ESPN March Madness Bracket Challenge - Atlas Group

[muon2]

I don't do a bracket, but I do watch the statistics of upsets. Statistically the lower seed should win about 22% of the time which means 7 upsets in the first round, though 6 or 8 is well within the margin of error. Obviously the closer the seeds are the greater the chance of the upset. So far it's running below expectations with only 2 upsets in the first 32 games.

[muon2]

6 upsets in the first round. On the low end of the expected range of 6 to 8. Interestingly no upsets above a 5 seed.

[muon2]

With 2 upsets today the overall average moves to 20%, very close to my expected average.

[muon2]

I glanced over at the leaderboard for the full ESPN challenge, which has 18.8 million entries this year. The top score is 470 out of 480, so there are no perfect brackets at ESPN after 40 games despite the huge pool of entries. Statistically it's really that hard.

Just consider the 1 seeds. They don't lose in the first round so those 4 games are easy to pick. Let's assume that there is a 1 in 6 chance that a 1 seed would be upset in the second round (the actual chance is pretty close at 85%), which means that the 1 seed wins 5/6 of the time. The chance that all four win the second round is (5/6)^4 or about 48%. That means there about an even chance that at least one of those top seeds will be upset. The best brackets would play the odds that each would advance, but over half the time that strategy loses. Guessing at one number one seed to lose may make some sense, but there's only a 1 in 4 chance of getting the right loser, so over all a bracket with one loser in this scenario will be right less than 10% of the time.

Picking the opponent for those number one seeds is tough, too. The 8-9 game in the first round is basically a toss up. So there is a (1/2)*(1/2)*(1/2)*(1/2) = 1/16 = 6% chance of getting them right.   Combine that with the probabilities for the second round match up and the chance of getting the first and second round winners of the 1-8-9 games (12 games) is down to about 4% or 1 in 25.

If each set of 12 games at the second round was equally hard, then the odds of a perfect bracket after the second round would be about 1 in 300K. But the other games don't have the gimme of a 1-16 match up, so the odds balloon to the point where none of the 18 million got it.

[muon2]

After 6/32 = 18% upsets in round 1, there were 4/16 upsets in round 2, for an overall upset rate of 21%. That's basically right on the average.

[muon2]

Do you guys think I still have a chance(my bracket is Oregon will Win)

If Oregon wins then it's very likely your bracket will win. Picking the champion gives you an extra 320 points compared to those who don't. There's also the guaranteed 160 that will at least match anyone who didn't pick Oregon in the semifinals. So unless you fall more than 320 points behind the leaders, you have a chance with Oregon.

[muon2]

Going into the final four there have been 14 upset seed wins out of 60 games or 23.3%. That's very close to my expectation of 22% and within the normal range of 19-25%. If you picked the higher seed to win every game so far you would have 820 points and be sitting in third place in Atlas pool.

Picking by seeds would leave 4 number 1 seeds in the Final Four. If you picked the last three games based on the postseason rank in the AP poll (1-Villanova, 2-Gonzaga, 3-Kansas, 6-North Carolina) then you'd have Villanova over Kansas in the final and finish with those same 820 points.