To figure out the fair value for the myriad of prop bets, many new features were added to the log5 tournament simulator. One tweak made to the actual simulation was the addition of partial home court advantage for teams playing within 90 miles of campus. The simulator loved the Big Ten, but there were only a few bets with sufficient edge. Also, I was in Vegas for the beginning of the tournament, so failed to make a pre-tournament post.
I used the Kelly criterion as a basis to determine relative bet sizing, while keeping an eye out for correlation. As opposed to a blackjack player, repeating many trials, my goal was more about maximizing one-time value, rather than bankroll preservation, so this was fairly aggressive, concentrated portfolio of prop bets. Disappointingly, many of the lines that appeared to have value on LVH or MGM’s printed prop sheet had moved against me by Wednesday night and lost their value — strong evidence that some sharp bettors were using similar methodologies. An example was Florida, a team that Kenpom loved and the committee didn’t. MGM opened them at 4-1 to make the Final 4, but by the time I got there, they were 6-5 and I passed. On a 62 unit bankroll, here were my plays:
Louisville to make the Final 4 + 120 (15 units) Edge: 27%
Kenpom log5 had this fair at +110, a small value, but when I gave Louisville a bonus for being nearly at home for its first two games, I had this line fair at -147. I didn’t even count that Indianapolis was a pseudo-home environment for them as well. As the best bet on the board, Louisville went largely unchallenged until they ran into Wichita St in the Final 4. It was especially hysterical to hear the talking heads saying what a tough draw Louisville got. Maybe the public believed them and that’s why this line didn’t get driven too far down. This prop opened at +150.
Result: Win (+18 units)
ACC Under 6.5 Wins Even (8 units) Edge: 23%
The simulator thought the ACC was trash.
Duke and Miami might have been 2s, but it didn’t like Miami and virtually no one else in the conference was respected. This became nerve-wracking when the ACC had 6 wins going into the Louisville – Duke game, so that game determined the two largest bets. A bit more correlation than I was hoping for. This was a rare one that moved towards me, after opening at -110.
Result: Win (+8 units)
Colorado St. wins more games than Wichita St. -125 (6 units) Edge: 10%
Kenpom loved Pitt and this was the most effective way to get long Pitt. The line fell from opening at -110. Wichita St. showed well. Whoops
Result: Loss (-6 units)
Kansas wins more games than Gonzaga -105 (5 units) Edge: 7%
Another synthetic Pitt long. The sim didn’t think Gonzaga was that much different than Kansas in a vacuum, but it hated Gonzaga’s second round matchup, in which the Shockers, not Pitt, did them in. This line fell from an initial +110.
Result: Win (+4.76 units)
Indiana to the Final 4 +120 (5 units) Edge: 8%
Indiana had a fairly easy bracket and this had a ton of value at its initial +180. In the Sweet 16, they were undone by the famous Syracuse 2-3 zone and a performance by Cody Zeller that probably has him rethinking the draft.
Result: Loss (-5 units)
ACC to not win -900 (4.5 units) Edge: 5%
One of those odd cases where Kelly says to bet a lot because a win is so likely, but the payoffs kinda suck and you don’t want to divert too much bankroll here. The ACC was thin behind Duke and Miami and this one also cleared when Louisville knocked off Duke. The line was -900 all week.
Result: Win (+.5 units)
North Carolina St. wins more games than Missouri -140 (4 units) Edge: 6%
A nice hedge against all the ACC short action, the simulator liked this because it thought Colorado St. was a tough matchup for Missouri (hence the Colorado St. > Wichita St. as well). That part was right, but NC State lost to Temple and this pushed.
Louisville to win +400 (4 units) Edge: 27%
Another Louisville long and despite the high edge, it was prudent to put a little less on this one, given its more longshot nature. Other books had it at +450, which was an even better take. The two games of the Final 4 were much more eventful that Louisville’s first four games.
Result: Win (+16 units)
Duke under 2.5 wins -140 (3.5 units) Edge: 5%
Yet another ACC short. Duke ended up in the Elite 8 with ease, so there’s not much else to say here.
Result: Loss (-3.5 units)
Big 12 under 6 wins -120 (3 units) Edge: 4%
When there’s a situation to hedge with edge, you pretty much have to do it. This was somewhat offsetting the Kansas long. The result was a middle, as Kansas outlasted Gonzaga, but the Big 12 managed just 3 total wins.
Result: Win (+2.5 units)
Wisconsin to the Final 4 +800 (2 units) Edge:10%
Wisconsin was a darling of power ranking systems due to its ability to beat anyone in the mighty Big 10. They seemed like a boom or bust team, one capable of losing in the first round or launching a Final 4 campaign. We know what happened.
Result: Loss (-2 units)
Big 10 to win +240 (2 units) Edge: 7%
This was about collecting some value for the field of non-Louisvilles. Indiana and Ohio State were the thoroughbreds here, but the sim thought Michigan State, Michigan and Wisconsin were all north of 2% to win it. The combination of this and the Louisville win ticket took a lot of sweat out of the final.
Result: Loss (-2 units)
Bets passed on because odds moved away:
- Missouri Valley Conference under 1.5 wins
- Pac 12 under 3.5 wins
- Florida to make the Final 4.
These three had one thing in common: They all lost. So, luck was helpful here. However, if I’d gotten the opening lines, I’d have gotten better odds on my winners, so luck/line movements were something of a wash.
Total return: +31.26 units, +50%
Obviously, there was an extremely strong bias towards Louisville and the Big 10, while going against the ACC. The rest was neither here nor there. To some degree, it’s fine to take a strong market opinion (Duke has a lot of retail /public money) and be willing to ride it. Variance is increased, but total return will be maximized. Obviously, this isn’t a strategy that will have a 2x or 3x return in any given year, but an expectation in the 10-15% range with some entertainment is nothing to sneeze at.