ESPN Bracket Predictor

[A-Tex Devil]

Guaranteed time suck (http://insider.espn.go.com/ncb/tournament/predictor?conf=1&action=upsell&appRedirect=http%3a%2f%2finsider.espn.go.com%2fncb %2ftournament%2fpredictor%3fconf%3d1) for anyone making brackets next week.

Unfortunately, I doubt it will be free after Sunday at about 6:00 Eastern. This thing basically lets you pick any two teams, then choose the statistical criteria you think have the most predictive value and it will spit out a winner and predicted score.

I may sign up for ESPN Insider simply for this. I'd be interested to what set of predictors is most accurate over this weekend's conference tourney games.

I could geek out over this for a good week. But I'm also a degenerate gambler.

[MulletMan]

Guaranteed time suck (http://insider.espn.go.com/ncb/tournament/predictor?conf=1&action=upsell&appRedirect=http%3a%2f%2finsider.espn.go.com%2fncb %2ftournament%2fpredictor%3fconf%3d1) for anyone making brackets next week.

Unfortunately, I doubt it will be free after Sunday at about 6:00 Eastern. This thing basically lets you pick any two teams, then choose the statistical criteria you think have the most predictive value and it will spit out a winner and predicted score.

I may sign up for ESPN Insider simply for this. I'd be interested to what set of predictors is most accurate over this weekend's conference tourney games.

I could geek out over this for a good week. But I'm also a degenerate gambler.

Not that I'm promoting gambling, but this thing is sweet! I bit the bullet and paid for Insider just so I can use it. Not gonna lie, I got it for the ummmmm... well... ah... analysis of spreads :rolleyes: but it certainley helps break the tie on some of those pesky matchups I can't decide on in my brackets. Its awesome how you can do pretty much any combo of stats and compare two teams. I also dig the "similar games" breakdowns which show you matchups from previous tourneys which have statistcial profiles a kin to those of the teams you're checking out... that way you get an idea of what actually played out in a game in the tourney with two teams that play similar styles to those you're analyzing. I dig it.

[ForeverBlowingBubbles]

http://insider.espn.go.com/ncb/ncaatourney09/insider/news/story?id=3986309




Duke makes it to the finals!

[FerryFor50]

Eh....

It tends to pick the higher seed EVERY time. Kind of boring and definitely not what I see happening.

[MulletMan]

Eh....

It tends to pick the higher seed EVERY time. Kind of boring and definitely not what I see happening.

Not really. It depends on what combination of statistics you use. In fact, they have a guy from ESPN who used offensive efficiency, defensive efficiency, turnover percentage, predictive rating, and points allowed as his criteria, and that resulted in Duke beating UNC. :D

[devildeac]

Not really. It depends on what combination of statistics you use. In fact, they have a guy from ESPN who used offensive efficiency, defensive efficiency, turnover percentage, predictive rating, and points allowed as his criteria, and that resulted in Duke beating UNC. :D

Doesn't sound like he included the hess or valentine factors then.:rolleyes:;).

[A-Tex Devil]

Eh....

It tends to pick the higher seed EVERY time. Kind of boring and definitely not what I see happening.

I think it depends on how you use it too. If you think offensive rebounding and low turnover rate are important, you can weigh those more.

I mean Pomeroy (other than 3-4 games) and virtually every predictor are going to pick higher seeds in almost every game. You aren't going to find a model that nails the upsets. But you can find edges, which is this whole point to <ahem> <chuckle> beating Vegas. :rolleyes:

[riverside6]

I think Bracket Brains (http://www.teamrankings.com/r/bracket-brains/?ref=53680) works better.

It does something similar. You can see some of their predictions on my site right now, as we shared some of their information on the ACC's first round games (http://www.scacchoops.com/tt_NewsBreaker_External.asp?NB=1844).

[feldspar]

I think Bracket Brains (http://www.teamrankings.com/r/bracket-brains/?ref=53680) works better.

It does something similar. You can see some of their predictions on my site right now, as we shared some of their information on the ACC's first round games (http://www.scacchoops.com/tt_NewsBreaker_External.asp?NB=1844).

Bracket Brains is where ESPN got their info from.

[MB in MD]

A couple of years ago I had my son write a little computer algorithm that picked brackets using a weighted random selection method based only on seeding. Thus, in a 1 v 16 the 1 seed would win 16/17 times, but in a 2 v 3 the edge was 60/40. Had him run it n times and selected one run at random and entered it in my pool along with my own "reasoned" entry. He finished third; I was way back in the pack.

I think over the long haul this method would do just as well as Bracket Brains.

[Namtilal]

I was just playing with the Accuscore program on ESPN, clearly inferior to the one mentioned here -- it gives no explanation for how it ranks teams, and just rolls the dice to see who wins.

Amazingly, this program believes Duke is the underdog in head-to-head matchups against *14* different teams. Whatever measuring stick they used overlooks the positives that led our team to a great regular season/ACC tourney. It's free, not Insider. You get what you pay for with this one.

http://sports.espn.go.com/ncb/ncaatourney09/bracketcaster

[riverside6]

Bracket Brains is where ESPN got their info from.

Interesting, didn't know. that. Unlike those above, I think BB impressive. I use it more as a guide but it doesn't get in the way of my instincts on certain picks.

[FerryFor50]

I think it depends on how you use it too. If you think offensive rebounding and low turnover rate are important, you can weigh those more.

I mean Pomeroy (other than 3-4 games) and virtually every predictor are going to pick higher seeds in almost every game. You aren't going to find a model that nails the upsets. But you can find edges, which is this whole point to <ahem> <chuckle> beating Vegas. :rolleyes:

Very true.

My factors were:
off efficiency
def efficiency
turnovers
def rebounds (covers skill of rebounding and opponent FG%)
possessions

[riverside6]

It seems a lot of the efficiency based predictions hate ACC teams not named Duke or UNC.

I know Bracket Brains (http://www.teamrankings.com/r/bracket-brains/?ref=53680) has Clemson, BC, FSU, and Maryland all with a good chance to lose. My question to that is that because the ACC beat each other up so much? Close games would make efficiency margin less and thus make that team less likely to win in those models.

[A-Tex Devil]

It seems a lot of the efficiency based predictions hate ACC teams not named Duke or UNC.

I know Bracket Brains (http://www.teamrankings.com/r/bracket-brains/?ref=53680) has Clemson, BC, FSU, and Maryland all with a good chance to lose. My question to that is that because the ACC beat each other up so much? Close games would make efficiency margin less and thus make that team less likely to win in those models.

I wonder if it's less they hate ACC teams more than they are playing teams with abnormally high efficiencies, but not so good records (Wiscy, USC, Michigan). Clemson is playing OU, a team computers don't really like, but it makes sense they'd favor them over Clemson.

[Fish80]

This one has us winning it all!

Madness (http://www.payscale.com/2009-march-madness-predictions)

[duketaylor]

brackets based on grads median income, good thing more of the Ivies missed out.