Sunday, March 13, 2011

Bracket Lesson #1: Which Upsets To Pick

I'm going to go through three "bracket lessons" before going through the entire bracket game-by-game. The first one: Which upsets should we pick?

A) Don't pick 13, 14, 15 or 16 seeds to win a game, but keep those seeds in mind later: Even if you think a 13 or 14 or 15 seed is going to win their game, don't pick it. The odds are the one or more of these teams will win, but the odds are that you won't pick correctly. And you're not going to lose your bracket contest because you failed to pick a 14 seed to win a game. You're never going to lose ground over that. However, keep these dangerous 13/14/15 seeds in mind. When you're filling out your bracket a lot of it is about picking the expectation value. You want to make sure that every team you pick for your Elite 8 is still around in the second round. So if you think a 14 seed has a good shot at an upset, then don't pick the 3 seed they're playing to go too far. If you can't decide who to pick in a 4/5 game, pick whoever you think is less likely to get upset in the first round.

B) Knock yourself out picking 9 & 10 seeds: First of all, 9 seeds win in the first round 54% of the time, so it's not even an upset if a 9 seed wins. 10 seeds over 7 seeds are upsets, but they win 40% of the time. So the odds are that we'll have three or four 9/10 seeds in the second round. And those teams almost always get wiped out by the 1/2 seeds anyway, so there's not much to lose. 1 seeds win their second round game 87.5% of the time (exactly one loss every other year) and 2 seeds win their second round games 67.6% of the time.

C) 12 seeds win more often than 11 seeds: In the 26 years of the modern bracket, 12 seeds have won in the first round 34% of the time, and 11 seeds have won in the first round 32% of the time. I've talked about this many times, but I think that NCAA Tournament performance is so dependent on player psychology. 5 seeds are just much more likely than 6 seeds to enter the Tournament thinking ahead to the second round and Sweet 16, and much more likely to not take their first round opponent seriously. You're going to want to pick at least one 12-over-5 upset. There's an extra wrinkle this year with the play-in games, but it's hard to tell exactly how that will play out. It will be very several years before we can figure out statistically whether it's a benefit or disadvantage for a team to be seeded against a play-in opponent.

D) Pick major conference 11/12 seeds to win, and mid-major 5/6 seeds to lose: This has been something I've been talking about as long as I've been doing this blog. You can go back and read some of my old posts on this, but a stat that should stand out is that teams from major (BCS) conferences that earn 12 seeds actually win more than 50% of the time. And the most likely 5 seeds to go down are mid-majors. I've been recommending this to readers since 2006, and over that time we've had 12 seeds from major conferences play 5 seeds from non-major conferences on six occasions, and gone 4-2 in those games. Unfortunately, this year we have no mid-major 5 seeds, and the only 12 seed that's from a major conference is Clemson, and they have to get through a play-in game. So still pick at least one 12 seed to win, but from those raw numbers alone there's no obvious pick. We need to go through the games one by one. Things are better in the 6/11 match-ups. We have a mid-major 6 seed in Xavier, and major conference 11 seeds in Marquette, Missouri and potentially USC (if they survive a play-in game). Marquette vs Xavier sticks out as an upset pick.

No comments: