RPI Forecast

[billyfa]

The "RPI Forecast by Final Record" section that stever20 is referencing is pretty awesome.

Agreed. Best part of the site. For reference, here is the conference record needed by each team to get into the RPI top 45 (according to their RPIForecast).

Villanova ---- (8-10, assuming an OOC win at Temple)
Creighton ---- (11-7)
Butler -------- (11-7)
Xavier -------- (11-7)
St. John's ---- (11-7, assuming an OOC win vs Dartmouth)
Georgetown - (11-7, assuming an OOC loss vs Michigan St)
Seton Hall --- (12-6)
Providence --- (12-6)
Marquette --- (13-5)
DePaul - (none of the 10,000 simulations had DePaul with a top 45 RPI)

Obviously RPI isn't the be-all and end-all of the selection process, but this gives a loose approximation. 11 seems to be the magic number for teams not named Villanova. The best case in term of maximizing the number of BE bids is for Nova to have a disappointing conference season. It would eliminate our best chance at a top seed, but they'd really need to tank to miss the tournament completely.

It's worth noting this doesn't take conference tournaments into account.

That is a pretty cool feature!

[stever20]

You've got the RPI part of it right, but the projection is based on Sagarin's predictive rating, which is a true power rating. That means that it uses margin of victory to make its changes and then predicts the outcomes of future games based on that before it calculates RPI.

The key factors are strength of the opponent and point spread, so you can't just look at won-record and treat it as though that's the only difference those unanticipated wins make.

I'd agree with that. I would say however that with relatively speaking far back Butler/Marquette are- for them to get even onto the bubble, they themselves are going to have to win some games they're currently projected to lose. in yesterday's run- both teams had 17 projected d1 wins. For them to even get on the bubble, they will have to likely both win 19 games.