Bracketology

[waterhead]

TeamRankings simulates all remaining games and uses an algorithm to generate an S-curve. Currently they have us finishing at 26-5, good for #6 on the S-curve and a 2 seed. They also have Wichita as a 3, and SMU and Houston both as 9 seeds. They give UC an 86% chance to get a 4 seed or better.

There is no way in hell we are getting a 2 seed with 5 losses. Not happening. I am not sure it will happen with 4 losses. Maybe 3. This is probably strictly a stats based projection which is fine...I understand.

The good thing is the committee will look at column 1 and 2 wins and they like wins away from home. If we beat WSU at home we will only have 1 column 1 home win and the rest will be away from home. That's an advantage.

If the committee is doing their homework they will also know we don't have any true home games. I think our home games should be put somewhere between home and neutral...even if closer to home. But I doubt we get that sort of benefit.

I think we need 4 losses or less to ensure a protected seed. I think 5 puts us on the outside.

 

[sedziobs]

The good thing is the committee will look at column 1 and 2 wins and they like wins away from home. If we beat WSU at home we will only have 1 column 1 home win and the rest will be away from home. That's an advantage.

The new group format accounts for home/away already. I don't think we'll earn any bonus points for away Group 1 wins.

We're one of only six teams in the country with at least four Group 1 wins and no losses outside of Group 1.

 

[kskenyon4]

Bracketmatrix has Cincy as the third #5 seed. THOUGH, as you all know, it takes time to update across the board, and it currently has us behind: ASU, UK, Seton Hall, etc that we clearly are a step ahead of as of today.

I think if today was selection sunday, we're a 4 seed.

If we manage to go 17-1 in AAC play, I think we're in heavy discussion for a 2 seed. If we go to the AAC final, we have 16 games remaining. 29-5 would be 4 seed IMO, anything better, hard to not give us a 2/3 seed.

For what its worth, 32-2 would be a 1 seed.

 

[waterhead]

The new group format accounts for home/away already. I don't think we'll earn any bonus points for away Group 1 wins.

We're one of only six teams in the country with at least four Group 1 wins and no losses outside of Group 1.

Sure but projections are we will lose at least two of our group 1 wins by end of season because they will fall in RPI. I love the fact we haven't lost at home or had a bad loss in over two years. Gessuz...that is good stuff!!!

 

[waterhead]

Bracketmatrix has Cincy as the third #5 seed. THOUGH, as you all know, it takes time to update across the board, and it currently has us behind: ASU, UK, Seton Hall, etc that we clearly are a step ahead of as of today.

I think if today was selection sunday, we're a 4 seed.

If we manage to go 17-1 in AAC play, I think we're in heavy discussion for a 2 seed. If we go to the AAC final, we have 16 games remaining. 29-5 would be 4 seed IMO, anything better, hard to not give us a 2/3 seed.

For what its worth, 32-2 would be a 1 seed.

I think if we beat ECU we should be in the 4 seed range with the rest of the losses going on this week. From there I think we can stay in the 4 seed range with 2 more losses and move up with only 1 more loss or move down with 3 more losses.

I think our tourney seed is going to be very very very close to our total number of regular season losses.

 

[sedziobs]

Sure but projections are we will lose at least two of our group 1 wins by end of season because they will fall in RPI.

This is why I don't like deterministic projections. One of those teams is UCF, who Warren Nolan projects at 16-14. Kenpom has them at 18-12. The other is Temple, who Warren Nolan has at 13-17. Kenpom has them at 15-15. Those two extra wins for each could allow them to finish top 75.

 

[waterhead]

This is why I don't like deterministic projections. One of those teams is UCF, who Warren Nolan projects at 16-14. Kenpom has them at 18-12. The other is Temple, who Warren Nolan has at 13-17. Kenpom has them at 15-15. Those two extra wins for each could allow them to finish top 75.

I agree with you. UCF is a wild card with Taylor back. Who knows what Temple is going to do. The stat prediction is they won't be there...but there is a lot more to it than stats.

 

[waterhead]

I agree with you. UCF is a wild card with Taylor back. Who knows what Temple is going to do. The stat prediction is they won't be there...but there is a lot more to it than stats.

I think we have to start with the stats based evidence and then determine as fans who we think aren't well represented by them. UCF getting Taylor back is a huge factor. SMU losing Foster would be another huge factor.

 

[sedziobs]

The stat prediction is they won't be there...

What I'm trying to say is there are different stat predictions, and Kenpom is superior. It's fair to say that one stat prediction has two of our current Group 1 wins falling out, but not stat projections in general. BJ Taylor's return isn't factored into either one.

 

[waterhead]

What I'm trying to say is there are different stat predictions, and Kenpom is superior. It's fair to say that one stat prediction has two of our current Group 1 wins falling out, but not stat projections in general. BJ Taylor's return isn't factored into either one.

Stat projections are just that. They are predictive but not reality.

Obviously things like Taylor returning is not factored in. Neither is a Jarrey Foster or Gilbert getting injured. So we move a team up or down a bit. We really have no clue other than predictive stats and then giving each team a bump or a minus from there based on what we as fans know.

In the end I don't think the stats based predictions will be TOO far off. One team can do better and another could do worse.

 

[waterhead]

Stat projections are just that. They are predictive but not reality.

Obviously things like Taylor returning is not factored in. Neither is a Jarrey Foster or Gilbert getting injured. So we move a team up or down a bit. We really have no clue other than predictive stats and then giving each team a bump or a minus from there based on what we as fans know.

In the end I don't think the stats based predictions will be TOO far off. One team can do better and another could do worse.

When you are into conference play this is about all you have. One team does better and the other does worse. I just hope we get the teams we play twice doing better and the teams we play once doing worse. I hope one winning doesn't mean the other falls out of a column 1 or 2...or vice versa.

 

[sedziobs]

In the end I don't think the stats based predictions will be TOO far off. One team can do better and another could do worse.

Yes, but you are using the inferior prediction to make your point. I think we're better off just using current RPI than assuming the favorite will win every single game for the rest of the season.

 

[sedziobs]

As an example, look at this past Saturday's slate of 134 games. Kenpom predicted the favorites to go 95-39. They actually went 93-41. That's a 2% error. Warren Nolan has the favorites going 134-0, for a 44% error. It just doesn't make much sense to use projections this early in the season with that kind of error rate.

 

[waterhead]

Yes, but you are using the inferior prediction to make your point. I think we're better off just using current RPI than assuming the favorite will win every single game for the rest of the season.

Inferior prediction? WTF is that? And who said we should assume the favorite will win every single game for the rest of the season? I remember saying it won't likely go anything like that. But it is the best predictor unless you have another.

Current RPI is bullshit in some cases. Current Kenpom or future RPI is much much better. If you want to use current RPI as the predictor you are on your own...it's not going to end like that...sorry

 

[waterhead]

As an example, look at this past Saturday's slate of 134 games. Kenpom predicted the favorites to go 95-39. They actually went 93-41. That's a 2% error. Warren Nolan has the favorites going 134-0, for a 44% error. It just doesn't make much sense to use projections this early in the season with that kind of error rate.

So which projection do you feel we should use? Yours? WE have to start with something but we can at least make adjustments from a starting point. There has to be a starting point or it's just a free for all.

 

[sedziobs]

If you want to use current RPI as the predictor you are on your own...it's not going to end like that...sorry

I think you'll be on your own with Warren Nolan predicted RPI. I don't know any bracketologist who uses that. Lots of them use current RPI.

 

[waterhead]

As an example, look at this past Saturday's slate of 134 games. Kenpom predicted the favorites to go 95-39. They actually went 93-41. That's a 2% error. Warren Nolan has the favorites going 134-0, for a 44% error. It just doesn't make much sense to use projections this early in the season with that kind of error rate.

That same error rate will be there in the last week. it's called college BB. As of right now we can only use predictive stats along with some good old fashioned fan intuition.

 

[sedziobs]

So which projection do you feel we should use? Yours? WE have to start with something but we can at least make adjustments from a starting point. There has to be a starting point or it's just a free for all.

I don't have a projection, so I don't know what that means. There doesn't have to be starting point. Bracketmatrix is crowdsourced predictions, which is more likely to outperform any individual prediction at this point. If you're talking about end of season RPI prediction, I don't think there is a good one. If there was, bracketology would be a piece of cake. I'm simply telling you, with evidence, that Warren Nolan's is bad.

 

[waterhead]

I think you'll be on your own with Warren Nolan predicted RPI. I don't know any bracketologist who uses that. Lots of them use current RPI.

It's roughly the same. He has a current and a future projected RPI. His current RPi is about exactly the same as any other current RPi formula.

I am looking at future RPI based on predicted wins. I know they are not going to be exactly accurate but it's a place to start The predicted results may or may not happen...which is why they are called predicted.

 

[sedziobs]

That same error rate will be there in the last week. it's called college BB. As of right now we can only use predictive stats along with some good old fashioned fan intuition.

And even in the last week predicted RPI will not be much, if any, better than the actual RPI at the time. I don't know why you think we have to use predictive stats. Most bracketologists have been using current RPIs for ages. The ones that do predict results use probabilistic Monte Carlo simulations, like TeamRankings.