My 2009 picks

To view my picks select the pool size and which estimate of actual probabilities you wish to use.
See my 2008 picks

How it works

Our methods area based on the premise the favorites tend to be picked to go farther in the tournament that they should. In order to win a large pool, you need to go against the crowd but still pick teams with good chances of winning. To make our picks we need to sets of inputs: estimated actual winning probabilities for each team and data on who everyone else is picking. Finally, you need to know how big of a pool you are in. In larger pools it is more important to avoid the crowd. In smaller ones, you should be picking the teams that have the best chance of winning.

Actual probabilities

These can derived from historical information (e.g. how often has a 1 seed beat a 4 seed, etc.), or from various rating systems (e.g. Massey, Sagarin or my own ratings)

Perceived probabilities

Both Yahoo and ESPN publish what percentage of people picked a given team to advance to each round. This gives you a fairly accurate picture of who is picking whom.

Discrepencies between actual and percieved probabilities

Note that their are major differences between how people pick teams and how likely they are to make the final 4 or win the title.

Team	Reaching the Final Four						Winning the Tournament
	Estimated Actual Prob				Perceived Prob		Estimated Actual Prob				Perceived Prob
	Historical	Massey	Sagarin	Letscher	Yahoo	ESPN	Historical	Massey	Sagarin	Letscher	Yahoo	ESPN
1 Louisville	43.4%	21.9%	22.9%	28.4%	65.3%	62.2%	13.7%	6.8%	6.9%	8.7%	20.3%	17.8%
1 Connecticut	43.4%	26.0%	24.7%	31.4%	45.6%	39.9%	13.7%	9.2%	8.6%	11.0%	8.0%	7.5%
1 Pittsburgh	43.4%	24.2%	24.6%	30.6%	56.9%	57.3%	13.7%	8.4%	8.5%	10.6%	15.8%	16.1%
1 North Carolina	43.4%	24.5%	27.0%	31.9%	61.9%	61.0%	13.7%	8.1%	9.8%	11.0%	28.7%	28.2%
2 Michigan State	22.3%	22.7%	17.0%	22.3%	12.0%	13.5%	6.1%	7.3%	4.4%	6.4%	1.8%	2.1%
2 Memphis	22.3%	17.0%	18.8%	20.3%	39.5%	44.3%	6.1%	4.8%	5.8%	6.0%	7.1%	8.8%
2 Duke	22.3%	19.9%	22.2%	23.4%	21.2%	21.3%	6.1%	6.5%	7.3%	7.5%	3.7%	3.9%
2 Oklahoma	22.3%	14.7%	13.7%	16.6%	17.0%	18.7%	6.1%	3.7%	3.5%	4.3%	4.5%	4.8%
3 Kansas	12.3%	9.6%	10.9%	10.8%	9.6%	9.9%	2.9%	2.1%	2.5%	2.3%	1.1%	1.2%
3 Missouri	12.3%	11.8%	13.0%	12.7%	6.0%	7.6%	2.9%	2.8%	3.4%	3.1%	0.7%	0.8%
3 Villanova	12.3%	13.2%	10.7%	12.2%	10.4%	12.1%	2.9%	3.5%	2.5%	2.9%	0.7%	1.0%
3 Syracuse	12.3%	12.4%	10.9%	12.2%	12.1%	13.1%	2.9%	2.9%	2.6%	2.8%	2.3%	2.6%
4 Wake Forest	6.4%	10.9%	10.9%	10.3%	8.2%	9.9%	0.9%	2.6%	2.5%	2.1%	0.7%	0.9%
4 Washington	6.4%	6.8%	7.6%	7.0%	1.8%	2.2%	0.9%	1.4%	1.7%	1.3%	0.3%	0.3%
4 Xavier	6.4%	5.0%	7.3%	6.0%	1.7%	1.7%	0.9%	0.8%	1.4%	0.9%	0.2%	0.1%
4 Gonzaga	6.4%	9.2%	12.7%	10.1%	3.0%	3.9%	0.9%	1.9%	3.3%	2.0%	0.5%	0.5%

Notice, for example, approximately 28% of people are picking North Carolina to win the champion ship. But using historical seed performance or various ranking systems they have an estimated 8-14% chance of winning the championship. So if you want to do better than the masses, maybe you shouldn't pick North Carolina. In contrast, look at Conneticut (also a #1 seed), people are more realistic about their chances of winning. This might be a better choice for picking a champion. Or maybe a #2 seed.

References

Our article in the 2006 NY Times sports section on picking your pool sheet.

My paper with Bryan Clair on "Optimal Strategies for Sports Betting Pools": pdf and our software for making picks.

Results and data from past tournament pools.