Quote Originally Posted by blacklistedbully View Post
Ok, so are we supposed to take the opposing team's RPI and multiply it by the appropriate factor? And since we're talking about which losses hurt more, we can ignore the UCSB games, as that would just compare how our road-trip compared to the HC games, not a relevant stat.

So, do we take Arizona's RPI, multiply it by 1.3, then subtract it from our RPI, followed by multiplying HC's RPI by .7 (x 2 for 2 losses) and subtract the sum from our RPI?

ETA - I know that can't be it, but perhaps something relational?
Correct in that it's not nearly that simple...

RPI is just an accumulation of overall mathematical data. You must have and apply this weighted w/l percentage to every team in the country including your own in order to accurately reflect it -- and it's only good for that "snapshot" in time. By time time you've calculated this using any "common sense" method, someone has played a game somewhere, and there's been a trickle down effect to literally every other team in the country(the 6 degrees of separation rule basically) rendering your data obsolete before you ever published it from an "exact" standpoint. In this, the teams that played the game's RPI is affected hugely, their immediate opponents are affected somewhat notably, their opponents' opponents are affected measurably, although minutely -- and the further you get from the actual matchup, the infinitely smaller the difference(it does make a difference to literally everyone though).

What I've found is that the "SOS factor" of it, while important, is basically immeasurable for the practical reasons I've shown -- unless you are a spreadsheet guru that can automate from online data and don't mind setting 3rd and 4th dimensional analysis -- which, in essence, Boyd and Nolan have both already done for us anyway.

Your own RPI-weighted W/L is 1/3 of it.
Your opponents' RPI-weighted W/L is 1/3 of it.
Your opponents' opponents RPI-weighted W/L is 1/3 of it.

You can get pretty damn close RPI Hypotheses by figuring current RPI W/L percentage -- and expected future RPI W/L percentage. Multiply both numbers times 1/3 and note the difference in the two numbers. From there, you assume your SOS is a constant(the other 2/3 of the formula which is constantly fluctuating, but unlikely to go up or down too drastically at one time and is much more "slowly moving" in notable terms overall).

EX:
Our current RPI W/L record is: 13.7-11.6
RPI WP: .5415
RPI win/loss effect: .1805

Taking 2 of 3 at home is "break even" in conference play, as is 1 of 3 on the road. Taking 2 of 3 from Arky will not help our RPI much. We'd have to sweep to see a big jumb. But if we took 2 of 3 in Baton Rouge, we'd be in business -- and we could almost quit worrying about RPI being a factor altogether. Still, even if we split these 2 series it will help our SOS -- and that would be enough to move up a couple of spots -- but not enough to make measuring it worthwhile.

Assuming we go 5-2 in the next 7 games, which is winning a midweek, 2-1 against Arky, and 2-1 in Baton Rouge(a tough goal -- but doable), this is how things will look.
Begin RPI: .5546
RPI W/L record in those games: 4.7-2
RPI W/L record overall: 18.4-13.6
RPI WP: .575
RPI win/loss effect: .1917

Improvement: .0112
New RPI: .5658
New current RPI position: 43rd
New position using final RPIs of last season(after regression of conference games for most teams): 24th

With the SOS factor included -- that would put us realistically around 35th currently and on pace to be in the top 20 at the end of the year...