Bracketology and SEC 7-9ers

[EricLA]

How can LSU be #12 in the RPI? Doesn't it give a much higher weight to wins? That's crazy and the RPI seems like it's becoming irrelevant, or at least lacking any common sense...

 

[Phil]

One-quarter of the RPI weight is wins. Three questers is who you play and who they play.

I've partially defended the RPI in the past. Given the constraint that it has to ignore useful information (MOV) it did OK.

The nature of the way it is constructed, it can be quite horrible in the early part of the season. It gets better and better throughout the season, and one hopes, since it is designed for end of season selection, that it might be OK by the end of the season.

That used to be sort of true, but now I'm having reservations.

It may have been UCMiami, but someone pointed out a corollary to something I had observed - I knew the formula was deliberately designed to ignore MOV, on the theory that they did not want the formula to reward blowouts. So the metric is not just a measurement, it intends to adjust behavior.

However, it doesn't just affect game behavior, it affects scheduling. That was deliberate, with an encouragement to schedule strong teams. However, I think teams have realized how to game the system. The weighting for schedule strength is atrocious, as has been discussed in the past. Clever teams will learn to schedule a number of teams 20-50 spots below them in the rankings, plus couple way above them, and avoid teams more than 100 spots below them. The ones way above them give them losses, but make up for it in schedule strength. The ones a few spots below them give them wins, without giving up schedule strength. The ones 100 spots below them hurt the schedule strength, and don't add more to the win count than a team 30 spots below them.

LSU is a perfect example of someone who has achieved (whether deliberately or not I cannot be sure) a favorable RPI SOS. They are number one as measured by RPI, but Massey who uses a sensible measure, rates their SOS as 14th. That's quite a difference.

According to RPI, UConn schedule strength is 37th. Do you need any more information to show it is flawed?

LSU gets most of their inflated RPI rating from their clever scheduling.

A better measure, such as Massey, would have their schedule strength at 14, and their overall rating as 35th. Still good enough to make the tournament if you look at raw numbers and ignore the "what have you done lately", but far below the inflated RPI value.

 

[Phil]

There are two contradictory stats for LSU, one of which says they almost surely cannot make the NCAAs, and the other says they have already clinched their spot.



Why the Tigers have a bid locked up
Despite their terrible play recently, the Tigers are #12 in the brain-dead RPI. That rank might change a tiny bit in the next few days, but the best RPI ranking for a team in the 2000s that did not make the tourney was #17. But that was not a major conference team (Western KY), and the highest RPI team from a major conference that did not make the tourney was #22 for a 13 loss Florida team back in 2000. By brain-dead RPI, LSU is in the tourney even if they lose the first-round SECT game to Bama.

Another reason why they are a lock? They are a host. That's not a guarantee—I think there have been situations where a host didn't qualify, but when it is a close call, they'll be in.

 

[UConnCat]

Another reason why they are a lock? They are a host. That's not a guarantee—I think there have been situations where a host didn't qualify, but when it is a close call, they'll be in.

Didn't know that. Forget everything I've said. LSU is in regardless of what happens in the SEC tournament.

 

[DobbsRover2]

How can LSU be #12 in the RPI? Doesn't it give a much higher weight to wins? That's crazy and the RPI seems like it's becoming irrelevant, or at least lacking any common sense...

In addition to all the other gaming techniques Phil noted for making the most RPI from your schedule is to adjust your games for home and away bonus\penalty advantage. For tourney type teams, a home loss hurts more than a road win helps, so you should try to schedule a lot of home games with teams near your level or a little below you and save your OOC road games (if you even have any nowadays) for the tough teams you need for sprucing up your resume plus a few of those more mediocre local teams that you know you can beat on the road. And you do not schedule teams rated above 200th at all, maybe even 150th.

This year LSU got those nice wins at home versus Rutgers, Michigan, and St. Joes that might have been losses on the road, and teams like Louisville and NC State might well have been losses at home anyway, so losing to them on the road didn't hurt that much.

RPI is just a stupid system that unfortunately is used to group teams together as a starting assessment tool even though it is badly flawed. Yet if you point out how screwed up the system is with simple easily understandable examples with a couple of teams playing each other, RPI proponents will say your examples are unreal and therefore trivial, a tactic which I've noticed that fans of flawed systems fall back upon. And unfortunately RPI is the often the fallback measure that a committee member can point to for any poor decision.

 

[Phil]

Good point about the home and away strategy.

I'm not opposed to a system that have incentives for behavior, but if you include those, you'd better be sure it incents behavior you want to happen. I'm not convinced this is the case. Time for the NCAA to do some hard thinking.

 

[triaddukefan]

Where are the quality wins by the SEC?

• 
  [ ]If I read correctly, the conference champion has one win over a ranked teams other than in conference.
  [ ]The number two team in the conference caught a freshman dominated UNC team in the second game of the season, before they jelled.
  [ ]The number three team in the conference, TAMU has zero wins against ranked teams out of conference.
  [ ]The number four team in the conference, Kentucky, actually won a legitimate game against a decent team.
  [ ]The number five team in the conference, Florida, has zero wins against ranked teams out of conference.

Why do some think the SEC is still an elite conference?

Shouldn't an elite conference manage more than three wins against ranked teams? UConn alone has as many wins before Thanksgiving against non-conference ranked teams as the entire SEC in their entire non-conference season.

The entire resume of the entire conference consists of one overtime win, a decent win by KY over Louisville and another win against a young team in their second game.

(edit oops, forgot the KY Louisville game, have edited)

South Carolina didnt have a win over a ranked team OOC.... their best win was vs the real USC. Oh yeah... they did beat UNC-CHarlotte too

 

[DobbsRover2]

South Carolina didnt have a win over a ranked team OOC.... their best win was vs the real USC. Oh yeah... they did beat UNC-CHarlotte too

The other issue for USCar is that even in conference they had only one road win over a team with a winning SEC record, though it was a plastering of KY, but Alabama also won at KY. They had chances against the second and third place teams but failed on the road at Texas A&M and UTenn.

WVU by contrast has road wins over the other two teams among the B12's top 3.

 

[Phil]

South Carolina didnt have a win over a ranked team OOC.... their best win was vs the real USC. Oh yeah... they did beat UNC-CHarlotte too

Heh, I was right the first time.