A grad school friend and I do it every year, but in a couple of dollars and put it in my pool. It never wins, however, it can give you some general trends on who to like/avoid. We have data going back the last 13 years for all the teams that are in the tournament a some stats (Sagarin rating, sos, rpi, conference Sagarin, conference rpi, offensive and defensive statistics as well) to create a profile of winners and losers for each round based on the past 13 years of data. In the analysis when two teams fit the "winners" profile, we have two different decision rules for the tie breaker (absolute value of winning score, difference of winner profile score - loser profile score). The absolute value tends to be more conservative, while the difference score tends to be more liberal...and boy is it this year.

Overall trends:
1. the data does not like N. Carolina, having them go out in the 2nd round both times (to Creighton).
2. Kentucky loses in the sweet 16 in both (once to Indiana, the other time to Wichita State).
3. With that in mind, I totally expect the national championship game to be Kentucky vs. N. Carolina.
4. Not a lot of upsets in the first two rounds, Other than North Carolina losing, both sets of brackets are pretty close to chalk.
Here are the final fours and championship games:
Absolute value
Indiana, Michigan State, Florida State, Kansas. With Indiana defeating Kansas.
Difference score:
Wichita State, Michigan State, Syracuse, Kansas. Syracuse defeating Wichita State

Things to note:
The analyses do not take injuries/suspensions in account, so um, I think the 2nd bracket is totally screwed...unless Wichita makes it to the finals.
We really need to take defensive statistics into account more, I think that might be my project after tenure.
If you want to see more/the analysis output, I'll bring it to bowling, it's interesting b/c the analyses for the first three rounds and the final round is significant (meaning it should do better than chance at predicting the winner) but rounds 4 and 5 it dies.

_________________

I hope your absolute value is right!

_________________
"He is a loathsome, offensive brute
--yet I can't look away."

I love the absolute value prediction.

_________________

I hate the absolute value prediction.

_________________
You do not talk to me like that! I work too hard to deal with this stuff! I work too hard! I'm an important member of the CSFMB! I drive a Dodge Stratus!

_________________

I hope the all chalk thing is right, just add North Carolina to it also.

Of course, I just heard somebody say Alabama will beat Creighton.

I wrote an app last year that would do mathematic modelling of NCAA basketball games and it was working fairly well on some of the more esoteric over/under matchups -- basically, it was good at estimating tempo and possession conversion. I was doing alright on matchbook for admittedly small levels of risk. Then matchbook shut down and I couldn't find any other places to take my crazy Furman vs Alcorn State action, so I backed off a bit. But when I ran it through the tourney, it was horrible at ranking the conferences. I put in a bracket anyway, and I think it finished 3rd from the last in a group of about 80.

_________________
"All crowds boycotting football games shouldn't care who sings or takes a knee because they aren't watching." - Nas

The problem we have, is that no matter want, we can't account for all variance. The profile analyses for rounds 1, 2 and 3 are predicting at around 75% accuracy which my gut tells me is about as high as we are going to do. The problem is that if you are using it to predict the whole tournament, you are going to get some people who are lucky and are picking at 90 or 95%. The one thing that it does tend to be good at is telling which high seeds are vulnerable for upsets.

_________________