Regional Projection 5-11

[CrazyCajun]

Do we want coastal to get a miracle or continue to lose?

Lose, or they will steal an at large spot. Hopefully Georgia southern can beat Georgia tech, USM beat ole miss and south al beat uno tonight. Need our previous opponents to get their ___ together.

[GoneGolfin]

Do we want coastal to get a miracle or continue to lose?

Lose.

Brian

[RaginDave]

I’m assuming we want georgia southern to beat ga tech tonight. uSA UNO was canceled. And ULM plays jackson state. We obviously want them to win.

[RaginScotsman]

Lose, or they will steal an at large spot. Hopefully Georgia southern can beat Georgia tech, USM beat ole miss and south al beat uno tonight. Need our previous opponents to get their ___ together.

Well, ya. I wish it came down to run differential like in soccer.

[GoneGolfin]

I’m assuming we want georgia southern to beat ga tech tonight. uSA UNO was canceled. And ULM plays jackson state. We obviously want them to win.

Affirmative. It may also help remove Georgia Tech from consideration.

Brian

[Big Buds]

GoneGolfin/Brian,

In what worst case scenario do you see Louisiana as an at large? As in, what is the very worse we can do the rest of the way, including the Conference Tournament, to get an at large?

[Cajunsmike]

RPI is frustrating. I think overall I would say it’s a good ranking system. But when it hurts, it really hurts. 9-1 in our last 10 games but dropping every day. I think we were as high as 38th in the last couple week? Now down to 56 and I’m sure will drop further even with a win tonight. That sucks.

RPI is 25% your record, 25% your opponents, and 50 % opponents of your opponents. Such a system that is so heavily weighted on who you play and even more so on who they play is deeply flawed. It does not reward how you play nearly enough. A 34/33/33 breakdown would be much more appropriate. Others have made the same observation in the past but it won't change as it gives the schools in the high profile leagues an advantage over those who have less of a favorable schedule situation.

[GeauxCajuns]

RPI is 25% your record, 25% your opponents, and 50 % opponents of your opponents. Such a system that is so heavily weighted on who you play and even more so on who they play is deeply flawed. It does not reward how you play nearly enough. A 34/33/33 breakdown would be much more appropriate. Others have made the same observation in the past but it won't change as it gives the schools in the high profile leagues an advantage over those who have less of a favorable schedule situation.

I think you have those percentages swapped

[GoneGolfin]

RPI is 25% your record, 25% your opponents, and 50 % opponents of your opponents. Such a system that is so heavily weighted on who you play and even more so on who they play is deeply flawed. It does not reward how you play nearly enough. A 34/33/33 breakdown would be much more appropriate. Others have made the same observation in the past but it won't change as it gives the schools in the high profile leagues an advantage over those who have less of a favorable schedule situation.

No, it is 50% your opponent's WP and 25% your opponent's opponents' WP.

One of the significant problems is that it is too heavily weighted on your opponent's WP, without regard to how strong their respective schedule is. The opposite of what you state above. But it is more than that.

A 34/33/33 breakdown also does not improve it much. It is more of a problem with how many levels the formula descends than the weights of the fixed three level formula. You will never arrive at a reasonable approximation of strength with three levels. The RPI is a pedestrian formula that never should have lasted as long as it has. As a mathematician and researcher in machine learning and artificial intelligence (statistical learning), I actually feel it is an embarrassment to the field of statistics.

Brian (1981 called and it wants its RPI formula back)

[J-Town Cajun]

No, it is 50% your opponent's WP and 25% your opponent's opponents' WP.

One of the significant problems is that it is too heavily weighted on your opponent's WP, without regard to how strong their respective schedule is. The opposite of what you state above. But it is more than that.

A 34/33/33 breakdown also does not improve it much. It is more of a problem with how many levels the formula descends than the weights of the fixed three level formula. You will never arrive at a reasonable approximation of strength with three levels. The RPI is a pedestrian formula that never should have lasted as long as it has. As a mathematician and researcher in machine learning and artificial intelligence (statistical learning), I actually feel it is an embarrassment to the field of statistics.

Brian (1981 called and it wants its RPI formula back)

Brian, can you elaborate more on what a better formula would look like? How many levels would you recommend and what other statistical formulas would you factor in?

[ZoomZoom]

Affirmative. It may also help remove Georgia Tech from consideration.

Brian

Tech 13-12 winner. GATA RPI 10.

[Jacob81]

Another bad night in the belt. Sunbelt conference rpi has gone from 5th to 8th in 2 weeks. Cusa, Missouri valley, and AAC have all moved passed us. Not much time to catch back up. With that said, a 5th ranked conf can get 3 maybe even 4 teams in. 8th place conf probably 2 teams maybe 3 if the tourney winner is an upset. I think it’s time to win “3 games in March” or win out.