The 3 seed is not prediction based, it was resume based up to that point. Obviously, they will drop in the next update because I'm not in love with certain teams and can think objectively unlike what some others on this thread want you to believe.
Lunardi Has Us As An #11
- Thread starter colbert17
- Start date
Agreed - but there are times when the media does highlight them nationally which can help. Tonight’s game is huge for a lot of reasons, but this can be tacked right onto the list of golden opportunities.
CBS updates their rankings daily and a lot of ranked teams lost last night. I could care less about rank recognition, but Parrish usually includes resume commentary to explain (justify) new additions to his list. Have to avoid the let down first (a tall ask) but if we win, he might put us in his top 26. In which case, I could see him mentioning the OSU ref error along with the fact that we were missing our starting backcourt for the Temple game.
I’m not sure how anyone considered them a 3 seed before this loss. Xavier and OSU are pretty good wins, but our wins are better than theirs and we’re nowhere near a 3 seed. I get that we lost to weaker teams than them but this is also their 3rd double digit loss.
I’ve been trying to put my finger around why I think this method of evaluating teams for field selection would be so flawed. This model is basically saying that no team should ever be penalized for losing to Purdue. It doesn’t matter how bad they look in the game. Similarly - a bubble team should beat Temple so no team should be rewarded for beating them - rather, only penalized for losing to them - correct?
So basically this method is saying to avoid playing middle of the road teams at whatever cost possible. No reward for winning, only penalized for losing. Unless I’m missing something - a team that schedules 8 non-conference games against all Columbia types and 3 games against Purdue, UConn and Gonzaga need only go 8-3 (avoid a Lafayette type loss) to have a better resume in your system than Rutgers with our non-conference schedule this year unless we won out and went perfect 11-0. Do I have that right? Compared to a bubble teams “we’re supposed to” beat Temple and SHU at home just as that other team is supposed to lose to Purdue and company (never mind if they get crushed). To do better than that team in this system we’d have to also win the Miami game to get a +1 vs the bubble. Do I have this right? It’d be better to get blown out 3 times by Purdue, UConn and. Gonzaga since those are games a bubble team is supposed to lose?
1st of all no one uses it or even looks at it.
If the bubble team had a 15% probability of winning in Purdue then a loss subtracts .15 from WAB score
If the bubble team had a 15% probability of winning in Purdue then a loss subtracts .15 from WAB score
T-Page for Rutgers - Customizable College Basketball Tempo Free Stats - T-Rank
T-Rank College Basketball Ratings and Rankings. And fun.
www.barttorvik.com
look at the WAB column. We got to a .2 WAB after beating Lowell (we were 3-0 while bubble would have been 2.8-0.2)
The Temple loss dropped us .8 WAB so we went to a -0.6
Fast Forward to Purdue. before game we were a -1.0, Beating Purdue gave us .8 so we are now -.2.
If we win tonight we go to a +.1, if we lose tonight we slip to -.9
| WIN | LOSS | ||||||
| Thu 1-05 | H | 61 (Ⅱ) | Maryland | -8.0, 65-57 | 82% | 0.31 | -0.69 |
| Sun 1-08 | H | 60 (Ⅱ) | Iowa | -9.0, 74-65 | 82% | 0.31 | -0.69 |
| Wed 1-11 | A | 62 (I) | Northwestern | -1.5, 57-56 | 58% | 0.6 | -0.4 |
| Sun 1-15 | H | 11 (I-A) | Ohio St. | -2.9, 67-64 | 62% | 0.55 | -0.45 |
| Thu 1-19 | A | 52 (I) | Michigan St. | -1.0, 61-60 | 55% | 0.63 | -0.37 |
| Tue 1-24 | H | 43 (Ⅱ) | Penn St. | -6.9, 66-59 | 78% | 0.36 | -0.64 |
| Sun 1-29 | A | 60 (I) | Iowa | -1.8, 70-68 | 57% | 0.6 | -0.4 |
| Wed 2-01 | H | 210 (Ⅳ) | Minnesota | -16.7, 68-52 | 96% | 0.07 | -0.93 |
| Sat 2-04 | H | 52 (Ⅱ) | Michigan St. | -7.4, 64-57 | 80% | 0.34 | -0.66 |
| Tue 2-07 | A | 27 (I-A) | Indiana | +1.0, 64-63 | 45% | 0.71 | -0.29 |
| Sat 2-11 | A | 45 (I) | Illinois | -0.6, 65-64 | 53% | 0.65 | -0.35 |
| Tue 2-14 | H | 97 (Ⅲ) | Nebraska | -10.4, 64-53 | 89% | 0.21 | -0.79 |
| Sat 2-18 | A | 56 (I) | Wisconsin | -1.2, 57-56 | 56% | 0.61 | -0.39 |
| Thu 2-23 | H | 58 (Ⅲ) | Michigan | -8.3, 70-62 | 81% | 0.32 | -0.68 |
| Sun 2-26 | A | 43 (I) | Penn St. | -0.5, 63-62 | 52% | 0.65 | -0.35 |
| Thu 3-02 | A | 210 (Ⅲ) | Minnesota | -10.5, 65-54 | 88% | 0.21 | -0.79 |
| Sun 3-05 | H | 62 (Ⅱ) | Northwestern | -7.4, 60-53 | 82% | 0.31 | -0.69 |
If I understand that chart correctly, this system is saying 80% of bubble teams would beat Purdue at Mackey. I'm calling BS on that.
You implied that it’s wrong that it’s not used - or in the least winning at Purdue really should just offset a neutral loss at Temple. I was responding to that comment. At the end of the day, unless it’s a loss like the one we had vs. Lafayette, I just don’t agree. Teams are going to have off days and who you lose to is often more a product of that than the opponent outside of the most elite teams. I do not think losing at Michigan by 30 was a “better” loss for Maryland than our loss to Temple was. But this system would disagree.
So yes WAB is not a thing, at least I don't think, for NCAA purposes.
It is my belief that a team should be viewed as what was their record and how hard was their schedule...that's it!
To me it is binary...you win or you lose. No style points and no good loses.
I also think we shouldn't overweight a team's good days and underweight their bad days.
I think the NET (which looks at scoring margin directly and indirectly) is great for strength of schedule reasons only.
Again my personal opinion.
It is my belief that a team should be viewed as what was their record and how hard was their schedule...that's it!
To me it is binary...you win or you lose. No style points and no good loses.
I also think we shouldn't overweight a team's good days and underweight their bad days.
I think the NET (which looks at scoring margin directly and indirectly) is great for strength of schedule reasons only.
Again my personal opinion.
The Purdue / Temple example......YES. In fact it was just about a wash. As the "bubble team" was expected to win 1.0 games.
I also think in the NFL losing to the Colts and beating the Eagles should equal beating the Colts and losing to the Eagles. As far as the NBA all wins are losses because the NBA stinks.
I also think in the NFL losing to the Colts and beating the Eagles should equal beating the Colts and losing to the Eagles. As far as the NBA all wins are losses because the NBA stinks.
20% or .2 wins
If you get a win you did .8 wins better than the bubble team
If you lose you did .2 worse than the bubble team
Tonight the average bubble team would get .69 wins playing against MD at home
If you get a win you did .31 wins better than the bubble
If you get a loss tow did .69 worse than the bubble.
The more you believe this is true the less it is probably true.
The opposite. It's saying a bubble team would win 20% of the time, so winning is 0.8 wins better than this.
I think this is what the committee is actually trying to do with NET. They are basically doing WAB in their heads with the quads and whatnot. They don't really appear to be directly using it to select teams.. it would be tough to pull the analysis I did last year but Kenpom, Bart, and my computer rankings that I know for certain the NCAA committee has never seen all had better correlation with seedings than the NET did.
I have a W/L only metric (basically RPI but without the flaws.. an actually properly formulated W/L only measure) that I would just use directly to select and seed the field if I were God of the NCAA.
It is, and it produces generally reasonable looking results, but it is not a well formed model.
I see my math error. I still meant to say I don't think winning at Mackey this year is 20% likely for true bubble teams. Maybe 10%.
It arguably does what Green says he wants to see. Rutgers rank in it is probably around where it is in the WAB.
The weightings in RPI were always an issue. Metrics were too heavily inflated by having played opponents with gaudy records who didn’t beat anyone. And the blended average impact was more problematic than anything.
Right, it does it but not particularly well. I mean in theory even just a simple comparison of winning percentages would satisfy Greene's requirement but I'm certain he and virtually anyone would agree that's a terrible system in a sport like college basketball with absurdly unbalanced schedules.
He did say SOS was part of his binary requirement. RPI accounts for SOS - just does so very poorly because of the way the blended averages work in the calculation.
Also - I’m not sure whether or not WAB would even address that issue either. I guess it depends on how the odds of a bubble team winning are being determined.
The flaw in a nutshell - team A goes 10-10. Plays the top 10 teams in the country on the road and the worst 10 D1 teams at home. Only wins the home games. The second iteration of RPI that accounts for SOS based on opponents’ opponents record would have a blended average factor of about a 500 record. The trouble is - team A showed no ability to beat 500ish teams and only beat up on the worst possible teams.
The most likely outcome would be 10 road losses for most bubble teams (@Houston, @Tenn, @ Kansas, etc.) - same 10-10 record so does that mean team A ought to be in the bubble mix?
An extreme example I know but intended just to get the point across.
The flaw in a nutshell - team A goes 10-10. Plays the top 10 teams in the country on the road and the worst 10 D1 teams at home. Only wins the home games. The second iteration of RPI that accounts for SOS based on opponents’ opponents record would have a blended average factor of about a 500 record. The trouble is - team A showed no ability to beat 500ish teams and only beat up on the worst possible teams.
The most likely outcome would be 10 road losses for most bubble teams (@Houston, @Tenn, @ Kansas, etc.) - same 10-10 record so does that mean team A ought to be in the bubble mix?
An extreme example I know but intended just to get the point across.
I've often had these kind of hangups when thinking about the design of my own metrics, but I don't think they're actually particularly important. The problem in your example is that the team has basically played a schedule that provides no info on how good they are. But that's a scheduling problem, not a metric problem.
If you go 10-10 with a schedule of ten games on the road against Houston and 10 games at home against IUPUI, the best guess I can give is that you are a middle of the pack team. But there is huge variance about that estimate; the real answer is something more like "need more data". But in real life no one plays a schedule like that, and if they do "middle of the pack" seems fine as a place to put them. They haven't done anything good or bad.
67/68 teams correctly predicted in the tournament last year. 46 exact seeds, 20 off by 1, 1 off by 2 and 1 team I had in Dayton (Texas A&M) didn't make it, while the team I had out (Notre Dame) went to Dayton instead. I think I'm doing just fine, thanks.
It will adjust itself this weekend if they beat Iowa. Many of the guys I follow had Rutgers 11/FFO before the Purdue win. Now, I'm seeing a lot of 5-8 and we're only going to keep winning. You can't discount the two quad 3 losses which are holding us down currently, but are starting to fade away with each victory.
I’m sorry if selection Sunday was tomorrow I can’t see how we are not a 4 seed. We have the best win if anyone this year - NET of 16 and no very bad losses.
The other 30 teams with a better current resume disagree. they didnt lose to temple and seton hall.
thats not how it works.....ranking does mean the seed is similar. I would say RU is probably a 6 now
Still don’t get the Indiana love. lunardi today has them as a 6. They have wins over unc and Xavier and nothing else.
Its because he is doing bracketology like its March when its early January where movement right now should be large.
Your binary logic doesn’t work because comprehensive SOS isn’t actually binary. NM (12 in WAB) doesn’t have a better resume simply because WAB “thinks” we’d have lost more than one game playing their schedule. Blended win/loss percentage statistics are complete garbage.
Our Temple loss (Kenpom 124) is equivalent to their loss to Fresno St (138). Both teams lost to one team as bad as or worse than Temple.
We went 4-3 vs. @ Purdue, Indiana (with Xavier and Race), @ OSU, @ Miami, Maryland, WF and Seton Hall.
No matter what WAB’s algorithm says - eeking out wins vs. St Marys, Iona and SF does not mean NM would do better than 4-3 against the above teams. And by the way - with Caleb and Paul healthy I do think we’d have a perfect record against their schedule.
I think he’s probably assuming Race and Xavier will be back and Indiana’s loss count will be low enough to warrant inclusion in the field. Injuries are largely disregarded for field selection but tend to be more reflected in seeding.
I don't want SOS to be binary. I want team evaluating to be based on binary wins and losses vs. non binary SOS.
I don't care about eeking out wins. Every sport I used to play and watch it is about a W or L.
New Mexico did win those games!
Injuries are part of the game.
If you want to really get me.......ask me to justify Sam Houston State resume better than RU. I can't
Similar threads
