Realtime Rpi Now Projection .............................

[MeanGreen61]

UNT season record 23-4, conference record 12-2

http://realtimerpi.com/rpi_405_Men.html

[KoldBeer]

UNT season record 23-4, conference record 12-2

http://realtimerpi.com/rpi_405_Men.html

I'd be very, very happy if that was our record at the end of the season.

Kenpom.com has us at #131 with a projected record of 23-6 (12-4)

http://www.kenpom.com/team.php?team=North%20Texas

[Mean Green Matt]

UNT season record 23-4, conference record 12-2

http://realtimerpi.com/rpi_405_Men.html

They are missing two conference games...

[UNTFan23]

They are missing two conference games...

RealtimeRPI lists as 23-4 (14-2). Kenpom failed rudimentary math and has us at 23-6 (12-4) yet only shows 2 conference losses (@WKU & @ASU).

[CMJ]

RealtimeRPI lists as 23-4 (14-2). Kenpom failed rudimentary math and has us at 23-6 (12-4) yet only shows 2 conference losses (@WKU & @ASU).

KenPom usually states that his record and game projections don't always match because he computes what the full record should be on his models and then game by game matchups separate from each other.

[Mean Green Matt]

KenPom usually states that his record and game projections don't always match because he computes what the full record should be on his models and then game by game matchups separate from each other.

Wouldn't it be easier if he just did all the game by game matchups and added up the record?

I can make him an excel spreadsheet for that if he needs one.

[UNTFan23]

Wouldn't it be easier if he just did all the game by game matchups and added up the record?

I can make him an excel spreadsheet for that if he needs one.

Were you a DSCI student or something?

[Mean Green Matt]

Were you a DSCI student or something?

Haha, no. I took a couple DSCI classes for my business degree though. I just really like spreadsheets, and data (I'm a sports nerd), so maybe I should have been one.

[UNTFan23]

Haha, no. I took a couple DSCI classes for my business degree though. I just really like spreadsheets, and data (I'm a sports nerd), so maybe I should have been one.

I hated stats as a BCIS major.

[CMJ]

Wouldn't it be easier if he just did all the game by game matchups and added up the record?

I can make him an excel spreadsheet for that if he needs one.

I think because he assigns percentages he calculates if a team is projected to win several 55/45 or 60/40 types of games -- odds are they will drop one or two over time.

[TheTastyGreek]

Pomeroy does cumulative probabilities. If you have a 90% likelihood of winning each of the 16 individual conference games, you've only got a 18.5% chance at actually going undefeated through all of them.

His predictive model gives a probability for each individual game then a collective estimate based on those individual probabilities. If his model predicts you as a 51% favorite in 16 games, he's not actually projecting you to go 16-0. He's projecting you to go 8-8 or (maybe) 9-7.

That's why there's a discrepancy. It's also why people shouldn't crap themselves when we lose to a team once or twice a year that we're "better than".

Edited December 23, 2010 by TheTastyGreek