Let me preface this by saying that this is purely an academic exercise.
I personally find KP to be the most reliable ranking system. It has its warts to be sure, but I find it paints a much more accurate picture than any of the other formulas.
Marquette's KP rank of 32 is significantly higher than its RPI of 70, which is the biggest factor putting a damper on our tournament aspirations.
If we lived in a world where tournament berths were not assigned by a committee but instead by the KenPom the following things would currently be true:
We would be an 8 seed in the tournament along with Miami (FL), Minnesota, and Dayton (ABD!).
No team currently behind us in bracketmatrix would be ahead of us (the closest would be Wake Forest at 33 who is currently first 4 out).
The following teams which are considered to be ahead of us (per bracketmatrix), would all be behind us:
Northwestern: KP 36, currently 10 seed
Xavier: KP 39, currently 8 seed
Virginia Tech: KP 41, currently 8 seed
Arkansas: KP 42, currently 9 seed
Maryland: KP 43, currently 7 seed
VCU: KP 45, currently 9 seed
Michigan State: KP 50, currently 10 seed
Providence, KP 55, currently 11 seed
Seton Hall, KP 57, currently 10 seed
USC: KP 67, currently 10 seed
Just an interesting perspective, purely academic. This is not the world we live in, nor do I think it should be. Even though I like KP, it is not perfect. I like that a committee picks the bids because there are some things that go beyond numbers. Though I am glad to hear that the committee is supposedly moving away from RPI. We have known that it has been outdated for years and they seem to be the last ones to have figured it out.
Wins above bubble is a metric I've been following this year. It uses kenpom and others, but also rewards teams for finishing games and winning (gulp).
I see no way this team should be an 8th seed, so I do not think KP is any more accurate. What is wrong with these systems is they do not take into account how teams change. For example I am not sure what Indiana's RPI is, but I am sure if they had not played their opening game against Kansas it would be significantly worse. They beat Kansas in that game and now they would have very little chance of beating Kansas, but their numbers are still benifiting from that win.
Quote from: MUDPT on February 28, 2017, 06:54:05 AM
Wins above bubble is a metric I've been following this year. It uses kenpom and others, but also rewards teams for finishing games and winning (gulp).
Yup, gulp indeed. WAB
Quote from: bilsu on February 28, 2017, 07:02:00 AM
I see no way this team should be an 8th seed, so I do not think KP is any more accurate. What is wrong with these systems is they do not take into account how teams change. For example I am not sure what Indiana's RPI is, but I am sure if they had not played their opening game against Kansas it would be significantly worse. They beat Kansas in that game and now they would have very little chance of beating Kansas, but their numbers are still benifiting from that win.
I think that's a good thing. It places equal value on the whole season, not just how they are performing lately.
Quote from: bilsu on February 28, 2017, 07:02:00 AM
I see no way this team should be an 8th seed, so I do not think KP is any more accurate. What is wrong with these systems is they do not take into account how teams change. For example I am not sure what Indiana's RPI is, but I am sure if they had not played their opening game against Kansas it would be significantly worse. They beat Kansas in that game and now they would have very little chance of beating Kansas, but their numbers are still benifiting from that win.
Not really how RPI works.
Quote from: Dr. Vinnie Boombatz on February 28, 2017, 07:12:34 AM
I think that's a good thing. It places equal value on the whole season, not just how they are performing lately.
Personally, I don't think that's a good way to judge teams. Xavier is a great example of the flaw in that mindset this season. Teams change considerably over the course of a season, and the most recent performances should be considered more relevant in a ranking system.
If Team A wins a Thanksgiving tournament, then finishes the conferences schedule on a 6 game losing streak, which set of games do you think is more indicative of how they'll play in the tournament?
Quote from: skianth16 on February 28, 2017, 09:55:01 AM
Personally, I don't think that's a good way to judge teams. Xavier is a great example of the flaw in that mindset this season. Teams change considerably over the course of a season, and the most recent performances should be considered more relevant in a ranking system.
If Team A wins a Thanksgiving tournament, then finishes the conferences schedule on a 6 game losing streak, which set of games do you think is more indicative of how they'll play in the tournament?
The tournament isn't about "who is playing better now." The tournament is about "who had the better overall season."
So your example of Team A above...they may lose in the first round of the tournament because they are playing poorly, but that doesn't negate the fact that they deserve to be there due to their Thanksgiving tourney victory. Game 1 counts just as much as Game 31.
Quote from: Dr. Vinnie Boombatz on February 28, 2017, 10:03:50 AM
The tournament isn't about "who is playing better now." The tournament is about "who had the better overall season."
So your example of Team A above...they may lose in the first round of the tournament because they are playing poorly, but that doesn't negate the fact that they deserve to be there due to their Thanksgiving tourney victory. Game 1 counts just as much as Game 31.
Agreed....there are really two different things: entry and seeding. Entry should be based on the season as a whole, seeding is determined on some sort of guidance around current performance.
Quote from: mu03eng on February 28, 2017, 10:14:47 AM
Agreed....there are really two different things: entry and seeding. Entry should be based on the season as a whole, seeding is determined on some sort of guidance around current performance.
OK, that's fair. I can agree with that concept for the most part.
I still have trouble with a team like Xavier, who is going to have some really high points on the resume, but comes to the tournament as a very different team than they were in November. I'm not 100% sold that these teams should always get the nod. In a weak bubble year like this year, sure, that's easier. But in a more competitive season, I think maybe they played their way out down the stretch.
crean sucks
Looks like they are considering going away from the RPI.
http://www.sbnation.com/college-basketball/2017/1/13/14264948/ncaa-tournament-selection-committee-process-rpi-advanced-stats
https://fivethirtyeight.com/features/the-ncaa-is-modernizing-the-way-it-picks-march-madness-teams/
Quote from: skianth16 on February 28, 2017, 09:55:01 AM
Personally, I don't think that's a good way to judge teams. Xavier is a great example of the flaw in that mindset this season. Teams change considerably over the course of a season, and the most recent performances should be considered more relevant in a ranking system.
If Team A wins a Thanksgiving tournament, then finishes the conferences schedule on a 6 game losing streak, which set of games do you think is more indicative of how they'll play in the tournament?
I'd actually love to see an analysis of this. How often does the "hot team" that everyone picks for the Elite Eight flame out in their first game? How often does the team everyone thinks is dead in the water suddenly catch fire and make a second weekend run?
My guess is those recent runs of good play are no more indicative of NCAA success than those Thanksgiving tournaments.
Quote from: brewcity77 on February 28, 2017, 05:32:16 PM
I'd actually love to see an analysis of this. How often does the "hot team" that everyone picks for the Elite Eight flame out in their first game? How often does the team everyone thinks is dead in the water suddenly catch fire and make a second weekend run?
My guess is those recent runs of good play are no more indicative of NCAA success than those Thanksgiving tournaments.
You don't have to go further than last year's Syracuse team to see an example of the latter. Lost 5 of 6 heading into the tournament and ended up in the Final Four.
Quote from: Jay Bee on February 28, 2017, 07:19:50 AM
Not really how RPI works.
If it does not work that way, then why play a tough schedule?
Because just playing Kansas on a neutral court is good for the RPI.....and it gives you an opportunity for a big win and if you lose it doesn't hurt at all.
A win over Kansas and a loss to Tennessee on a neutral court would be the same as a loss to Kansas and a win over Tennessee on a neutral court as far as the RPI is concerned.
At least that is my understanding.....JB can correct me if I'm wrong.
Quote from: Dr. Vinnie Boombatz on February 28, 2017, 05:43:01 PM
You don't have to go further than last year's Syracuse team to see an example of the latter. Lost 5 of 6 heading into the tournament and ended up in the Final Four.
Seton Hall was the flip side of this.
Quote from: skianth16 on February 28, 2017, 09:55:01 AM
Personally, I don't think that's a good way to judge teams. Xavier is a great example of the flaw in that mindset this season. Teams change considerably over the course of a season, and the most recent performances should be considered more relevant in a ranking system.
If Team A wins a Thanksgiving tournament, then finishes the conferences schedule on a 6 game losing streak, which set of games do you think is more indicative of how they'll play in the tournament?
You're seriously saying that a team that "finishes the conferences schedule on a 6 game losing streak" is going to make the tournament. If they did, they would have had to have been in the top ten the entire season before that losing streak.
And UConn's run through the BEast and NCAA tourneys a few years ago confirms that it's not just entry (does the team deserve to get in based on what it has done over the whole season) or seeding (how does the team stack up against other teams that are in the bracket) but also performance (how is the team trending).
I know they've moved away from the results of the last 10 games, but I find that to be a useful measure. Not perfect as a predictor, but teams that are trending well can ride that groove. If I recall, Izzo is pretty explicit about his commitment to peaking at tournament time, even at the cost of earlier performance.
Quote from: bilsu on February 28, 2017, 05:43:07 PM
If it does not work that way, then why play a tough schedule?
I think Jay Bee would argue that simply playing Kansas impacts Indianas RPI more so than the fact that they beat Kansas. That is to say, Indiana's RPI if they lost to Kansas would be better than it would be if they hadn't played them at all.
If anything it probably reinforces your point.
KPom's strength is it is stable. Its weakness is it is not situational.
Quote from: bilsu on February 28, 2017, 05:43:07 PM
If it does not work that way, then why play a tough schedule?
It's a reason TO play a tough schedule. Sit back, analyze the calcs, & you'll better understand
You credited the WIN v KU as a special WIN relative to their RPI ranking, but that's not how RPI works.
Clue: you could flip almost any loss on the schedule to a win, but in exchange assume they lost to KU... the result is their RPI would be essentially flat as compared to what it is now
A couple thoughts/notes.
1) I believe the KenPom ratings include a recency factor. Seem to remember reading something about that in his blog at the beginning of the season.
2) Analytics are a big step forward in how we evaluate sports performance. But a system based primarily on points per possession excludes the most important result of any matchup: who actually won the game.
I know KenPom accounts for this to some degree with the Luck factor. But I'm not sure it gives winning or losing enough weight/credit.
Imagine a hypothetical team in an average conference with an average strength of schedule. The team loses every single game by 1 point (say 0-30). If I understand the system at all, KenPom would rate this team somewhere near the middle of the pack — say in the 160 range — with a Luck rating approaching -0.999. But by definition, this isn't an average team; it's the worst team in the country. They couldn't win a game, which is ultimately what counts.
This isn't a defense of RPI. I find KenPom far more informative about a team's strengths and weaknesses on both ends of the court. At the same time, it's not the Unified Theory of college basketball.
Quote from: Jay Bee on March 01, 2017, 03:40:31 AM
It's a reason TO play a tough schedule. Sit back, analyze the calcs, & you'll better understand
You credited the WIN v KU as a special WIN relative to their RPI ranking, but that's not how RPI works.
Clue: you could flip almost any loss on the schedule to a win, but in exchange assume they lost to KU... the result is their RPI would be essentially flat as compared to what it is now
Makes me wonder why Marquette is never willing to play blue bloods on the road. Fu** do a 2 or 3 for 1 with Kansas, or Arizona, or UNC, or hell...Duke. Why not. Losses aren't going to hurt you, and just playing the games is going to immensely help your RPI. I think this season has shown you that as long as you can get to .500 or better in a top 3 or 4 conference, losing 4 games in the non conference isn't a back breaker.
Quote from: JamilJaeJamailJrJuan on March 01, 2017, 09:50:17 AM
Makes me wonder why Marquette is never willing to play blue bloods on the road. Fu** do a 2 or 3 for 1 with Kansas, or Arizona, or UNC, or hell...Duke. Why not. Losses aren't going to hurt you, and just playing the games is going to immensely help your RPI. I think this season has shown you that as long as you can get to .500 or better in a top 3 or 4 conference, losing 4 games in the non conference isn't a back breaker.
Because every additional blue-blood road game is one less game that's played at home. IOW, scheduling for RPI can't be done at the expense of game revenue.
Quote from: Benny B on March 01, 2017, 09:58:55 AM
Because every additional blue-blood road game is one less game that's played at home. IOW, scheduling for RPI can't be done at the expense of game revenue.
Well seeing as though we've burned a non-con game 2 of the past 3 seasons, this doesn't hold much water.
Quote from: JamilJaeJamailJrJuan on March 01, 2017, 10:08:50 AM
Well seeing as though we've burned a non-con game 2 of the past 3 seasons, this doesn't hold much water.
This isn't an all-or-none proposition. Burning one non-con game is much, much different than burning 3 or 4. In fact, it's 3 or 4 times the difference.
Quote from: Benny B on March 01, 2017, 10:16:23 AM
This isn't an all-or-none proposition. Burning one non-con game is much, much different than burning 3 or 4. In fact, it's 3 or 4 times the difference.
Well they've burn 1 in two of the past three seasons. If they did a 2 for 1 with a blue blood, it'd be the same amount of homes games lost, but guess what, that 1 on the backend is going to drum up far more interest for the program than Southwest Idaho A&M State.
Maybe this scenario is not super realistic, but just wish MU scheduled better. That's it. I see other teams do it, so it is certainly possible.
Quote from: Marcus92 on March 01, 2017, 09:42:26 AM
A couple thoughts/notes.
1) I believe the KenPom ratings include a recency factor. Seem to remember reading something about that in his blog at the beginning of the season.
I didn't realize this was the case but you are right, in a blog post earlier this year he mentioned that the recency factor has been tuned to be even smaller than it used to be. Does anyone have a link to how that impacts the rankings? I couldn't seem to find it.
Quote from: Marcus92 on March 01, 2017, 09:42:26 AM
2) Analytics are a big step forward in how we evaluate sports performance. But a system based primarily on points per possession excludes the most important result of any matchup: who actually won the game.
I know KenPom accounts for this to some degree with the Luck factor. But I'm not sure it gives winning or losing enough weight/credit.
Imagine a hypothetical team in an average conference with an average strength of schedule. The team loses every single game by 1 point (say 0-30). If I understand the system at all, KenPom would rate this team somewhere near the middle of the pack — say in the 160 range — with a Luck rating approaching -0.999. But by definition, this isn't an average team; it's the worst team in the country. They couldn't win a game, which is ultimately what counts.
I would argue that a team that plays an average SOS but loses every game by 1 point is not the worst team in the country. There would be a number of teams that they would be able to beat, because they were good enough to go play-for-play with all the "average" teams in college.
Quote from: Dr. Vinnie Boombatz on February 28, 2017, 05:43:01 PM
You don't have to go further than last year's Syracuse team to see an example of the latter. Lost 5 of 6 heading into the tournament and ended up in the Final Four.
You can also look at UNI last year. They struggled throughout the year, won their conference tournament, and went on to win their first round game against a 20 win Texas team that had 4 top 10 wins on its resume. And then for 39 minutes and 58 seconds they looked poised to make it to the Sweet 16 as well.
Anecdotes are easy to come by in March. But I have to think that in general, the teams playing better in February than November perform better in the tournament than the teams who play well in November and struggle in February.
We got this.
Quote from: Marcus92 on March 01, 2017, 09:42:26 AM
.
Imagine a hypothetical team in an average conference with an average strength of schedule. The team loses every single game by 1 point (say 0-30). If I understand the system at all, KenPom would rate this team somewhere near the middle of the pack — say in the 160 range — with a Luck rating approaching -0.999. it's the worst team in the country.
Sounds like you may not know the system at all
Quote from: Benny B on March 01, 2017, 09:58:55 AM
Because every additional blue-blood road game is one less game that's played at home. IOW, scheduling for RPI can't be done at the expense of game revenue.
That would not of been true this year. MU could of scheduled a 31st game, but did not. They simply could of scheduled to play a game at a blue blood and not of had any less home games this year. Giving up a home game is just an excuse.
Quote from: bilsu on March 01, 2017, 03:38:27 PM
That would not of been true this year. MU could of scheduled a 31st game, but did not. They simply could of scheduled to play a game at a blue blood and not of had any less home games this year. Giving up a home game is just an excuse.
They had a game (home and home?) with Utah scheduled, but it either fell through/was delayed. After that, they made the decision to skip the 31st game instead of playing a game that would drag down their RPI.
Quote from: Jay Bee on March 01, 2017, 03:30:08 PM
Sounds like you may not know the system at all
I'm always the first to admit a lack of knowledge, especially when it comes to statistical analysis.
But you didn't exactly enlighten anyone with this statement. I'm not trying to be argumentative — just looking for a little explanation beyond "you're wrong." Where did I misrepresent how KenPom evaluates the relative performance of basketball teams?
Quote from: MUeagle1090 on March 01, 2017, 03:41:47 PM
They had a game (home and home?) with Utah scheduled, but it either fell through/was delayed. After that, they made the decision to skip the 31st game instead of playing a game that would drag down their RPI.
Yes this is correct because of a change in their conf schedule next year would not allow. All that was left at that point were terrible buy games.