2ndGenBeaver
Sophomore
Posts: 1,837
Grad Year: 1991 (MS/CS) 1999 (PhD/CS)
|
Post by 2ndGenBeaver on Jan 28, 2019 22:12:16 GMT -8
Interesting to take a gander at RPI at this moment.... Basically every top-25 team is ranked pretty much in line with RPI (or within 2-3 spots of it as you get into the teens and twenties) - with one exception. Us. RPI of 41. Ranking of 9ish. Is this evidence that we have made lasting impact on the hearts and minds of the voting public? Does RPI just lag reality considerably? Is this just an anomaly this season? Or is there some other explanation for this? Enquiring minds want to know. Go Beavers!
|
|
|
Post by bvrbooster on Jan 28, 2019 22:30:11 GMT -8
This has always been a mystery to me; I have no idea how they figure these things. Obviously, it can be quite hurtful come tournament time - recall last year we wound up as a 6 seed, and had to go to Tennessee.
But, it is fixable by winning. Beat Utah, beat Stanford, then see where they have us.
|
|
|
Post by blackbuttebeaver80 on Jan 28, 2019 22:46:29 GMT -8
Interesting to take a gander at RPI at this moment.... Basically every top-25 team is ranked pretty much in line with RPI (or within 2-3 spots of it as you get into the teens and twenties) - with one exception. Us. RPI of 41. Ranking of 9ish. Is this evidence that we have made lasting impact on the hearts and minds of the voting public? Does RPI just lag reality considerably? Is this just an anomaly this season? Or is there some other explanation for this? Enquiring minds want to know. Go Beavers! RPI is measured by a combination of our win/loss percentage, our opponent’s win/loss percentage, and our opponents’ opponent’s win loss percentage. So, most of the RPI is a reflection of strength of schedule only measured by our oponent’s success and their opponent’s success. Once we’ve scheduled the games the only thing we can control is winning the games on the schedule. The polls allow for much more subjectivity which leads to much higher ratings for us after voters in the polls actually see us play.
|
|
|
Post by bennyskid on Jan 29, 2019 9:03:01 GMT -8
Only the NCAA takes RPI seriously. If the NCAA didn't make it a thing, no one but no one would pay any attention to it. KenPom and Massey are infinitely better.
As for the reason our RPI stinks: Too many of our early season opponents turned out to be particularly lousy this year. Who knew that LaSalle would be 3-18 or Cal-Poly 4-13? Replace LaSalle with DuQuesne and Cal-Poly with Cal-Irvine and Ark-Pine Bluff with Ark-Little Rock and our RPI would probably be in the top 20. That's problem #1 with the RPI - it places way too much emphasis on meaningless tune-up games.
The good news is that our RPI will improve steadily as we run through the conference schedule. We need it to get into the 20's if we aren't going to get bowbed by the committee again for hosting.
|
|
|
Post by beaverstever on Jan 29, 2019 9:06:25 GMT -8
It's a complete joke - the RPI needs completely trashed. OSU's SOS is #26 per Massey. So world does Rutgers, with a worse record, lost to worst teams, and worse SOS somehow ranked #14 and OSU is #41?? www.ncaa.com/rankings/basketball-women/d1/ncaa-womens-basketball-rpiwww.masseyratings.com/cbw/ncaa-d1/ratingsIn general Massy aligns with the polls pretty well with the human polls, but still has OSU at #11 instead of #9 and ahead of Utah. That probably does make sense as the human polls trust OSU more than Utah, despite their better record. But the RPI is just off, and not just on OSU. For instance, Iowa State is at #11 in RPI, where the human polls have them outside of the top 20 ... that's a big difference as well. Or that there are 6 teams better than UConn ... yea,sure
|
|
|
Post by Henry Skrimshander on Jan 29, 2019 11:43:45 GMT -8
We have played Cal Poly (314) and UA-Pine Bluff (344) and Bakersfield (284) and La Salle (286). They are RPI killers.
|
|
|
Post by rmancarl on Jan 29, 2019 21:44:29 GMT -8
|
|
|
Post by beaveragain on Jan 29, 2019 23:01:32 GMT -8
warrennolan has an RPI sort of rating called ELO ranking. They ignore everything except DI games. And on that ranking without the garbage games OSU is ranked 10th.
|
|
|
Post by bennyskid on Jan 30, 2019 7:25:30 GMT -8
warrennolan has an RPI sort of rating called ELO ranking. They ignore everything except DI games. And on that ranking without the garbage games OSU is ranked 10th. Those garbage games are DI games. Basketball only has three divisions - DI, DII, DIII. DIII is non-scholarship. DII is limited to 10 scholarship equivalents (most scholarships are partial). Western Oregon and Northwest Nazarene are DII. The only DII schools we ever play is in that "exhibition" game we're allowed to play before the real season starts.
The Elo ranking is a system based on the system used in chess. Honestly, it's not a great formula for college basketball - it is very abstract and is best for sports with lots and lots of games and no extraneous factors like margin of victory or home-court advantage. In other words, chess.
(And it's Elo, not ELO. As in Arpad Elo, the physicist who devised it.)
|
|
|
Post by jefframp on Jan 30, 2019 14:34:05 GMT -8
warrennolan has an RPI sort of rating called ELO ranking. They ignore everything except DI games. And on that ranking without the garbage games OSU is ranked 10th. Those garbage games are DI games. Basketball only has three divisions - DI, DII, DIII. DIII is non-scholarship. DII is limited to 10 scholarship equivalents (most scholarships are partial). Western Oregon and Northwest Nazarene are DII. The only DII schools we ever play is in that "exhibition" game we're allowed to play before the real season starts.
The Elo ranking is a system based on the system used in chess. Honestly, it's not a great formula for college basketball - it is very abstract and is best for sports with lots and lots of games and no extraneous factors like margin of victory or home-court advantage. In other words, chess.
(And it's Elo, not ELO. As in Arpad Elo, the physicist who devised it.)
Sounds like a case of Quantum Entanglement to me.
|
|
2ndGenBeaver
Sophomore
Posts: 1,837
Grad Year: 1991 (MS/CS) 1999 (PhD/CS)
|
Post by 2ndGenBeaver on Jan 30, 2019 14:48:49 GMT -8
Those garbage games are DI games. Basketball only has three divisions - DI, DII, DIII. DIII is non-scholarship. DII is limited to 10 scholarship equivalents (most scholarships are partial). Western Oregon and Northwest Nazarene are DII. The only DII schools we ever play is in that "exhibition" game we're allowed to play before the real season starts.
The Elo ranking is a system based on the system used in chess. Honestly, it's not a great formula for college basketball - it is very abstract and is best for sports with lots and lots of games and no extraneous factors like margin of victory or home-court advantage. In other words, chess.
(And it's Elo, not ELO. As in Arpad Elo, the physicist who devised it.)
Sounds like a case of Quantum Entanglement to me. Ah - thanks for the clarification, I thought this ELO ranking system might be a living thing..... Go Beavers!
|
|
|
Post by beaveragain on Jan 30, 2019 16:51:08 GMT -8
"The Elo ranking is a system based on the system used in chess. Honestly, it's not a great formula for college basketball - it is very abstract and is best for sports with lots and lots of games and no extraneous factors like margin of victory or home-court advantage. In other words, chess."
Yes, it's best used in games with a large sample size, such as basketball with 350 teams competing. And it is used in soccer, football, baseball as well as basketball. Margin of victory is ignored in chess and rightfully so I think and I hope they do so in their version. Home-court advantage is an easily tacked on fudge factor.
"(And it's Elo, not ELO. As in Arpad Elo, the physicist who devised it.)" I used the name that the site used so as to not be confusing.
|
|
|
Post by bennyskid on Jan 30, 2019 19:10:06 GMT -8
I was a chess Tournament Director a very long time ago, and I have a MS in Applied Math. In my youth I calculated Elo formulas by hand. The Elo formula is no more sophisticated than the RPI - it is nothing more than this:
If you win, you get so many points. (Typically 16) If you lose, you lose the same number. (Typically 16) And, regardless of whether you win or lose, you gain some smaller number of points in proportion to the difference in the two team's ratings. (Typically 0 to 14)
That's it. It's barely different than the RPI. It works in chess because in chess, ratings can be based on all the games in a player's career - not just the current season - and the ratings of active players can be based on hundreds or even thousands of games. In college sports, you can't really carry ratings over from season to season, so ratings are based on a fraction as many games. In chess, players aren't even given Elo ratings until they have at least 25 games recorded.
(The number of teams is completely irrelevant. It's the number of games that matters.) It's used in lots of places that it shouldn't for one reason: it's stupidly simple. You don't have to know anything about the sport and it can be implemented in a few lines of code. And, frankly, it sounds sophisticated - it was invented by a physicist!
|
|
|
Post by Werebeaver on Jan 30, 2019 19:44:07 GMT -8
I was a chess Tournament Director a very long time ago, and I have a MS in Applied Math. In my youth I calculated Elo formulas by hand. The Elo formula is no more sophisticated than the RPI - it is nothing more than this: If you win, you get so many points. (Typically 16) If you lose, you lose the same number. (Typically 16) And, regardless of whether you win or lose, you gain some smaller number of points in proportion to the difference in the two team's ratings. (Typically 0 to 14) That's it. It's barely different than the RPI. It works in chess because in chess, ratings can be based on all the games in a player's career - not just the current season - and the ratings of active players can be based on hundreds or even thousands of games. In college sports, you can't really carry ratings over from season to season, so ratings are based on a fraction as many games. In chess, players aren't even given Elo ratings until they have at least 25 games recorded. (The number of teams is completely irrelevant. It's the number of games that matters.) It's used in lots of places that it shouldn't for one reason: it's stupidly simple. You don't have to know anything about the sport and it can be implemented in a few lines of code. And, frankly, it sounds sophisticated - it was invented by a physicist! Does it work for checkers?
|
|