Stats are broken again?

Anyone notice the huge difference between Wichita State's RPI and KenPom rating? KenPom loves these guys at 20 (they were 21 yesterday afternoon), while RPI has them at 70. I'm curious how that is happening.

Indiana's another one, with KP liking them at 39, 80 in RPI.

I looked at Lunardi's bracket...it just makes everything worse. Richmond, Indiana (at a 9!), Michigan State...in.

Nothing says these guys have to agree. It would be pretty disappointing if they were all the same, as if one guy did the work and the others cribbed off of him, wouldn't it? Different opinions are good.

Another one to look at is Bracket Matrix.

KenPom takes into account more than just wins and losses. Wichita has been beating the crap out of opponents which helps them in a lot of the computer rankings outside of the RPI. Wichita is hurt in the RPI because the couple of good shots we had at "quality" wins were early in the year and we have a very young team that had lost its on court leadership. We struggled. I'm not sure if the leadership void has been completely filled but it's been better and the play has as well. Unfortunately, the MVC as a whole isn't strong so it's hard to say just how good we are by watching. We are much better than we were in Atlantis but nobody knows how much so since the past few weeks our average margin of victory is around 25 or so. We should know a little more Saturday when we face the only other team in the conference this year that could actually be a problem if they get the chance come March.

I had noticed something similar when I looked at the ACC teams, since Lunardi had so many of them in his bracket. So I looked at all the ACC teams using five different rankings: RPI, BPI, KenPom, Massey Composite and Sagarin.

The top 6 ACC teams' rankings were mostly consistent, with a couple of exceptions. KP has Virginia at #2, but their RPI is only #13. Duke's RPI is #20. but their BPI is a hard to understand #6.

But for the teams that would be seeded lower than #8, there were some real head scratchers. Pitt's RPI is 56, while KP has them at 86. That may be because he can take into account that the Panthers are reeling right now. Miami's RPI is only 63, while everyone else sees them as 30-ish. Syracuse is probably out of the running anyway, but not as bad as their #99 RPI would suggest. KenPom, Sagarin and Massey all put them between 43-60.

I suppose these rankings will tighten up by the end of the regular season, but it's clear that the selection committee has already realized the RPI is flawed. They recently announced they will be reducing the weight they give this in the future. Whether they could ever find a different metric everybody would agree with is hard to say.

(01-30-2017 01:20 PM)Wedge Wrote:  Nothing says these guys have to agree. It would be pretty disappointing if they were all the same, as if one guy did the work and the others cribbed off of him, wouldn't it? Different opinions are good.

Another one to look at is Bracket Matrix.

Oh, no complaints about the differences, just that it's not that common to see such a gap between RPI and KP numbers.

Looking at Bracket Matrix, it looks like Lunardi's power source.

Interesting to see how Wichita doesn't even make it into half of the brackets Matrix assesses. Conversely, Indiana shows up in all but six...with lower RPI and KP numbers against WSU.

Quote:KenPom takes into account more than just wins and losses. Wichita has been beating the crap out of opponents which helps them in a lot of the computer rankings outside of the RPI. Wichita is hurt in the RPI because the couple of good shots we had at "quality" wins were early in the year and we have a very young team that had lost its on court leadership. We struggled. I'm not sure if the leadership void has been completely filled but it's been better and the play has as well. Unfortunately, the MVC as a whole isn't strong so it's hard to say just how good we are by watching. We are much better than we were in Atlantis but nobody knows how much so since the past few weeks our average margin of victory is around 25 or so. We should know a little more Saturday when we face the only other team in the conference this year that could actually be a problem if they get the chance come March.

RPI will reward a tough schedule, but, that's relative when you think you're scheduling something competitive, to see it flop. WSU's SOS is not so great in that regard. But, winning those games can't hurt. RPI, if it punishes anyone, it's those who lose games they shouldn't, especially when they're at home. It looks like KenPom puts a lot of it together. Margin, location, SOS, number of victories against average number of victories, that kind of stuff...

It's weird, though...UNI had a top-10 RPI figure before even playing five games, because they got Xavier twice, and Xavier was a metric monster. UNI, if they somehow pull it together and win the MVC tournament...I suspect the committee would overlook the garbage record and bad numbers because those guys can play when they're healthy.

Yeah, Bracket Matrix shows where the consensus is: They compile 79 different bracket projections. The top 36 teams listed are in at least 75 of those 79 brackets.

Farther down in the field, there is no consensus. 37 bracket projections have Wichita in the tournament; the other 42 don't. 38 have Miami in; 41 have the Canes out. Wake Forest is "in" on 40 of the 79 brackets.

1 thing that is deceiving about Bracket Matrix is that it's taking into account brackets done for an entire week.

So like take TCU for instance. they are in 66/79 brackets on the matrix.....
BUT they are only in 5/14 that were done yesterday. or as many as Georgetown.

(01-30-2017 02:08 PM)The Cutter of Bish Wrote:  

(01-30-2017 01:20 PM)Wedge Wrote:  Nothing says these guys have to agree. It would be pretty disappointing if they were all the same, as if one guy did the work and the others cribbed off of him, wouldn't it? Different opinions are good.

Another one to look at is Bracket Matrix.

Oh, no complaints about the differences, just that it's not that common to see such a gap between RPI and KP numbers.

Looking at Bracket Matrix, it looks like Lunardi's power source.

Interesting to see how Wichita doesn't even make it into half of the brackets Matrix assesses. Conversely, Indiana shows up in all but six...with lower RPI and KP numbers against WSU.

Quote:KenPom takes into account more than just wins and losses. Wichita has been beating the crap out of opponents which helps them in a lot of the computer rankings outside of the RPI. Wichita is hurt in the RPI because the couple of good shots we had at "quality" wins were early in the year and we have a very young team that had lost its on court leadership. We struggled. I'm not sure if the leadership void has been completely filled but it's been better and the play has as well. Unfortunately, the MVC as a whole isn't strong so it's hard to say just how good we are by watching. We are much better than we were in Atlantis but nobody knows how much so since the past few weeks our average margin of victory is around 25 or so. We should know a little more Saturday when we face the only other team in the conference this year that could actually be a problem if they get the chance come March.

RPI will reward a tough schedule, but, that's relative when you think you're scheduling something competitive, to see it flop. WSU's SOS is not so great in that regard. But, winning those games can't hurt. RPI, if it punishes anyone, it's those who lose games they shouldn't, especially when they're at home. It looks like KenPom puts a lot of it together. Margin, location, SOS, number of victories against average number of victories, that kind of stuff...

It's weird, though...UNI had a top-10 RPI figure before even playing five games, because they got Xavier twice, and Xavier was a metric monster. UNI, if they somehow pull it together and win the MVC tournament...I suspect the committee would overlook the garbage record and bad numbers because those guys can play when they're healthy.

Our schedule fell apart this year. Long Beach has been much less than expected. South Dakota State has had their worst year since moving to D1, or close to it. Oklahoma fell off much more than anticipated. Saint Louis has taken a nose dive the last couple years. Tulsa is young. Then we fell just short in early games against Michigan State and Louisville. I think we've been favored in every game this year. That's all had a negative impact on our RPI. I do believe we're much better than the RPI indicates but I'm not sure we're quite as good right now as the likes of KenPom say. Though just about all of the other computer rankings have us in the top 30. I do think we should be if we were a little more consistent.

I do believe the advanced metrics of the different computer rankings are probable more accurate than RPI but there's no perfect system.

(01-30-2017 02:28 PM)SubGod22 Wrote:  Then we fell just short in early games against Michigan State...

Yuck. That will be the trap. Because, MSU is not a strong team this year, and very inconsistent. If you lob MSU out of the tournament, saying they weren't good enough, it's a bad call if you're Wichita and you fall just shy of the outright auto bid and onto that bubble.

The schedule...it seems like it gets some sympathy here, none there. It's that head-to-head versus potential tournament teams where it really cooks. And you really want Michigan State to turn it on now and not look back, because it will, in turn, help Wichita's numbers profoundly.

BPI is an ESPN creation with a special sauce secret formula... ESPN leans on the scales with its Bilas Pressure Index, which probably involves assisting their most-favored college hoops properties.

(01-30-2017 12:48 PM)The Cutter of Bish Wrote:  Stats are broken again?

Anyone notice the huge difference between Wichita State's RPI and KenPom rating? KenPom loves these guys at 20 (they were 21 yesterday afternoon), while RPI has them at 70. I'm curious how that is happening.

Indiana's another one, with KP liking them at 39, 80 in RPI.

I looked at Lunardi's bracket...it just makes everything worse. Richmond, Indiana (at a 9!), Michigan State...in.

Makes sense, television will do whatever they can to make sure Izzo unjustly makes the tournament.

(01-30-2017 02:48 PM)The Cutter of Bish Wrote:  

(01-30-2017 02:28 PM)SubGod22 Wrote:  Then we fell just short in early games against Michigan State...

Yuck. That will be the trap. Because, MSU is not a strong team this year, and very inconsistent. If you lob MSU out of the tournament, saying they weren't good enough, it's a bad call if you're Wichita and you fall just shy of the outright auto bid and onto that bubble.

The schedule...it seems like it gets some sympathy here, none there. It's that head-to-head versus potential tournament teams where it really cooks. And you really want Michigan State to turn it on now and not look back, because it will, in turn, help Wichita's numbers profoundly.

That is one of the concerns for us. Our resume' is definitely lacking. Our team I feel is more than good enough. This is the problem of playing in a conference with very few chances for quality wins and one of the reasons Wichita has been looking to move. Our whole post season hinges on our noncon performance. Last year we had injuries early in the season, including our PG Fred VanVleet as well as a few others and this year we were trying to replace the leadership provided by our two NBA guards and others. It made for a tough go.

I won't feel comfortable about getting into the NCAA unless we win the conference tournament. Any chances at an at-large do depend on how others do, including MSU, and us winning on Saturday against Illinois State. We don't have what would be considered a "bad" loss, but we're lacking in "good" wins. I know the committee isn't supposed to spend much time on the advanced metrics this year, but as a Shocker fan, I have to hope they do.

(01-30-2017 04:45 PM)SubGod22 Wrote:  I know the committee isn't supposed to spend much time on the advanced metrics this year, but as a Shocker fan, I have to hope they do.

They won't need advanced metrics to look at how the Shockers did against tournament teams. If it's a "0 for" number, the season kind of ends (well, they'll probably get a top NIT spot)...unless they win all the way on out until the MVC tournament final, and amass something like a 29-5 record. That probably puts them in a play-in at least.

What's garbage is that if they lose a few more games, but still win the MVC AQ, you wonder where they seed? THAT'S where the bias lies, I feel. Their numbers are either decent or they aren't. By RPI, which is valued to an extent, these guys are going up against 5 and 4 seeds as 12 and 13-seeds. Is the committee REALLY going to burden those seeds with a data oddity like Wichita, who we know can play up in a big way when it counts? They could win MVC...I'd wager they'd stick them in the 7-10 range, those death seeds when they're getting the 1 or 2 the next round. But, again...how do you make that call when the progs don't like them that much?

(01-30-2017 04:59 PM)The Cutter of Bish Wrote:  

(01-30-2017 04:45 PM)SubGod22 Wrote:  I know the committee isn't supposed to spend much time on the advanced metrics this year, but as a Shocker fan, I have to hope they do.

They won't need advanced metrics to look at how the Shockers did against tournament teams. If it's a "0 for" number, the season kind of ends (well, they'll probably get a top NIT spot)...unless they win all the way on out until the MVC tournament final, and amass something like a 29-5 record. That probably puts them in a play-in at least.

What's garbage is that if they lose a few more games, but still win the MVC AQ, you wonder where they seed? THAT'S where the bias lies, I feel. Their numbers are either decent or they aren't. By RPI, which is valued to an extent, these guys are going up against 5 and 4 seeds as 12 and 13-seeds. Is the committee REALLY going to burden those seeds with a data oddity like Wichita, who we know can play up in a big way when it counts? They could win MVC...I'd wager they'd stick them in the 7-10 range, those death seeds when they're getting the 1 or 2 the next round. But, again...how do you make that call when the progs don't like them that much?

I know our wins against tournament teams is and will be lacking, I was just going off of the fact that we're a top 30 team on those other computer numbers and may end up top 20 on a couple. Hell, we somehow got votes in both polls this week. Though that means next to nothing as we've seen a top 25 team or two left out of the tournament in the past.

But you're right, the best chance of sneaking in is to win out and at least make it to the championship game against Illinois State. Even then, I wouldn't be comfortable. We'd be 29-5 if we did that and I think the previous calculations I've seen others run would give us an RPI in the 40s which would at least appear more respectable. But we'd still be lacking those wins you referenced but would also still have no bad losses. It will be interesting. If we do that and get in we would probably be in the play in game again. I'd take it if it meant we got a shot. If we win out and win the tournament we'd probably end up around an 11. All we need is in though and we can hold our own. I think we were an 8 or 9 the year we went to the Final Four.

On the bright side, we have no seniors and this team should be scary good next year. They could still make some noise this year if they get a chance.

the problem with Wichita is they have absolutely beaten no one. They have 3 wins this year vs winning teams. Tulsa, Colorado St, and Loyola-Chicago. 2/3 of their wins vs teams in the RPI 151 or worse. That's putrid. If they don't beat Illinois St, I don't see much of a chance.

(01-30-2017 05:26 PM)stever20 Wrote:  the problem with Wichita is they have absolutely beaten no one. They have 3 wins this year vs winning teams. Tulsa, Colorado St, and Loyola-Chicago. 2/3 of their wins vs teams in the RPI 151 or worse. That's putrid. If they don't beat Illinois St, I don't see much of a chance.

Yet KenPom has them at #20, Sagarin at 18, Massey Composite at 31, the AP poll 29 and the coaches poll 35. If they continue to win, and given their recent tournament history, I would expect them to get in as about a #9-10 seed.

(01-30-2017 05:33 PM)ken d Wrote:  

(01-30-2017 05:26 PM)stever20 Wrote:  the problem with Wichita is they have absolutely beaten no one. They have 3 wins this year vs winning teams. Tulsa, Colorado St, and Loyola-Chicago. 2/3 of their wins vs teams in the RPI 151 or worse. That's putrid. If they don't beat Illinois St, I don't see much of a chance.

Yet KenPom has them at #20, Sagarin at 18, Massey Composite at 31, the AP poll 29 and the coaches poll 35. If they continue to win, and given their recent tournament history, I would expect them to get in as about a #9-10 seed.

and last year Wichita was #12 in Ken Pom and they barely made the tourney as the next to last team in the tourney.

Recent tourney history does not mean anything. NOTHING .

(01-30-2017 12:48 PM)The Cutter of Bish Wrote:  Stats are broken again?

Anyone notice the huge difference between Wichita State's RPI and KenPom rating? KenPom loves these guys at 20 (they were 21 yesterday afternoon), while RPI has them at 70. I'm curious how that is happening.

Indiana's another one, with KP liking them at 39, 80 in RPI.

I looked at Lunardi's bracket...it just makes everything worse. Richmond, Indiana (at a 9!), Michigan State...in.

is Duke a 2 seed or do they have to lose a couple of more times for that to happen?

The final formula will be whatever one screws the little guys the most.

(01-31-2017 01:01 AM)HarmonOliphantOberlanderDevine Wrote:  The final formula will be whatever one screws the little guys the most.

This year's cast: UNCW, MTSU, Akron, UNR, and making a return visit...Valpo!

I saw the UNCW-W&M game this weekend. The Tribe beat those guys like a drum. Only uglier thing on that floor than the game were UNCW's uniforms.

I don't envy UNCW were they to get the AQ...they'd line up in that 7-10 seed pool with a nice weekend treat after the first game.