RPI predictions

[jesterondirt]

Ok don't shoot me for bringing this up in November, I'm not so much curious about the our current RPI, because I do understand how it doesnt have enough data to be relevant this early. What I am curious about is their win/loss predictions.

I would assume they're computer generated correct?

I noticed they have us currently projected at 10 losses and one of those being @Tulane. I'm not going to argue the others but how would they have arrived to this prediction?

http://realtimerpi.com/rpi_424_Men.html

 

[cmullinsTU]

They use their own created GAMER analysis to predict winners. I suspect we are projected to lose almost all of our away games (home court advantage is one of the variables) because we haven't played a road game and all of our opponents have won at least 1 home game.

 

[ctt8410]

That site also does a poor job of handling probabilities. If you're a 1-point underdog in every game, then you're projected to go 0-30.

I prefer to use Pomeroy for projections and then RPIforecast to look at the RPI corresponding to each record. That would put the current projection at 20-10 with a 41 RPI. Then you can tweak up or down from there if you think the computers are wrong.

 

[astonmartin708]

My predictions for this season: Note the T's are tossups. W's are good chances for wins, and L's are good chances for losses.

So I'm saying approximately 15 sure wins and 6 sure losses, with 9 games being up in the air. I would set our win total o/u at 19.5