year some outlets predicted of him, and Kevin Stallings' 6-0 team has already had two nice wins over Indiana and Michigan. Granted, those came at home in that freaky benches-on-the-baselines Memorial Gym, but they're definite head-turners for a team that went 11-18 last year (and a miserable 3-13 in the SEC). We won't find out much more about Vandy until they meet Auburn on Jan. 7 -- its schedule between now and then is cake -- but I've got a good vibe about this team. If the Commodores can stay near .500 in the SEC (and battle with Tennessee for the No. 3 spot in the Eastern Division behind Kentucky and Florida) they'll be an NCAA team. One man's prediction: If Freije stays healthy, he'll be a first-team All-America.
(Random question: Have Lionel Richie's Commodores ever played on the Commodores' campus? Is there any way we can rectify this? And what's up with Lionel's Paris Hilton-sidekick of a daughter anyway?)
Now for a couple of questions on scheduling:
Which ranked teams have the hardest non-conference schedules?--Richard Travassos, Spokane, Wash.
To answer your question I re-ranked the most recent AP Top 25 from hardest to easiest non-conference schedules. Keep in mind that I didn't take into account any future matchups that are contingent upon results in, say, a four-game holiday tournament.
Here are the results (with the number of current Top 25 teams on each schedule in parentheses):
Tom Izzo is certifiably insane. But if you look closely, three of Michigan State's five back-breakers (Duke, Oklahoma and Kentucky) will end up having Spartan-friendly crowds. Izzo probably expected to win at least two of those three games, not to get whitewashed and lose all three (which is what will happen if Kentucky beats MSU at Detroit's Ford Field on Saturday).
Pitt puts together a cowardly non-conference schedule every year. That certainly hasn't ruined the program, but it does put a damper on the excitement for unfortunate Panthers fans, who deserve better.
Learning from past tournament-seeding nightmares, Gonzaga has done a better job than any other mid-major of pumping up its non-conference schedule. If the Zags can beat Missouri in Seattle this Saturday and Stanford at the Pete Newell Classic next week (a less likely proposition, to be sure) then their 3-1 mark against those four teams will have been well worth the risk.
Oddly, none of Duke's three difficult non-conference games will take place in Cameron Indoor Stadium.
After a nice road win at Ohio State to validate the UConn upset, Georgia Tech could easily take a spotless record into 2004.
Michigan State has one of the toughest schedules in the country. Do you believe teams are better prepared for the long season by scheduling such tough non-conference opponents or are they better off with a lighter schedule to handle the long season?--David Walby, Troy, Mich.
Scheduling depends on a couple factors, but I think the key is to find a balance. Obviously, in college basketball losing a game early in the season won't prevent you from winning the national title, so generally I think it's smart to get some heavyweights on your slate. The one thing you want to avoid, as Izzo understands, is for your team to completely lose its confidence after tough losses and have that lead to a downward spiral from which you can't escape.
Do you think that Coach K will eventually stress a slower inside offense as compared to previous years at Duke? It seems to me that Shavlik Randolph and Shelden Williams could have monster seasons along with newcomer Luol Deng on the interior. The Blue Devils' outside shooting has been less than stellar to date. Chris Duhon and Daniel Ewing seem better equipped to slash and dish than to spot up with J.J. Redick anyway.-- Kevin, Big Spring, Texas
Good question, Kevin. As I wrote in a March 2002 article for SI, one of the main reasons Duke has the best spacing of any team in the country is the way the Blue Devils have used the three-pointer to set up their inside game (much like an NFL team that uses the pass to set up the run). Duke traditionally has employed the trey in all manner of situations, especially in transition, after offensive rebounds and off kick-outs. That, in turn, draws defenders away from the basket, making it difficult for them to help as Duke's perimeter players penetrate with the ball and cut to the basket without it. As Devils assistant Chris Collins told us, the three-pointer has caused teams to start running their offenses out to 22 feet (as opposed to the pre-trey 15- to 17-foot range).
That's why your question is important. Duke is always at or near the top of the country in three-point attempts, but Redick and Duhon are both hitting under 30 percent this year from beyond the arc. (That should change for Redick, but I doubt it will for Duhon.) And though Deng and Ewing are above 40 percent so far, I do wonder if defenses will start daring the Dookies to prove they can hit the trey -- particularly when Randolph and Williams appear to be the Devils' most effective one-two inside presence in several years.
My take: Coach K is a master at making the fine adjustments, but I don't think he'll blow up an attack that has served him well for years. In other words, better hit your threes, guys.
Why do all of the Duke and UNC alumni seem to run things at the big sports magazines? They seem to constantly downplay Kentucky. And how can no player from UK make the preseason All-American lists? Chuck Hayes was the best player on the U.S. team at the Pan American Games this summer and Gerald Fitch is averaging over 20 points a game thus far. Why do you guys love to pass over UK in all of your popularity contests?--Dave Kaelin, Bagram, Afghanistan (via Louisville, Ky.)
First off, Dave, if you're doing what I suspect you're doing in Afghanistan (that would be serving our country, not running an opium cartel) then thank you for your efforts. (Do you have a way of seeing games over there? Somehow I get tickled thinking there are folks in Afghanistan right now screaming for their Cats.) To answer your first question, you might be surprised at how few Tobacco Road folks there are at SI, and at how little an impact writers' and editors' own college backgrounds have on their coverage. (I'd be lying if I said it never happened in the industry, but we're all professionals here.)
Less hype than Kentucky deserves? Perhaps. It's a bit odd that in UK's three-year reign from 1995-96 to 1996-98, during which the Wildcats won two titles, they were never once ranked preseason No. 1 by SI. Then again, SI's only in-season college hoops cover of the past four years (i.e., non-preview, non-tournament) has been ... Kentucky last February. You do make a good point about the lack of individual honors for Kentucky players. I think it's directly tied to the team-first mentality that Tubby Smith has instilled in the Wildcats, an approach that has brought his guys a lot of success over the years (especially last season) though perhaps at the expense of individual awards.
I was very happy to see Virginia beat Minnesota. But I can't fall into the Pete Gillen trap of early season wins and ACC-season collapses. How long can he continue to produce average teams and still have a job?--TG, Chicago
Gillen's job is definitely on the line this year, and regrettably for him it appears that the ACC is even better than most people thought it would be (with Georgia Tech and Wake Forest exceeding expectations so far). There's no reason Virginia can't join those two teams, but there will be very few easy conference games in the ACC, and the Cavs can't afford another swan dive.
Williams College, the defending Division III national champions, went into Holy Cross and beat the Crusaders 78-71. Shouldn't the fact that a Division III team handled them at least make some appearance on the college basketball scene?--Eric, Boston
Major, major snaps for the Ephs, who shot 52 percent (from two-point and three-point range) to end Holy Cross's record-tying 14-game home winning streak at the Hart Center. Williams coach Dave Paulsen rode Ben Coffin, Michael Crotty and Chuck Abba (63 combined points) to the upset. Keep in mind, this is the same Holy Cross program that has made three straight NCAA tourneys, in which it had second-half leads on two subsequent Final Four teams (Kansas '02 and Marquette '03) and lost by four to another power (Kentucky '01). And though Ralph Willard's Crusaders lost three 1,000-point scorers after last season, you can bet Willard had no idea Williams would beat his guys at home.
The Ephs are my new favorite D-3 team -- and happen to own a 45-game home-court winning streak of their own.
How many coaches are successful at their alma maters? How many are there currently?-- Bob Hamilton, Brasfield, Ark.
Can we get any help from readers on this one? (Respond below.) We'll revisit this question in detail next week.
I got a lot of response on last week's comments regarding Kansas, Roy Williams and North Carolina -- and a few of them were even thoughtful. The basic complaint from Kansans is that the media won't let the issue go, and I can understand that (especially when you listen to what's being said on Kansas TV broadcasts). Still, a lot of people (including, surprisingly, a fair number of Jayhawks fans) agreed with me too. I've said my piece, but there were two somewhat related questions I wanted to address:
Though I am a huge Carolina fan who is glad Roy is home, Matt Doherty is not a bad basketball coach. With some improvement of his people skills isn't he likely to land a job somewhere soon, as many programs could use a coach with his credentials?--Matt Cole, St. Petersburg, Fla.
I think Doherty is extremely likely to land a good job for next season. Remember, this guy was the national coach of the year three seasons ago, and he was the guy who recruited the current North Carolina team, which resides in the Top 10.
Grant Wahl will periodically answer questions from SI.com users in his mailbag.
Doherty and I spoke about his future in coaching in late September. Here's what he told me when I asked him if he was interested in coaching again next year:
"I think so," Doherty said. "This year I want to spend time with my family, do some TV and see where the fire burns after the season. I was watching Remember the Titans the other day, and it made me think about the emotions you go through. There's nothing that moves you emotionally like coaching. You have high highs and low lows, and to be able to have an impact on young people, to get a group of young men together and accomplish a goal, that is a real exciting challenge for me. You don't get that in any other profession."
Does he have any major regrets about his experience at North Carolina?
"Major regrets? Obviously I could have stayed at Notre Dame. It was a great time for me to be at Notre Dame. But I made the best decision I could [to come to North Carolina] given the information that I had. There are certain things I would do differently. I've taken some executive MBA classes studying business leaders and leadership, people who deal with takeovers. I've made some mistakes I would handle differently if I had to do it over again. That's experience. Maturity. I've never been in that position before. There were other things out of my control. But I'm 41. I'm young, and I'll be that much smarter and more prepared for the next opportunity."
Has there ever been a situation like the Roy Williams-KU-UNC triangle before, i.e. 1) a coach at the absolute top of his game leaving a team at the apex and 2) departing and creating this much hostility?-- Dan Tolhurst, Toronto
The best current example is Rick Pitino, who didn't set off Kentucky fans until he took the Louisville job. The circumstances are slightly different than the Williams case, of course: 1) Pitino didn't truly anger Kentucky fans until several years after he left Lexington; 2) Pitino has always been viewed as something of a carpet-bagger, whereas Williams hasn't, and 3) Pitino has to deal with upset Cats fans in person, which Williams gets to avoid (more or less) with Jayhawks supporters. Then again, the enmity toward Pitino has died down quite a bit over the last two years, just as I expect it will for Williams.
Where Are They Now
My bad. I didn't get a hold of Larry Eustachy (the cell phone number I had for him wasn't working), but the 'Bag has learned that Eustachy is busy traveling to games again. He was spotted at Oklahoma and Oklahoma State home games last week, and word is that his recovery from alcoholism is going well and he's playing a lot of golf. Having written SI's story on Eustachy last May and spent five days in Ames, I know that he and his loved ones don't view his admitted alcoholism as a laughing matter. I'm inclined to agree. I expect that like Doherty, Eustachy (himself a recent national coach of the year) will be back on the sidelines soon.
Next week's WATN comes from Greg De Vries of Telkwa, B.C., who asks:
Where in the world is Scotty Thurman?
See you next week.
Sports Illustrated senior writer Grant Wahl covers college basketball for the magazine. His story about the Princeton offense, which ran in the Feb. 17, 2003 issue of SI, was awarded first prize in the magazine-length feature category of the 2003 U.S. Basketball Writers Association writing contest.