A Wise Crowd Selects the NCAA Winner

[UPDATED | National Bracket image | March 17 @ 7:45 AM (cst)]

Conventional wisdom tells us the NCAA Men’s Final Four will consist of all #1 seeds (Illinois, Washington, North Carolina, and Duke). It also tells us the winner will be Illinois, college basketball’s top-ranked team. However, since 1985, #1 seeds have reached the Final Four only 42.5% of the time and top-seeded times have won the tournament 55% of time. Hmm … conventional wisdom doesn’t seem to produce the best results when filling out your NCAA bracket.

What would happen if we applied James Surowiecki’s The Wisdom of Crowds theory to filling out our NCAA bracket?

Well, it would probably look like this:

[Images captured at 7:45 AM (cst) on Thursday, March 17.]

This National Bracket is from the collective wisdom of those playing ESPN.com’s online NCAA Tournament game. And as of Monday night, it mirrors conventional wisdom with Wake Forest being the only non-#1 seed to make it to the Final Four.

According to Surowiecki, four qualities must be present for a crowd to be wise:
(1) diversity
(2) independence
(3) decentralization,
(4) aggregation

This ESPN.com National Bracket seems to conform to Surowiecki’s outlined qualities needed for collective wisdom. With over 1.3M submissions last year, the ESPN National Bracket is inclusive of diverse opinions. Each submission is independently entered and the selections are not contingent upon other submissions. It is decentralized in that each person uses their own knowledge to select winning teams. And, ESPN.com aggregates all the submissions to arrive at a collective decision.

It’ll be interesting to see how the collective wisdom of ESPN.com’s National Bracket changes as thousands more diverse opinions are incorporated. You can track the changes to the National Bracket by clicking here.

Comments

[peter]

i'd argue that the opinions are not very independent. most people are heavily swayed by rankings, past performance and other's opinions.

[johnmoore (from brandautopsy)]

Peter ... valid point. People's decisions in filling out their brackets could be more independent as we are prone to being persuaded by pundits. However, each decision is independently made in so far as my decision to have Pittsburgh defeat Pacific has no real affect on your bracket.

Could this National Bracket be more perfectly independent? Yeah sure, you betcha.

[jbr]

so, what did the wisdom of the crowd predict in 2004,2003...etc? I think we need to weigh that factor in before we assume that the "crowd" has it correct...

also, you really do need to understand the demographics of the crowd before assuming their wisdom...for example, if a statistically significant segment of the population is made up of NC grads/fans, the "wisdom" is statistically suspect.

also, my personal opinion is NC has very little chance of making the title game...they are a good team, but too flawed to go the distance. They may get beaten by Iowa State or Florida. Illini has a weak bracket in Chicago...seems the NCAA wants them to make the final 4.

[johnmoore (from brandautopsy)]

jbr ... I ain't assuming the crowd is right. I'm just interested in comparing collective wisdom with conventional wisdom.

[Andrew Teman]

Interesting stuff. I have a semi-similar expiriment going right now that I am documenting on my blog, after years of frustration losing these pools to girls (no offense ladies)

http://andrewteman.org/blog/?p=90

[jbr]

Andrew, sorry to break the news, but you have doomed your experiment. Remember, (not that I really took the course, but have read enough fiction) in physics, if you observe a particle, then the particle will react to your observation. Somehow, the particle knows it's being watched and reacts to this new stimulus. Don't ask me how it knows....remember I didn't take physics....it just knows. Likely, we send out forms of energy from our eyes or something like that.

So, for you to announce your observation/experiment, you may have altered the outcome. Chances are thousands of other people, girls included, are doing the same thing in response to your efforts. This will alter your "particle" and the behavior change your are seeking will be nullified or something like that....

Hopefully, your altered experiment results will be positive ( you winning the pot ) and then, you will not care too much about the particle watching phenom. Like any serious research scientist, who cares about results as long as you get money for more research....

enjoy the hoops!

[Basil Valentine]

One year, I made my picks based on the Scrabble score of each team's name (averaged per letter so as not to penalize those with shorter names). I had Gonzaga going all the way, and finished second to last in the pool. I was able to make quite a bit of fun of the person I beat, who actually spent time making decisions.

This year, I have calculated each team's average height and weight. (In case you are interested, the average player in the tournament is 6'5" and weighs just under 206 lbs.) I also have scored each team based on the average class of the team (4 points for each Senior, 3 for Junior, etc), and also have rank, win/loss, conference, region, and seed.

I think I'm going to apply some various rules based on height and weight and see what happens. There is the "Bruisers" model (tall and heavy), the "Lanky" model (tall and lighter), the "Squat" model (short and heavy), and the "Compact" model (short and light). I may also score on proximity to the total average player scores (which by the way, almost exactly match Oklahoma's team averages).

Hail the Geek King!

[kareem]

I work for ESPN.com... let me see if I can get the National Brackets for past years. Funny enough, I was going to pick my bracket based on the NB, but then I saw it and saw that the 1-4 seeds make the Sweet 16 in all four regions, and knew that the NB would never win the pool I'm in...