|
Post by trollhunter on Feb 15, 2018 14:13:44 GMT -5
Please let me know results of your validation. I'm guessing that both methods should come up with same result, if modifications are done uniformly. It is the average of the opponents w/l% after modifications for #2. It is clearly different than the sum of the modified wins and losses - but usually not by much. Playing a team like Texas who has a good record, but doesn't play a lot of matches is a better opponent using average compared to sum.
Thanks for explaining the slight difference, that makes sense. Your work is much appreciated!
|
|
bluepenquin
Hall of Fame
4-Time VolleyTalk Poster of the Year (2019, 2018, 2017, 2016), All-VolleyTalk 1st Team (2021, 2020, 2019, 2018, 2017, 2016)
Posts: 12,381
|
Post by bluepenquin on Feb 22, 2018 8:47:58 GMT -5
I have now completed all validations - which will make next year's RPI Futures more accurate. Here is the precise RPI Calculation:
1) W/L% * 25% 2) Opponent W/L% * 50%. Opponent W/L% is the average of all opponent w/l% excluding all games not involving the named team. Play a team 3 times, all 3 matches are excluded from the opponent's wins and losses. 3) Opponent's Opponent W/L% * 25%. This is the average of each Opponent's Opponent W/L%. This is 'simply' averaging all opponent's #2 calculation from above, no other modifications.
Bonus Points: .0026 scheduling bonus for any team that as 50% or more matches out of conference against teams ranked in the Top 75 in unadjusted RPI. .0013 bonus points for a win against a team in the Top 50 in unadjusted RPI and an additional .0013 points if defeating a team in the Top 25.
Penalty Points: Penalty Points mirrors the Bonus Points in reverse, .0026 penalty if out of conference schedule has 50% or more against teams ranked 260 or worse in unadjusted RPI. .0013 penalty points for losses against teams ranked worse than 285 in unadjusted RPI and an additional .0013 for losses against teams ranked 310 or worse. Finally, I think there is an additional penalty for losses against non D1 teams, but I don't really care about this. I had 3 SWAC teams that had a lower final RPI than my calculation related to 'extra' penalties - the only 3 teams where my RPI calc was different than the NCAA.
|
|
|
Post by trollhunter on Feb 22, 2018 17:57:50 GMT -5
I have now completed all validations - which will make next year's RPI Futures more accurate. Here is the precise RPI Calculation: 1) W/L% * 25% 2) Opponent W/L% * 50%. Opponent W/L% is the average of all opponent w/l% excluding all games not involving the named team. Play a team 3 times, all 3 matches are excluded from the opponent's wins and losses. 3) Opponent's Opponent W/L% * 25%. This is the average of each Opponent's Opponent W/L%. This is 'simply' averaging all opponent's #2 calculation from above, no other modifications. Bonus Points: .0026 scheduling bonus for any team that as 50% or more matches out of conference against teams ranked in the Top 75 in unadjusted RPI. .0013 bonus points for a win against a team in the Top 50 in unadjusted RPI and an additional .0013 points if defeating a team in the Top 25. Penalty Points: Penalty Points mirrors the Bonus Points in reverse, .0026 penalty if out of conference schedule has 50% or more against teams ranked 260 or worse in unadjusted RPI. .0013 penalty points for losses against teams ranked worse than 285 in unadjusted RPI and an additional .0013 for losses against teams ranked 310 or worse. Finally, I think there is an additional penalty for losses against non D1 teams, but I don't really care about this. I had 3 SWAC teams that had a lower final RPI than my calculation related to 'extra' penalties - the only 3 teams where my RPI calc was different than the NCAA. There is a penalty for losing to non-D1 teams, but you are correct to ignore it - it will not occur in top 75 RPI teams. You can also ignore the entire Penalty Points section. Nobody in RPI top 100 will lose to a 285+ team, nor schedule a majority of RPI 260+. They should have just removed that section instead of making it ridiculous. It actually came up occasionally in the past when penalties were around the 150-175 mark (midpoint).
|
|