This is really a cool model. Thanks for creating it. Based upon what I have been reading in your blog, it appears to be really close the model used by the state. How often are you updating it....for instance is it accurate through today?
I start updating it on Friday when the games in Utah finish up. I will then update again Saturday for out of state opponents who don't finish as early Friday nights, or for the occasional Saturday game. I've had to update it on Sunday the last few weeks because Kahuku will often play Saturday evenings.
That's the long way of saying it's updated through today. And with games happening on Wednesday, Thursday, and Friday of this week, I will try and update it at the end of each day.
Thanks for doing this. I'm not sure why UHSAA won't at least show the results as of Oct 9th when they stopped calculating. In fact, it really doesn't make any sense to not publish it continually. Maybe they feel that a team can try to improve their matchup in the playoffs but I think there are way too many pieces to "fix" a matchup with an intentional loss.
I'm still not sure I'm a fan of the formula. It has Provo (who I do think is one of the best teams they've had in a long time) at #2 in 5A even after losing to #11 Timpview and #12 Lehi.
It probably pays to not play up a division in the non-region games. At least for rankings. Maybe not for experience.
I imagine there will be quite a few "upsets" come playoff time.
I think playing up a division is worth it, because it will get you an extra 15% for your win, though I suppose it could cost you an easy win. The issue is playing down more than once -- and unfortunately UHSAA gives all out of state teams the same value (which is essentially the value of a 4A team). So Timpview playing Kahuku (who are ranked 110 in the country by MaxPreps), probably isn't worth it for the RPI, but Timpview playing Granger (6A, but ranked 8457 in the country) would be worth it for Timpviews WP.
There are some bugs they need to work out for sure. As much as I hate to say it, it might be better to give more strength to OWP and take some of the strength from WP.
I updated the dashboard, but I am waiting for it to refresh online. As much as I would like to put in the 3 games from tomorrow, I'm going to pass since there were a few upsets today.
Traditional RPI is split 25 percent modifier on the WP, 50 percent modifier on the OWP, 25 percent modifier on the OOWP (which is reflected in the dashboard up above), and the UHSAA model is 40/40/20. There is also an additional modifier of 1.4 for away games, no modifier for neutral games, and 0.6 for home games in the traditional RPI -- however that is not taken into account on the traditional modifier column in the dashboard above.
Excluding any typos (which I double checked, but I it's possible I may have miskeyed something), I am 100% confident in the WP and OWP columns, and 95% confident in the OOWP column. The OOWP column is occasionally off on teams with out of state opponents. The only reason I could find is some of the out of state opponents had opponents who possibly didn't record the result officially/correctly. When they are off, it rarely makes a difference. Only Monticello and East have been affected (negatively) in the past. East should still get a bye, and Monticello would play the same opponent, but they would be on the road.
Impressive!! U are the king of numbers!! Nice Hey 1 last question, with Bingham playing tomorrow and should win. How much will that change Weber High RPI since they played them. Bingham will go from 0.750000 to .770000 thanks can u run the number thanks
I saw that. The PG data was by far the most confusing. I double checked to make sure I didn't forget to give them a loss, but everything looked right. I will have to triple check later tonight. Unless I made a typo, I am confident in my rankings because I was able to test them throughout the season to the official numbers. I don't know how DesNews did theirs, but if they did it in the last week or two, they weren't able to check with the official numbers. But if they are right and I am wrong, I'll gladly ask them how they got it right and how I can make my model better in the future.
Thanks for all your help!! Desert News seems to be off. The only caveat is Kearns. They wer ranked #5 before this week . Don't u think they would be 4 or ?
This is really a cool model. Thanks for creating it. Based upon what I have been reading in your blog, it appears to be really close the model used by the state. How often are you updating it....for instance is it accurate through today?
ReplyDeleteI start updating it on Friday when the games in Utah finish up. I will then update again Saturday for out of state opponents who don't finish as early Friday nights, or for the occasional Saturday game. I've had to update it on Sunday the last few weeks because Kahuku will often play Saturday evenings.
DeleteThat's the long way of saying it's updated through today. And with games happening on Wednesday, Thursday, and Friday of this week, I will try and update it at the end of each day.
Thanks for doing this. I'm not sure why UHSAA won't at least show the results as of Oct 9th when they stopped calculating. In fact, it really doesn't make any sense to not publish it continually. Maybe they feel that a team can try to improve their matchup in the playoffs but I think there are way too many pieces to "fix" a matchup with an intentional loss.
ReplyDeleteI'm still not sure I'm a fan of the formula. It has Provo (who I do think is one of the best teams they've had in a long time) at #2 in 5A even after losing to #11 Timpview and #12 Lehi.
It probably pays to not play up a division in the non-region games. At least for rankings. Maybe not for experience.
I imagine there will be quite a few "upsets" come playoff time.
I think playing up a division is worth it, because it will get you an extra 15% for your win, though I suppose it could cost you an easy win. The issue is playing down more than once -- and unfortunately UHSAA gives all out of state teams the same value (which is essentially the value of a 4A team). So Timpview playing Kahuku (who are ranked 110 in the country by MaxPreps), probably isn't worth it for the RPI, but Timpview playing Granger (6A, but ranked 8457 in the country) would be worth it for Timpviews WP.
DeleteThere are some bugs they need to work out for sure. As much as I hate to say it, it might be better to give more strength to OWP and take some of the strength from WP.
Provo did Beat #1, #7, and #9 all on Road as well as losing to #11 and #12 both on road
DeleteThis changes a lot since LP lost their appeal.
ReplyDeleteCan u please do an update with LP record and how this all shakes out. Bingham has 1 game left against Copper Hill
ReplyDeletemight as well put a win.
Thanks:)
I updated the dashboard, but I am waiting for it to refresh online. As much as I would like to put in the 3 games from tomorrow, I'm going to pass since there were a few upsets today.
Deletewhat is the traditional RPI? How accurate do u feel this is 100%? thanks for doing all this
ReplyDeleteTraditional RPI is split 25 percent modifier on the WP, 50 percent modifier on the OWP, 25 percent modifier on the OOWP (which is reflected in the dashboard up above), and the UHSAA model is 40/40/20. There is also an additional modifier of 1.4 for away games, no modifier for neutral games, and 0.6 for home games in the traditional RPI -- however that is not taken into account on the traditional modifier column in the dashboard above.
DeleteExcluding any typos (which I double checked, but I it's possible I may have miskeyed something), I am 100% confident in the WP and OWP columns, and 95% confident in the OOWP column. The OOWP column is occasionally off on teams with out of state opponents. The only reason I could find is some of the out of state opponents had opponents who possibly didn't record the result officially/correctly. When they are off, it rarely makes a difference. Only Monticello and East have been affected (negatively) in the past. East should still get a bye, and Monticello would play the same opponent, but they would be on the road.
Impressive!! U are the king of numbers!! Nice Hey 1 last question, with Bingham playing tomorrow and
ReplyDeleteshould win. How much will that change Weber High RPI since they played them. Bingham will go from 0.750000
to .770000 thanks can u run the number thanks
Have U seen the projections from the desert news they have PG 14?
ReplyDeleteI saw that. The PG data was by far the most confusing. I double checked to make sure I didn't forget to give them a loss, but everything looked right. I will have to triple check later tonight. Unless I made a typo, I am confident in my rankings because I was able to test them throughout the season to the official numbers. I don't know how DesNews did theirs, but if they did it in the last week or two, they weren't able to check with the official numbers. But if they are right and I am wrong, I'll gladly ask them how they got it right and how I can make my model better in the future.
DeleteThanks for all your help!! Desert News seems to be off. The only caveat is Kearns. They wer ranked #5 before this week . Don't u think they would be 4 or ?
Deletecc .676
ReplyDeleteHerriman .603
bingham .589
kearns.572
skyridge .559
east.556
cuse .545
AF .543
roy .518
hunter .5317
weber .5135