Louisiana Softball Makes BR Regional

[Cajun90]

They have high RPIs, SOS, and quality wins because they are ranked highly in the preseason and play each other. They can't help but have an artificially high RPI.

Ummmm..... RPI and SOS have nothing to do with rankings. Everyone starts off at zero from day one and its nothing more than a math formula. Now you can argue the formula has some bias built into it but other than that there is no subjectivity to it.

[LaCajunsFan]

If they want to still call us ULL or Louisiana Lafayette, or UL Lafayette, we call them State instead of LSU to get them riled up.

But I don't get what that has to do with my post of "NVM" that he quoted. BTW: I had removed my post because it was intended for the baseball thread, not SB.

[UncleCajun]

I get where you are coming from but the SEC had an extremely good year. Eleven of the thirteen teams were in the RPI top 20. There lowest ranked team was Ole Miss at 40. I don't think it would have been hard to leave Missouri or Ole Miss out of the field but it is also hard to argue against having them in the field.

I think a measure that should get more attention is how teams faired on the road. With that said keep in mind that is a metric that the Cajuns actually take advantage of by not playing many non-conference road games.

As far as hosting goes you once again can't argue with what happened based on the process that is in place. That process right now is that "in general" the top 16 teams are going to host. Now I'm not saying that process is the right process. If you want to grow the sport and find out just how much home field advantage plays then maybe you open things up a little bit. I certainly think the top 8 deserve to host but maybe the next eight sites have additional criteria. You could spread the wealth a little bit by looking at say the top thirty teams and awarding some bids to schools who have never hosted before. You could also try and spread it out between more conferences etc. I don't think that is being "unfair" but for now that isn't the process.

With that said..... Baylor got hosed. They should have hosted over Arkansas.

The use of RPI weighted toward the SEC belies the argument the SEC "had a very good year". More like, "the SEC had an artificially-attained extremely good year based on benefits of a built-in pre-season enhanced RPI and Strength of Schedule."

Take the majority of the SEC host teams out of the SEC and use Big 12 RPI and SOS - will see a marked difference in their placement. Missouri and Ole Miss would likely be nowhere near the top 64.

[Cajun90]

The use of RPI weighted toward the SEC belies the argument the SEC "had a very good year". More like, "the SEC had an artificially-attained extremely good year based on benefits of a built-in pre-season enhanced RPI and Strength of Schedule."

Care to explain what you mean by those statements?

Honest question. Do you understand how RPI & SOS are calculated?

[UncleCajun]

Care to explain what you mean by those statements?

Honest question. Do you understand how RPI & SOS are calculated?

Yes, I truly do understand how they are calculated. Do you? Do you understand why the Selection committee does not use only RPI to select the top 16 teams to host? Do you understand strength of schedule is built into RPI?

I get it. You are a fan of the SEC and the Selection Committee. I suppose you were okay last year when UL, Minnesota and James Madison had high RPI's, but were passed over for hosting for SEC teams.

Also, sorry you do not understand what I said in the statements. If you want to go to private message, I will explain to you at length what I meant. It's not fair to the ones who did understand and agree to have to do it again.

[Turbine]

Care to explain what you mean by those statements?

Honest question. Do you understand how RPI & SOS are calculated?

Make sure your teams win pre conference.

Then once conference play begins, almost universal cross play locks the entire conference into a strong SOS based on the overall winning record going in.

With this formula you can end up 1 game over .500 in conference and still have a tremendous RPI

[UncleCajun]

Make sure your teams win pre conference.

Then once conference play begins, almost universal cross play locks the entire conference into a strong SOS based on the overall winning record going in.

With this formula you can end up 1 game over .500 in conference and still have a tremendous RPI

Thanks, Turbine, that's what I meant to say?

Cajun90, I think I came off a little short with you and I apologize. I am just a little too passionate about the inequity on the process in place..

[Cajun90]

Thanks, Turbine, that's what I meant to say?

Cajun90, I think I came off a little short with you and I apologize. I am just a little too passionate about the inequity on the process in place..

Uncle you might want to look into my background a little more and ask around before blasting me. I can assure you I'm no fan of the SEC but to each their own.

The RPI is a statistical piece of junk in this day and age but the terms you use "weighted" and "built in preseason" just add to the confusion for many people who read this board.

Gonegolfin and I were talking RPI and post season picks long before there even was a Ragin Pagin. Now with his background he has taken the study to a level that I certainly haven't but other than him there is probably no one else on RP that has studied the issue in as much depth as I have regarding baseball and softball post-season at-large selections and how hosting sites are determined.

With that said I believe if you did have a correct statistical model in place you would still find that the SEC would put in at least 11 teams this year. Eleven of the thirteen SEC teams had a NON-conference RPI rated in the top 25 and they performed well against that schedule winning .8749 percent of those games. Now they certainly haven't done that every year but as a league they are very strong this year.

In regards to what should change with the RPI..... It needs to be scraped. It is extremely antiquated. The model was first built in the 70's where we didn't have the computational power we have today and needed something that could make sense of the numbers but not require the purchase of a supercomputer (at the time) and/or a lot of man power.

Today we have the computational power and we also have the historical data to validate the models. Ultimately you would like to see an AI solution with machine learning. You could load in a lot of additional variables to the formula to increase the accuracy as well. The NCAA has been very much against margin of victory because of the potential for manipulation but statistically it is a valid input. Other items could be road/home/neutral etc.

So why hasn't this been done yet? I've got two theories and they are just that theories. Unless something has changed in the last year or two I believe many in the NCAA statistical group are "old school" and have been around a while. They probably don't have the skill sets to implement an AI system. The other theory is that if they do have the skill sets they are concerned about getting "buy in" from the membership at large. It's not difficult to explain the RPI formula, even when you have a bonus/penalty structure. An AI based recursive system isn't going to be as clear cut for many people who don't have a computer/stat background. For most people in the world it becomes a black box kind of approach. You can tell them the inputs and you can show them the outputs but trying to explain what happened in between (black box) isn't going to be nearly as easy.

Like I said Gonegolfin can talk circles around me on this subject but I'm no newbie either. I do believe the committee when they say they didn't take conference affiliation into account when awarding host sites. I believe based on the criteria they are tasked with utilizing that they were pretty darn close but I would have changed a few. Now if your argument is that the criteria are wrong........ you might have a point.

[alum81]

It appears that we got a really good draw this year. Houston barley got in and LSU is one of the lesser host teams.

Offense and base running are our weaknesses but our girls have heart and will not give up. We will advance as far as Summer's arm will take us. GEAUX GET EM LADIES.

[UncleCajun]

Uncle you might want to look into my background a little more and ask around before blasting me. I can assure you I'm no fan of the SEC but to each their own.

The RPI is a statistical piece of junk in this day and age but the terms you use "weighted" and "built in preseason" just add to the confusion for many people who read this board.

Gonegolfin and I were talking RPI and post season picks long before there even was a Ragin Pagin. Now with his background he has taken the study to a level that I certainly haven't but other than him there is probably no one else on RP that has studied the issue in as much depth as I have regarding baseball and softball post-season at-large selections and how hosting sites are determined.

With that said I believe if you did have a correct statistical model in place you would still find that the SEC would put in at least 11 teams this year. Eleven of the thirteen SEC teams had a NON-conference RPI rated in the top 25 and they performed well against that schedule winning .8749 percent of those games. Now they certainly haven't done that every year but as a league they are very strong this year.

In regards to what should change with the RPI..... It needs to be scraped. It is extremely antiquated. The model was first built in the 70's where we didn't have the computational power we have today and needed something that could make sense of the numbers but not require the purchase of a supercomputer (at the time) and/or a lot of man power.

Today we have the computational power and we also have the historical data to validate the models. Ultimately you would like to see an AI solution with machine learning. You could load in a lot of additional variables to the formula to increase the accuracy as well. The NCAA has been very much against margin of victory because of the potential for manipulation but statistically it is a valid input. Other items could be road/home/neutral etc.

So why hasn't this been done yet? I've got two theories and they are just that theories. Unless something has changed in the last year or two I believe many in the NCAA statistical group are "old school" and have been around a while. They probably don't have the skill sets to implement an AI system. The other theory is that if they do have the skill sets they are concerned about getting "buy in" from the membership at large. It's not difficult to explain the RPI formula, even when you have a bonus/penalty structure. An AI based recursive system isn't going to be as clear cut for many people who don't have a computer/stat background. For most people in the world it becomes a black box kind of approach. You can tell them the inputs and you can show them the outputs but trying to explain what happened in between (black box) isn't going to be nearly as easy.

Like I said Gonegolfin can talk circles around me on this subject but I'm no newbie either. I do believe the committee when they say they didn't take conference affiliation into account when awarding host sites. I believe based on the criteria they are tasked with utilizing that they were pretty darn close but I would have changed a few. Now if your argument is that the criteria are wrong........ you might have a point.

90,
that is why I apologized - I looked back a little on some of your posts, BUT I did not like the tone on your response earlier - seemed kind of condescending, a mini-blast, if you will - of course, I believe a lot of the criteria is wrong. I also think hosting a regional if your conference record is under .500 does not pass the smell test. But then, that is a subjective opinion.
Truthfully, while I do know about RPI, I don't know as much as you do and appreciate the opportunity to learn more. Believe me, there are a lot us that think the SEC is somehow manipulating the process, possibly through the "old school" committee, and kind of like the "old school" NCAA/NIT committee that gave LSU the home court advantage over us in basketball this year.

Missouri at 28/27 overall and 6/17 in conference is a stretch to be in the tournament, jmo.