rochsioux Posted December 2, 2014 Share Posted December 2, 2014 The pairwise on SiouxSports does not match USCHO or collegehockeynews. The biggest difference is with Harvard which is ninth on SiouxSports and sixth on the other two sites. The problem appears to be with the RPI calculation (.5929 vs .6031). The other two sites are excluding some games from the calculation while SiouxSports is not. There are also some minor differences in RPI (e.g. Minnesota: .6004 vs .6007, UNO: .6152 vs .6143, etc.) Quote Link to comment Share on other sites More sharing options...
Goon Posted December 2, 2014 Share Posted December 2, 2014 The pairwise on SiouxSports does not match USCHO or collegehockeynews. The biggest difference is with Harvard which is ninth on SiouxSports and sixth on the other two sites. The problem appears to be with the RPI calculation (.5929 vs .6031). The other two sites are excluding some games from the calculation while SiouxSports is not. There are also some minor differences in RPI (e.g. Minnesota: .6004 vs .6007, UNO: .6152 vs .6143, etc.) IRC, last season Jim was right on with his Pairwise calculations. Quote Link to comment Share on other sites More sharing options...
jimdahl Posted December 5, 2014 Share Posted December 5, 2014 Sorry, my PWR time this week has been sucked up by working on this (awesome!) article -- When to start looking at PWR (revisited). But, I'm looking into it. Because each is independent, If two of the three agree, odds are the third (me!) is wrong. Since the formulas haven't changed, it's probably a score discrepancy. Though I have automated double-checks of USCHO and CHN against my scores, now that neutral vs. not matters it's still possible to get the scores right but come up with different PWRs. Quote Link to comment Share on other sites More sharing options...
jimdahl Posted December 5, 2014 Share Posted December 5, 2014 IRC, last season Jim was right on with his Pairwise calculations. But yes, historically when there are formula changes, USCHO is last to get it right. But there were no formula changes this year and apparently USCHO and CHN agree, which strongly suggests that I've got a bad score (or location) somewhere. I'll dig into it tomorrow and post an update before the weekend's games start. Quote Link to comment Share on other sites More sharing options...
jimdahl Posted December 5, 2014 Share Posted December 5, 2014 Ok, they agree now. I was including the exhibition game between Dartmouth and Brown (because I was just testing to see if both were NCAA teams). I'm now properly excluding exhibition games between NCAA teams. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.