OT: The demon that is RPI

[80shusky]

Posters often complain about RPI being a flawed statistic and clearly it is but like or not it is a demon we have to live with. There have been many posts about what if this happened and what if that happened. So, just for kicks I was messing around with RPI wizard today which some of you have referenced and it is amazing what relatively small tweaks do to your RPI.
Our current RPI is 57 according to RPI wizard.
Our worst and I believe most avoidable losses this year were Temple and Cincinnati at home. If you change those two games to wins we are 23-8 with an RPI of 29 and solidly in the field.
If you change either of those games, either Temple or Cincinnati to win and make them a split, we are 22-9 with a 37 RPI and in much better shape than we are now. So just one more win against one of those teams makes a huge difference.
Now I got creative (for me anyway). I said what about those five cupcakes we played early? I removed Sacred Heart and Umass Lowell from our schedule. They were RPI 261 and 260 respectively. Instead I looked for and found a couple of local "mid-majors" we are all familiar with. I substituted them with Siena at RPI 104 and URI at RPI 106. Now (assuming those were wins) our RPI calculated at 40 and a 44 SOS. Again, an improvement.
So what does the team do going forward? Should we drop two cupcakes and look for some teams like Siena and URI that aren't great teams but usually don't carry 250+ RPIs I don't think? In our area there are several of those teams that tend to be 150+ right like Yale, Albany, Manhattan, St Joes, etc.
Anyway, I realize many of you are probably saying "um, duh!" but I didn't realize how pretty small changes make pretty big differences.

 

[zls44]

Stop scheduling sub-300 teams everyone knows will be sub-300 teams like UML and CCSU.

 

[Robertelamin]

stop losing home games to bubble teams.

 

[Shabizness]

stop losing home games to bubble teams.

Stop losing.

 

[BoatShow]

With Howie gone there's no reason to play CCSU anymore. If you must play an instate team Quinnipiac or Fairfield usually have much better RPIs. Someone once mentioned scheduling UML may be a nod to Hockey East. Let's try and get PC or ND instead.

 

[Mr. Wonderful]

Scheduling America East and Northeast Conference games are relatively inexpensive "buy" games, but they rape your RPI without lube. Substitute MAC or A10 teams instead, which are more expensive but are winnable RPI boosters.

 

[80shusky]

@Mr. Wonderful exactly! With USF, UCF, E Carolina and Tulane we will have plenty of RPI vampires in conference. Let's take more risks early instead by scheduling 150+ minimum teams if we can. We can't control their rankings ultimately but we can at least schedule teams that are better historically.

 

[Jerry1714]

The RPI is not nearly as important as it use to be. Colorado St was left out with an RPI 29 last year but their Kenpom was 68 and their BPI was 57.

Committee chair Scott Barnes said last year the RPI will help pool which teams deserves to be in consideration together, but it's not a determining factor for who gets in.

The committee last year definitely gave more value to Kenpom and BPI rankings over RPI.

 

[BFieldHusky]

@Jerry1714. With that thought in mind...UConn is actually one spot ahead of Cincy in Kenpom going in tonight (30 v.s. 31). Let's say UConn lost on Friday, and their KenPom dropped to 34/35...what was the lowest rated team left out last year from Kenpom?

 

[Jerry1714]

@Jerry1714. With that thought in mind...UConn is actually one spot ahead of Cincy in Kenpom going in tonight (30 v.s. 31). Let's say UConn lost on Friday, and their KenPom dropped to 34/35...what was the lowest rated team left out last year from Kenpom?

I believe it was Florida at 46th

 

[upstater]

Posters often complain about RPI being a flawed statistic and clearly it is but like or not it is a demon we have to live with. There have been many posts about what if this happened and what if that happened. So, just for kicks I was messing around with RPI wizard today which some of you have referenced and it is amazing what relatively small tweaks do to your RPI.
Our current RPI is 57 according to RPI wizard.
Our worst and I believe most avoidable losses this year were Temple and Cincinnati at home. If you change those two games to wins we are 23-8 with an RPI of 29 and solidly in the field.
If you change either of those games, either Temple or Cincinnati to win and make them a split, we are 22-9 with a 37 RPI and in much better shape than we are now. So just one more win against one of those teams makes a huge difference.
Now I got creative (for me anyway). I said what about those five cupcakes we played early? I removed Sacred Heart and Umass Lowell from our schedule. They were RPI 261 and 260 respectively. Instead I looked for and found a couple of local "mid-majors" we are all familiar with. I substituted them with Siena at RPI 104 and URI at RPI 106. Now (assuming those were wins) our RPI calculated at 40 and a 44 SOS. Again, an improvement.
So what does the team do going forward? Should we drop two cupcakes and look for some teams like Siena and URI that aren't great teams but usually don't carry 250+ RPIs I don't think? In our area there are several of those teams that tend to be 150+ right like Yale, Albany, Manhattan, St Joes, etc.
Anyway, I realize many of you are probably saying "um, duh!" but I didn't realize how pretty small changes make pretty big differences.

URI isn't going to give UConn free games. They'll ask for a 2 and 1 at least, so they are out of the equation.

We've obviously played Yale, Albany in the past. Yale we lost to. So it didn't help that year to schedule them, but it should help in the future. St. Joe's? No, they're not coming to Storrs for free.

 

[upstater]

Scheduling America East and Northeast Conference games are relatively inexpensive "buy" games, but they rape your RPI without lube. Substitute MAC or A10 teams instead, which are more expensive but are winnable RPI boosters.

You have to win those games though. It doesn't help your RPI at all if you lose them. Put those into the machine.

 

[BFieldHusky]

I believe it was Florida at 46th

Something to keep in mind this week. I don't think we make it if we lose on Friday, but we would be about 10 spots ahead of last year's cut off team then.

 

[Mr. Wonderful]

You have to win those games though. It doesn't help your RPI at all if you lose them. Put those into the machine.

Playing the America East doesn't help your RPI, win or lose. In fact, I see no good reason to ever play an America East team if you're in the AAC.

 

[tcf15]

Something to keep in mind this week. I don't think we make it if we lose on Friday, but we would be about 10 spots ahead of last year's cut off team then.

Along with that, the teams that were ranked 40+ per-tourney in KenPom that received at-large bids:

Dayton (25-8)
KP: 40
RPI: 32
11 seed

UCLA (20-13)
KP: 41
RPI: 49
11 seed

St. John's (21-11)
KP: 42
RPI: 36
9 seed

LSU (22-10)
KP: 43
RPI: 54
9 seed

Mississippi (20-12)
KP: 44
RPI: 57
11 seed

Oregon (25-9)
KP: 46
RPI: 26
8 seed

Purdue (21-12)
KP: 49
RPI: 53
9 seed

Indiana (20-13)
KP: 53
RPI: 58
10 seed

 

[uconnfan68]

Stop losing.

Even that isn't foolproof. UConn women are #4, and no one is close to them in any way, shape, or form.

 

[BLUEDOGHOUSE]

Posters often complain about RPI being a flawed statistic and clearly it is but like or not it is a demon we have to live with. There have been many posts about what if this happened and what if that happened. So, just for kicks I was messing around with RPI wizard today which some of you have referenced and it is amazing what relatively small tweaks do to your RPI.
Our current RPI is 57 according to RPI wizard.
Our worst and I believe most avoidable losses this year were Temple and Cincinnati at home. If you change those two games to wins we are 23-8 with an RPI of 29 and solidly in the field.
If you change either of those games, either Temple or Cincinnati to win and make them a split, we are 22-9 with a 37 RPI and in much better shape than we are now. So just one more win against one of those teams makes a huge difference.
Now I got creative (for me anyway). I said what about those five cupcakes we played early? I removed Sacred Heart and Umass Lowell from our schedule. They were RPI 261 and 260 respectively. Instead I looked for and found a couple of local "mid-majors" we are all familiar with. I substituted them with Siena at RPI 104 and URI at RPI 106. Now (assuming those were wins) our RPI calculated at 40 and a 44 SOS. Again, an improvement.
So what does the team do going forward? Should we drop two cupcakes and look for some teams like Siena and URI that aren't great teams but usually don't carry 250+ RPIs I don't think? In our area there are several of those teams that tend to be 150+ right like Yale, Albany, Manhattan, St Joes, etc.
Anyway, I realize many of you are probably saying "um, duh!" but I didn't realize how pretty small changes make pretty big differences.

Well URI will be the pre season favorite to win the A-10 next year. They should be a lock to make the tournament assuming Mathews comes back healthy.

 

[upstater]

Playing the America East doesn't help your RPI, win or lose. In fact, I see no good reason to ever play an America East team if you're in the AAC.

Again, put it into the machine. I bet you that beating an Am East team is better than losing to a MAC team.

Also, there is an America East team this year coached by a former UConn Husky. They have an RPI in the 60s. We have scheduled them in the past.

 

[Mr. Wonderful]

Again, put it into the machine. I bet you that beating an Am East team is better than losing to a MAC team.

Also, there is an America East team this year coached by a former UConn Husky. They have an RPI in the 60s. We have scheduled them in the past.

I know this is your M.O. (redefining and then framing someone else's arguments along your own narrow parameters), but I'm not gonna let you do that here.

Asking to compare a win to a loss is irrelevant.

Sure there is a better chance of losing, but the reward of winning, IMO, outweighs the added risk of losing by a wide margin. Our existence in the AAC changes the way we need to schedule. The lower half of the AAC drags down our RPI so severely compared to the way things were in the Big East that every potential non-conference opponent needs to be scrutinized with an eye towards the chances that their RPI will be 250+. If the chances are good, then they can't be scheduled. Our margin for error here is very small, as the total lack of respect shown our league by the NCAA during tournament selection time has proven.

 

[Palatine]

Stop scheduling sub-300 teams everyone knows will be sub-300 teams like UML and CCSU.

That's not the problem. It doesn't help but it's not what hurts us.

Don't blow a 12 point lead against Temple. Play better down the stretch of games. That's what matters.

 

[tzznandrew]

That's not the problem. It doesn't help but it's not what hurts us.

Don't blow a 12 point lead against Temple. Play better down the stretch of games. That's what matters.

Beat Temple (both times!) and Cincy at home.

Voilà! 24-7 (14-4), RPI 24, 9-7 vs. RPI Top 100. Conference champs. Probably a 5 or 6 seed, maybe higher depending on conference tournament performance.

 

[Mr. Wonderful]

That's not the problem. It doesn't help but it's not what hurts us.

Don't blow a 12 point lead against Temple. Play better down the stretch of games. That's what matters.

The committee uses both RPI and SOS as part of a matrix of indicators when evaluating at-large bids. Although not an exact science, there is a high degree of predictability when it comes to where on the spectrum you're going to fall with these metrics based on who is on your schedule.

Everyone knows playing strong teams helps. What seems to be hard to convey to some is that yes, math can prove beyond a shadow of a doubt that the weak side of your schedule matters quite a bit.

So what is hard for me to grasp is that 1) you know you will be judged by the criteria of RPI and SOS, among other opponent based metrics, and 2) you can maximize your return on those metrics by using some basic predictive math, then why would you nor do it?

 

[tcf15]

That's not the problem. It doesn't help but it's not what hurts us.

Don't blow a 12 point lead against Temple. Play better down the stretch of games. That's what matters.

Scheduling 300+ teams definitely hurt us.

UConn's current RPI with Maine and CCSU as 300+ RPI teams is 57.

Replace those two with teams ranked 199 and 201 in the RPI and our RPI jumps to 41.

Taking it a step further, if you remove UMass-Lowell (RPI 260) and Sacred Heart (RPI 261) and replace with teams ranked 210 and 212 and UConn's RPI jumps to 35.

Just removing the four 250+ teams this year does wonders for UConn's RPI.

 

[pepband99]

yeah, @tcf15 and i are on the same wavelength. Just posted about CCSU, which is a cinder block tied to our necks.

 

[tzznandrew]

Not true at all. Scheduling 300+ teams does hurt us.

UConn's current RPI with Maine and CCSU as 300+ RPI teams is 57.

Replace those two with teams ranked 199 and 201 in the RPI and our RPI jumps to 41.

Taking it a step further, if you remove UMass-Lowell (RPI 260) and Sacred Heart (RPI 261) and replace with teams ranked 210 and 212 and UConn's RPI jumps to 35.

Going from 57 to 35 just by removing the 250+ teams does wonders for your RPI.

This is true. Still, this is less a problem if they won the games they should have.

If you do what I said in my previous post, and pair it with yours, our RPI jumps from 24 to 16 and our SOS is 28. Looking like a 3 or 4 seed.

That said, some of those are hard to nail. 212 was a CUSA team. While CCSU was easily a bad team, sometimes what you think will offset that (Georgetown, say) doesn't help as much as you think. Replace Georgetown with Seton Hall (who we thought Georgetown would be) and we jump 8 spots!