[Bill Dozer]

Gaming the Game in March Madness Brackets
By: Elihu Feustel

Picking winners is key to any March Madness contest, but maximizing your points is just as critical. Moneylines and the number of contestants are two criteria to keep in mind filling out your brackets.

Thousands of basketball experts enter brackets every year, but a vast majority of them have virtually no chance at all to win. While it is often fairly straightforward to produce a bracket that maximizes your expected points, most players do not know what adjustments to make to win a "lottery type contest." Always remember that your goal is to win the contest, and NOT to pick the teams most likely to win.

There are two things you should do to maximize your chances of winning a contest. First, identify what selections will give you the highest expected number of points. This process is boring – if no upsets occur, what will happen?

I use moneyline odds to estimate the chances for any given team to win a matchup in the first round. For example, if TheGreek lists the West Virginia/Arizona matchup at -135/+115, I convert this to a win percentage, which is about 56% for West Virginia. For the second round and on, I use Sagarin ratings from USA Today as a guideline for which teams are better. Using spreads and power ratings, it is straightforward to at least decide which team should be favored.

The second thing to focus on is number of competitors you must beat. The more people you are up against, the more you should drift from 'proven and true' teams towards long-shot underdogs. In brackets with a lot of competitors, you want to ensure no one has picks too similar to yours. One way to do this is to pick several upsets in the mid-to-later rounds. If the rest of your selections are conservative and you win your two 'upset specials,' you’ll probably finish high in the money.

It is important to only pick a few underdogs though – many people hurt their chances by picking too many upsets. If you have fewer than two No. 1 seeds in the Final Four, you might want to rethink a more conservative approach.

I’ll explain how I filled out my brackets for two contests, starting with the free brackets contest SBRforum.com is offering. The base scoring is standard – the points per game double each round (except the last two, which are slightly less). There are two key variations though.

First, Round 1 awards more for each seed differential if you pick an underdog; if you pick a No. 13 seed to beat a No. 4, that winning selection would award 10 times as much as picking the No. 4 seed. Second, the final round multiplies the points won by the seed of the winner meaning a No. 4 seed will be worth four times as much as a No. 1 seed. These two rule changes drastically affect the basic selection process, even before adjusting for the number of entrants.

One way to maximize your points is to first decide which 16 teams are most likely to advance to the Sweet 16. After filling in those teams for the first two rounds, mark upsets for all of the other Round 1 games. This will give you a lot of upsets, half of the Round 1 games to be exact, but also should give you the most expected points. You don’t want too many upsets though, or you won’t have good teams left to advance into the Sweet 16.

Once you’re to the Sweet 16, you need to incorporate your gaming theory. Which one team do you think will win it? The rule rewards picking a long shot that has a legitimate chance. In my bracket, I picked Wisconsin to win it all. As a No. 3 seed, they have a higher Sagarin rating than the No. 2 seed in its bracket. If they do win it all, I’ll get three times as many points as any No. 1 seed that might win it.

I also picked Stanford over Texas – this isn’t a huge upset, but keeps moving my bracket away from all favorites. Once I’ve picked those 'underdog specials,' I finished the rest of the bracket with boring selections that are most likely to win – in this case, UNC, Memphis and UCLA (with UCLA losing to Wisconsin in the final).

Now consider A-rated TheGreek.com’s bracket contest. The last time they had this contest there were several thousand entrants, with prizes paid out to the Top 300. To 'make the money,' your goal should be making the top 10% of entrants. They have a fairly typical scoring method, with the points for each game doubling each round (although the last two rounds are slightly less). In TheGreek’s contest, your default – where you have the highest expected points – is simply picking which team that is favored in any matchup. Using spreads, there are three matchups where a lower seed is likely to win the match – Kent State, Davidson and Mississippi State. Other than these three matchups, I took every favorite in Round 1.

In the next couple rounds, I’ll keep using boring selections of the team more likely to win. The one exception – I’m picking two underdogs to over-perform; if they do I’m a favorite to be in the money. Duke has the highest Sagarin rating for a non-No. 1 team, so they are a good long-shot candidate. My Final Four had three No. 1 seeds plus Duke, with Kansas beating Duke in the final. Now I don’t feel strongly about any team winning it all, or even Duke being that much better than other teams – it’s all about maximizing points, and minimizing other peoples’ chances of having similar brackets.

© Copyright 2008 Elihu Feustel

[INVEGA MAN]

Very interesting

[bigboydan]

Very good read Mr.Feustel.

[mofome]

great read