To figure out the fair value for the myriad of prop bets, many new features were added to the log5 tournament simulator. This year it was tweaked to include the play-in games and a comprehensive mileage from home adjustment. The code will not be updated on the link.
#2 seeds weren’t too popular. Giant blip in the distribution of three #1s and one #2 making the Final Four.
The big theme was that the markets were treating Michigan State and Louisville much better than any model would suggest. Louisville fell from 12-1 to win it all on Sunday to 9-2 by Monday, without even playing! When I ran my Monte Carlo with kenpom values, the odds for each of those to make the Final 4 and then win were egregiously far off. I gave Michigan State a slight bump because they’ve had some key players return from injury. To align Louisville’s Monte Carlo odds with what the market was implying would’ve required a bump to make them far and away the best team. The plan was to structure prop bets to get short Louisville and Michigan State, while picking up what little value lie elsewhere.
Virginia also seemed to offer great value, though it was directly correlated to the model not liking Michigan State as much as the public and sports books. Virginia has a much easier path to their potential Sweet 16 match-up and all of their games are in North Carolina, a bus ride away.
I used the Kelly criterion as a basis to determine relative bet sizing, while keeping an eye out for correlation. On a 64 unit bankroll, here were my plays:
American Athletic Conference Under 5.5 Wins Even (6 units) Edge: 19%
Line opened +120 and quickly slid to Even and later into minus territory. Louisville is the only real team here. UConn, Cincy, Memphis are unlikely to win more than a game. So it’s a short Louisville play. Left of the dash indicates the winning part of the distribution.
Big East Under 5.5 Wins Even (10 units) Edge: -10%
This was a mix-up. New simulator missed picking up Xavier’s play-in game and any possible wins after that. When Xavier lost, the line looks like the one I thought I was betting on at 18% edge. Lucky break and lesson learned. Funny enough, the line slid from +120, so someone else went the same way in size.
Big Ten Under 10.5 Wins -145 (8 units) Edge: 20%
This was the one to really back the truck up on. Obviously, Michigan State is driving some of this, but holy cow, look at all that territory to the left of the over under.
One Seeds Over 11 Wins + 120 (8 units) Edge: 27%
The market was insisting that Louisville will walk through Wichita State and Michigan State through Virginia. If the #1s all get to the Sweet 16, this should be pretty easy to cover.
Two Seeds Under 8.5 Wins + 120 (6 units) Edge: 14%
The sim didn’t think much of the two seeds, as evidenced by the bump in the first histogram.
A 14 or 15 Seed Wins – 110 (11 units) Edge: 39%
Because Louisville and Michigan State, who are admittedly good teams, were #4s, the #2s and #3s are weak. The sim thinks that there’s an upset 71% of the time. It’s a fun one to root for as well!
Arizona Wins More Games than Louisville – 120 (6 units) Edge: 17%
Louisville has a much tougher path than Arizona. This was borne out by the head to head simulations.
Arizona to Win + 700 (4 units) Edge: +28%
The sim likes Arizona, as they have a relatively easy bracket (aside from the second round minefield). They also play all of their games in SoCal, a very easy trip. This line has floated higher, which seems kind of odd.
Virginia to Win + 1500 (1 unit) Edge: +122%
That line was grabbed Sunday as they were beating Duke. It actually went up to 18-1 later in the day before settling at 12-1. Should’ve grabbed some more!
Virginia to Final 4+ 300 (3 units) Edge: +48%
Already super long Virginia, so no need to go too crazy here, despite the nice edge.
Tennessee to Win + 7500 (1 unit) Edge: + 20%
This was taken before Selection Sunday. When they got tossed into the play-in round, the fair value plummeted to negative value. I took this instead of putting a unit on Louisville back when they were 12-1. Dumb in retrospect, because this line was going nowhere, but Louisville could only get richer. Also, the sizing was wrong. A fractional unit at 75 to 1 would’ve been more appropriate.
Unlike the last year where there some intriguing head to head bets (Team X wins more games than Team Y), none of them were any good unless they involved shorting Louisville or Michigan State. This is a very correlated portfolio whose variance is going to swing on the Round of 16.
Survival Odds for each round
Fair Odds for each interval