So long, RPI

[TAMU, Knower of Ball]

I believe my preview for Presbyterian is tomorrow, but I think they are a solid add regardless. Not just a "not bad for July" add, but simply a good buy game.

I know you believe that, brew but I don't see it. They were the 334th best team in the country last season and lose their two best players, including their only true PG. They do play a favorable non-conference schedule and should be able to pick up a few wins there but I think they finish dead last in the Big South along with Longwood.

[WarriorDad]

will be interesting to see coaches and ADs reaction to the NET as it is not a verifiable measurement.

I'm guessing there will be sites that continue to do the RPI and a team or two will be left out this season that may have a strong RPI, that is when the reactions will come out even more.

[MUMonster03]

It said it considering using game dates, but decided not to. I think they should of considered dates. I do not think a win or loss in November should mean as much as a win or loss in February.

All games should be treated equal. So basically if you did this there would be no reason to have difficult games early in the season. So if we beat Kansas and Louisville/Tennessee in the NIT it shouldn't mean that much? Those early season wins can often be the difference for a bubble team.

Now if you beat that team because they had significant injuries that is something else, but if you try and say that results in February should mean more than November, you have no reason to schedule any difficult non conference games.

[brewcity77]

I know you believe that, brew but I don't see it. They were the 334th best team in the country last season and lose their two best players, including their only true PG. They do play a favorable non-conference schedule and should be able to pick up a few wins there but I think they finish dead last in the Big South along with Longwood.

You're wrong. Their point guard is returning. Bart Torvik was also wrong and has since upgraded them to around 289 since I pointed out that error (feel free to fact-check that).

And their best player was Francois Lewis, who is back. They plummeted last year after he was declared academically ineligible. Lewis is the type of 6'9" badass that tends to dominate lower leagues, but didn't get the chance last year.

(EDIT: To be clear, I just meant you're wrong specifically about Davon Bell. He is back.)

[mu03eng]

Does it? I honestly don't follow college football.

Yep, last year OSU got into the playoffs with a loss on the record that came early in the season versus an Alabama who had one right before the playoff field was determined.

[mu03eng]

I feel pretty confident one of the defining features of this new analytic is that they will be doing nested correlations and bilevel optimizations to evaluate whos playing well when as well as to get a more accurate accounting for stength of schedule of opponents of opponents, etc

Conspiracy mu03eng thinks part of the reason they aren't disclosing the analytic/formula is because it is a ML-based solution and they are going to have a couple of data scientists "optimizing" in real time.

Buckle up kids.

[TAMU, Knower of Ball]

You're wrong. Their point guard is returning. Bart Torvik was also wrong and has since upgraded them to around 289 since I pointed out that error (feel free to fact-check that).

And their best player was Francois Lewis, who is back. They plummeted last year after he was declared academically ineligible. Lewis is the type of 6'9" badass that tends to dominate lower leagues, but didn't get the chance last year.

(EDIT: To be clear, I just meant you're wrong specifically about Davon Bell. He is back.)

Huh, you are correct about Bell. I didn't realize he had another year of eligibility. That makes me feel a lot better about the game. I still don't think they are a good buy game if RPI was still a thing, but they wouldn't have been terrible....not that it matters now!

[Dr. Blackheart]

Huh, you are correct about Bell. I didn't realize he had another year of eligibility. That makes me feel a lot better about the game. I still don't think they are a good buy game if RPI was still a thing, but they wouldn't have been terrible....not that it matters now!

RPI no matta, hey.

[TAMU, Knower of Ball]

RPI no matta, hey.

I'll drink to that

[brewcity77]

Huh, you are correct about Bell. I didn't realize he had another year of eligibility. That makes me feel a lot better about the game. I still don't think they are a good buy game if RPI was still a thing, but they wouldn't have been terrible....not that it matters now!

Check out the preview I have up now. I'll follow up in there with some extra thoughts.

[Dr. Blackheart]

I'll drink to that

RPI was the pioneer into early use of statistics in basketball (hat tip).  However, why the NCAA held on to this dinosaur for so long must have been political.  It became the US News and World report college rating metric where schools tried to manipulate the outcome versus use it as an evaluator of the value of the quality of their play.

Pomeroy, while pedantic, has continually tried to improve his methods and quality of his metrics, on the other hand. The game is changing in subtle ways and these metrics need to move with the game.

#machinelearningmattas

[Marcus92]

Looming NCAA stupidity alert. Matt Norlander is reporting that the NCAA won't release the formula/foundation for the NET because it's based on ML/AI and isn't "readable".

No surprise there. The NCAA never released the exact formula for the RPI, either.

[Jay Bee]

No surprise there. The NCAA never released the exact formula for the RPI, either.

#FakeNews
#Lies

Yes they did. You could recalculate it.

[muwarrior69]

Why not let Watson decide?

https://www.ibm.com/blogs/business-analytics/mens-basketball-bracket/

[Nukem2]

#FakeNews
#Lies

Yes they did. You could recalculate it.

Yep, there were a number of sites that recalculated it every day and some in real time.

[bilsu]

Second, individual games don't keep teams out of a tournament. Its the entire resume.

Does anyone here want to argue that the loss at DePaul did not keep us out of the tournament?

[TAMU, Knower of Ball]

Does anyone here want to argue that the loss at DePaul did not keep us out of the tournament?

It didn't. Our entire resume which included the loss at DePaul kept us out of the tournament.

[Its DJOver]

Does anyone here want to argue that the loss at DePaul did not keep us out of the tournament?

It was a factor, but if I could change any "L" to a "W", it would not be that game.  The DePaul game wouldn't even make the list if I could change 8 "L's" to "W's".

[bilsu]

Up until a few years ago the NCAA did look at how you performed in the last 10 games. I am basically only talking about bubble teams fighting for the last bids. I think the team that is playing better in the last 10 games should get a bid over a team that got an upset win in November. Unless there is an injury or transfer there is no team in the country that is as good in November as they will be in March. However, all teams do not improve at the same rate.
Team A that is better than Team B in November may not better than Team B in March. Maybe that is the so called eye test.

[Its DJOver]

Up until a few years ago the NCAA did look at how you performed in the last 10 games. I am basically only talking about bubble teams fighting for the last bids. I think the team that is playing better in the last 10 games should get a bid over a team that got an upset win in November. Unless there is an injury or transfer there is no team in the country that is as good in November as they will be in March. However, all teams do not improve at the same rate.
Team A that is better than Team B in November may not better than Team B in March. Maybe that is the so called eye test.

The problem with that is that you can't control your conference scheduling.  Last year, our conference schedule was extremely front loaded.  5 of our first 9 games were against ranked opponents, then before we got to New York we didn't play a ranked team again so our "last 10 games" record was not an accurate reflection of our team.  Yes we improved, but as Nova proved in NY it clearly wasn't enough. 

[Jay Bee]

Up until a few years ago the NCAA did look at how you performed in the last 10 games. 

Not sure it was 10 games, but certain that either your definition of “a few years ago” is bonkers or you’re just wrong.

[MU Fan in Connecticut]

RPI RIP.

[MUDPT]

RPI was the pioneer into early use of statistics in basketball (hat tip).  However, why the NCAA held on to this dinosaur for so long must have been political.  It became the US News and World report college rating metric where schools tried to manipulate the outcome versus use it as an evaluator of the value of the quality of their play.

Pomeroy, while pedantic, has continually tried to improve his methods and quality of his metrics, on the other hand. The game is changing in subtle ways and these metrics need to move with the game.

#machinelearningmattas

I think Gasaway interviewed the RPI creators son a couple years back. The creator had passed away. Anyway, the guy said there is no way his father would have wanted it used now with better evaluation tools out there.

[Nukem2]

Not sure it was 10 games, but certain that either your definition of “a few years ago” is bonkers or you’re just wrong.

It was 10 games. In a quick google , I saw a quote from 2011 in which Gene Smith said the committe looked at the last ten games “a while ago”.  So it’s been a long time since the committee looked at that as a factor. 

[MUMonster03]

The problem with that is that you can't control your conference scheduling.  Last year, our conference schedule was extremely front loaded.  5 of our first 9 games were against ranked opponents, then before we got to New York we didn't play a ranked team again so our "last 10 games" record was not an accurate reflection of our team.  Yes we improved, but as Nova proved in NY it clearly wasn't enough.

This is why I think you need to look at the entire schedule

Team A: Big wins in November/December, play tough conference schedule at end goes 5-5.

Team B: No big out of conference wins, but goes 8-2 down stretch in conference against mid tier teams.

Is Team B really better then Team A just becuase they did better in final ten games? This is why bas losses and good wins matter so much, it is one of the few ways to compare two teams who play very limited (if any) shared opponents.