Dan
Admin
Illinois a 4 seed in the Bracket Matrix
http://www.bracketmatrix.com/
The Selection Show
Sunday, March 17th, 5:00pm CT, CBS
http://www.bracketmatrix.com/
The Selection Show
Sunday, March 17th, 5:00pm CT, CBS
If the committee's understanding is that McCullar and Dickinson won't play, I agree, they have historically been super harsh seeding teams that don't have the players that built their resume.Kansas still as a 4 seed does not pass the eye test IMHO. Just loss 2 games in row by a total of 50 points.
If the committee's understanding is that McCullar and Dickinson won't play, I agree, they have historically been super harsh seeding teams that don't have the players that built their resume.
Yep, we definitely need to make it to Sunday in order to be a #3 seed (IMO), but I really like our chances if we do. We got a LOT of help around the nation this week, and we need to hold up our end of the bargain ... beat two teams we have already beat! Maryland also slipped from #81 to #72, so by quitting the season a day early, they have cemented us as having a Quad 3 loss and one fewer Quad 1 win. However, we have these opportunities in front of us:Seems like after Duke's loss, we have a shot for the 3 seed...but I think we need to win the next 2.
The old bracketology thread had the NET rankings up there in the last couple posts. Looking at that, is there any world houston is not the number one overall seed?
Same record as Purdue. Best team in best conference in the country. Better quad 1. I’ve seen it talked about like Purdue is a lock for that spot, but houston is it to me.
You have to dig a little deeper than NET and Quad wins. Purdue has, arguably, a much more polished resume as a whole. Houston's best win is at home against Iowa State and @Baylor. Purdue had a three game stretch where they beat Gonzaga, Tennessee, and Marquette. They later beat (then top ranked) Arizona. Without a doubt, Purdue has much higher quality wins in the Quad 1A field. That likely gets Purdue the top overall seed.The old bracketology thread had the NET rankings up there in the last couple posts. Looking at that, is there any world houston is not the number one overall seed?
Same record as Purdue. Best team in best conference in the country. Better quad 1. I’ve seen it talked about like Purdue is a lock for that spot, but houston is it to me.
But Purdue's losses were @OSU (#53 in NET), @Neb (#37), and @NW (#52), while Houston's losses were @Kansas (#19), @TCU (#42), @ISU (#9)You have to dig a little deeper than NET and Quad wins. Purdue has, arguably, a much more polished resume as a whole. Houston's best win is at home against Iowa State and @Baylor. Purdue had a three game stretch where they beat Gonzaga, Tennessee, and Marquette. They later beat (then top ranked) Arizona. Without a doubt, Purdue has much higher quality wins in the Quad 1A field. That likely gets Purdue the top overall seed.
You have to dig a little deeper than NET and Quad wins. Purdue has, arguably, a much more polished resume as a whole. Houston's best win is at home against Iowa State and @Baylor. Purdue had a three game stretch where they beat Gonzaga, Tennessee, and Marquette. They later beat (then top ranked) Arizona. Without a doubt, Purdue has much higher quality wins in the Quad 1A field. That likely gets Purdue the top overall seed.
Honestly, if bracket was released today, the three losses for Purdue, Houston, and UCONN are all a pretty much a wash (all Quad 1 road games). The key differentiator is the extent of top tier wins. This is where Purdue holds the edge over Houston and UCONN.But Purdue's losses were @OSU (#53 in NET), @Neb (#37), and @NW (#52), while Houston's losses were @Kansas (#19), @TCU (#42), @ISU (#9)
NET, Coaches, AP, and Pomeroy all have Houston #1. Massey Composite might be diluted by inferior systems, but FWIW, it also has them #1. When the committee goes against such information, perhaps they're taking into account things that a computer can't, or perhaps the computers are taking into account things that the humans can't. I'm more inclined to think it's the latter.
Honestly, if bracket was released today, the three losses for Purdue, Houston, and UCONN are all a pretty much a wash (all Quad 1 road games). The key differentiator is the extent of top tier wins. This is where Purdue holds the edge over Houston and UCONN.
Purdue top wins: Arizona (NET 4), Tennessee (NET 5), Alabama (NET 8), Marquette (NET 13), Illini twice (NET 16), Gonzaga (NET 17), Wisconsin twice (NET 21), and Michigan State now twice (NET 24). That is 11 wins total against teams in the top 25 NET! That is very indicative of the top team in the nation for seeding.
Again, that is as of today at 2PM CT. However, I would be surprised if Purdue did not get top overall seed if they at least make it to the BTT Final.
This.
The Big 12 hit the NET jackpot this year and that's helped inflate Houston's resume a bit after a fairly pedestrian non-conference schedule.
Houston is a fantastic team and deserving of a 1 seed but their wins, while of similar Quad variety, don't stack up to Purdue's.
I just think this sort of analysis is better handled by computers than humans trying to boil it down to a neat "resume". A computer can take into account a team's performance in all their games, not just the best wins or worst losses.Honestly, if bracket was released today, the three losses for Purdue, Houston, and UCONN are all a pretty much a wash (all Quad 1 road games). The key differentiator is the extent of top tier wins. This is where Purdue holds the edge over Houston and UCONN.
Purdue top wins: Arizona (NET 4), Tennessee (NET 5), Alabama (NET 8), Marquette (NET 13), Illini twice (NET 16), Gonzaga (NET 17), Wisconsin twice (NET 21), and Michigan State now twice (NET 24). That is 11 wins total against teams in the top 25 NET! That is very indicative of the top team in the nation for seeding.
Again, that is as of today at 2PM CT. However, I would be surprised if Purdue did not get top overall seed if they at least make it to the BTT Final.
I just think this sort of analysis is better handled by computers than humans trying to boil it down to a neat "resume". A computer can take into account a team's performance in all their games, not just the best wins or worst losses.
The betting market can do even better by taking into account extra information through a wisdom of the crowds, but there isn't really a practical way to use that to determine seeds.
It's amazing how much our culture has changed since the early days of the BCS in college football.I just think this sort of analysis is better handled by computers than humans
Kenpom (and .NET) style analysis do not work on football. There is not enough data.It's amazing how much our culture has changed since the early days of the BCS in college football.
Agree that football is different due to small sample size and even fewer non-conference games.Kenpom (and .NET) style analysis do not work on football. There is not enough data.
I'd love to see the .NET recalculated two different ways:
1) Cap margins of victory at 15 pts. A big win is a big win. All we learn is which coached are jerks and push for blowouts vs. those who play subs and walkons.
2) Re-evaluate the at-large teams dropping all results against teams ranked 250+
Running up efficiency numbers against weak teams causes the conference to be rated higher. Once conference play starts, as you observed, it becomes a closed loop. I want to see how the conferences rate without the noise in the data from playing clearly overmatched teams.Agree that football is different due to small sample size and even fewer non-conference games.
Scoring margin is no longer a direct input (though it still impacts net efficiency). When it was an input, it was capped at 10 points.
Beating really bad teams doesn't do much to help anyone's ranking directly. Perhaps - since so many results are closed loops within conference - it could have an additional hidden impact on how conferences are rated relative to each other.
Net efficiency is adjusted for strength of opponent and location, so it doesn't help a team to beat a very weak team unless they do even better than "expected". It also increases the out-of-conference sample size, so I doubt removing them would make the rankings more accurate.Running up efficiency numbers against weak teams causes the conference to be rated higher. Once conference play starts, as you observed, it becomes a closed loop. I want to see how the conferences rate without the noise in the data from playing clearly overmatched teams.
Yes, I agree there isn't a perfect algorithm. That's why I said the betting markets can do better, but not by much when compared to an average of ranking systems (link). Their main edge is taking into account information that's new and/or isn't generally included in computer ranking systems, particularly injuries and roster changes.Computers are algorithms made by humans.
Betting markets are definitely a bunch of humans.
I'm not against objective measures of performance, but college bball will always have apples to oranges comparisons for schedules, wins, etc.. as long as there are 350+ teams. There simply isn't a perfect algorithm for capturing those differences. Even the existing ones come to different answers.