NE Region General Questions

[Hugenerd]

The numbers have been a work in progress... this year's numbers have actually be adjusted from the previous which I think where 1.4 and .6 (memory not serving right now).

Remember, the key seems to have been they want to stop rewarding teams who pack their schedules with home games... in a sense giving them an unfair advantage. The previous set-up seemed a bit heavy-handed... I haven't made up my mind on this year's version... though I don't have any negative feelings toward it. As far as knightslappy's numbers... I worry that the closer we get to 1.0... the further away we are from the purpose of the numbers in the first place.

Since the adjustment is made to both home and away games, 0.85/1.15 still weighs away games 35% higher than home games. 0.75/1.25 weights away games 67% higher than home games. I don't care what the numbers are, just that they are justified and not pulled from thin air (the previous factors were 0.6/1.4, which weighted away games 133% higher than away games). Rather than just tweeking the numbers every year until people don't complain, it seems alot easier to spend the time looking at what a true home/away advantage is for a good team (a team that would be considered for a pool c spot in a typical year) by compiling data from the last 5-10 years. Seems pretty straightforward to me.

[Dave 'd-mac' McHugh]

They are justified... I think they used the numbers D2 was using initially... when that didn't seem to work well for D3... they adjusted.

[Hugenerd]

They are justified... I think they used the numbers D2 was using initially... when that didn't seem to work well for D3... they adjusted.

What did they adjust based on? If they actually did a thorough analysis that justifies their numbers, why not publish them?

[augie_superfan]

They are justified... I think they used the numbers D2 was using initially... when that didn't seem to work well for D3... they adjusted.

What did they adjust based on? If they actually did a thorough analysis that justifies their numbers, why not publish them?

Hugenerd, got a few answers for you here...well maybe not definite answers but what I've "deduced" while studying all these numbers.  I think the general thinking is that the original 1.4/0.6 modifier came straight from the Division I RPI calculation.  However, the DI RPI uses the modifier on the winning percentage, not the OWP or OOWP.  There is very good reason why this is done.

The reason they use the 1.4/0.6 modifier is because historically, the home team wins about 67% of the games.  Thus, using the modifier works out to correcting a team's winning percentage for this fact.  See example below of 3 teams with actual WP's of 50% but playing 14,10,and 6 home games respectively:

Team  # Home Gms  Home WP  Away WP   Predicted WP   Actual WP    RPI WP
  A              14             0.50         0.50             0.57             0.50         0.42
  B              10             0.50         0.50             0.50             0.50         0.50
  C               6              0.50         0.50             0.43             0.50         0.58

So as you can see, Team A was expected to have a better WP because they played more home games and vice versa for Team C.  But since Team A "underperformed" the modifier makes up for this with the weighting of the games so that their "RPI WP" is lower than what their true WP is.  This makes sense to me, it applies the home/away modifier to something that is easily measureable.

NOTE:  Why the RPI doesn't use this adjusted WP when calculating a team's OWP and OOWP is beyond me however.  So they still have issues IMHO.

Now, for D1 baseball, they used this same modifier probably without any thought.  However, the home team doesn't win quite as often in baseball, closer to 62% I think.  So, as of next year, they will use 1.3/0.7 for baseball.  So they are adapting, which is nice to see.

Over the past 2 years, my data shows that for D3 mens basketball, that the home team wins about 58-60% of the time. Thus a modifier of 1.2/0.8 would make more sense (this is actually close to what my rough calculations showed last week).  Maybe the NCAA realized this and that is why the change was made from 1.4/0.6 to 1.25/0.75 for this season.

But, all this still doesn't explain why our modifier is not used to modify the WP.  If you are using WP data to create the modifier then my arguement would be why aren't you applying that modifier to the WP data?  Now, I haven't studied it close enough yet to determine if it makes any difference at all but it is slightly baffling why D1 and D3 would apply the modifier differently.  I plan to look at it closer after the end of the season.  If I find something fun it might be worth writing up and sending it to the statistics gurus at the NCAA, as their emails are listed on the NCAA website.

[Hugenerd]

They are justified... I think they used the numbers D2 was using initially... when that didn't seem to work well for D3... they adjusted.

What did they adjust based on? If they actually did a thorough analysis that justifies their numbers, why not publish them?

Hugenerd, got a few answers for you here...well maybe not definite answers but what I've "deduced" while studying all these numbers.  I think the general thinking is that the original 1.4/0.6 modifier came straight from the Division I RPI calculation.  However, the DI RPI uses the modifier on the winning percentage, not the OWP or OOWP.  There is very good reason why this is done.

The reason they use the 1.4/0.6 modifier is because historically, the home team wins about 67% of the games.  Thus, using the modifier works out to correcting a team's winning percentage for this fact.  See example below of 3 teams with actual WP's of 50% but playing 14,10,and 6 home games respectively:

Team  # Home Gms  Home WP  Away WP   Predicted WP   Actual WP    RPI WP
  A              14             0.50         0.50             0.57             0.50         0.42
  B              10             0.50         0.50             0.50             0.50         0.50
  C               6              0.50         0.50             0.43             0.50         0.58

So as you can see, Team A was expected to have a better WP because they played more home games and vice versa for Team C.  But since Team A "underperformed" the modifier makes up for this with the weighting of the games so that their "RPI WP" is lower than what their true WP is.  This makes sense to me, it applies the home/away modifier to something that is easily measureable.

NOTE:  Why the RPI doesn't use this adjusted WP when calculating a team's OWP and OOWP is beyond me however.  So they still have issues IMHO.

Now, for D1 baseball, they used this same modifier probably without any thought.  However, the home team doesn't win quite as often in baseball, closer to 62% I think.  So, as of next year, they will use 1.3/0.7 for baseball.  So they are adapting, which is nice to see.

Over the past 2 years, my data shows that for D3 mens basketball, that the home team wins about 58-60% of the time. Thus a modifier of 1.2/0.8 would make more sense (this is actually close to what my rough calculations showed last week).  Maybe the NCAA realized this and that is why the change was made from 1.4/0.6 to 1.25/0.75 for this season.

But, all this still doesn't explain why our modifier is not used to modify the WP.  If you are using WP data to create the modifier then my arguement would be why aren't you applying that modifier to the WP data?  Now, I haven't studied it close enough yet to determine if it makes any difference at all but it is slightly baffling why D1 and D3 would apply the modifier differently.  I plan to look at it closer after the end of the season.  If I find something fun it might be worth writing up and sending it to the statistics gurus at the NCAA, as their emails are listed on the NCAA website.

Sorry for not giving the credit you deserve for doing your analysis (I mistakenly said knightslappy).  As for the multipliers, I have no problem with whatever numbers they use, as long as they explain why they are using them, how they came up with them, and actually understand their proper use. Thank you for the explanation.

[augie_superfan]

No problem...there is some good info out there on D1 on why they've done what they have but obviously little to none for D3.  And that's pretty understandable.  It's much different when you have CBS and ESPN staring over your shoulder when doing things as compared to a few people on a message board.  And unfortunately, I don't think we can expect that our regional or national committees have those answers because they are just trying to do the best with the info they are given.

[Dave 'd-mac' McHugh]

The numbers, as shown, were not pulled out of the air... they took a previous model (again, think there was a D2 model there as well). After a couple of years with that model, they found that the numbers did not work. I do believe they looked at different statistical models (though, I have to check in with who told me that in the off-season) to come up with the numbers they are now using.

The reason the numbers were adjusted this year was because they didn't like the numbers they were getting... and felt it was weighted too much... so they did do a stats breakdown and came up with these numbers - that I was told officially.

[Hugenerd]

The numbers, as shown, were not pulled out of the air... they took a previous model (again, think there was a D2 model there as well). After a couple of years with that model, they found that the numbers did not work. I do believe they looked at different statistical models (though, I have to check in with who told me that in the off-season) to come up with the numbers they are now using.

The reason the numbers were adjusted this year was because they didn't like the numbers they were getting... and felt it was weighted too much... so they did do a stats breakdown and came up with these numbers - that I was told officially.

I still don't understand why they just wouldn't publish their analysis, even if it is just an abstract or bullet point list of the key points in an appendix in the handbook.  I'm certain there are publications out there that deal with statistics that would love this stuff because of the readership it would pull (that would be the 'academic' thing to do), or just put it some where in the handbook.  It is just the secretive manner in which these things are handled that make people suspicious of the methods used.

In fact, augie superfan, I bet if you put enough time into your database of games and developed thoughtful and rigorous methodology for analyzing this problem, you could get your work published also. I'm not saying it would be a high impact factor journal, but I guarantee you could find a place for it.

[amh63]

Hugenerd and others.........Interesting points but you must sometimes step back and face "reality" a bit before continuing chasing "windmills".  Even if I could do all the number crunching to obtain a rational model/equation to apply to Div3 or any division....giving the data base...and offer it to the NCAA free and thereby eliminate the over riding money/funds factor....it won't happen.  I will refer to an OLD exercise I went through in a graduate systems  course at MIT.  We were to take a given data base and come up with the best/optimum transportation system to alleviate the problem for our island state in the Pacific.
New land roads?, water transportation systems?, etc.  Though the class came up with many new and novel ideas...the ones that won dealt with new roads...even one around the island.  Why?.....bottom line....Money!  the US Highway funding that existed at the time, provided the key weighting factor in the exercise.  Not always rational, maybe unfair and so forth....but a reality.  The teacher was a visiting teacher that had worked in the real world and was political savy.
Look at the mess in big time college football and even in div 1 BB.  Yes MONEY is an unwanted blanket in college sports of student-athletics.
Besides, if the answer was here/now....we will have little to post here on this board?

[Hugenerd]

Hugenerd and others.........Interesting points but you must sometimes step back and face "reality" a bit before continuing chasing "windmills".  Even if I could do all the number crunching to obtain a rational model/equation to apply to Div3 or any division....giving the data base...and offer it to the NCAA free and thereby eliminate the over riding money/funds factor....it won't happen.  I will refer to an OLD exercise I went through in a graduate systems  course at MIT.  We were to take a given data base and come up with the best/optimum transportation system to alleviate the problem for our island state in the Pacific.
New land roads?, water transportation systems?, etc.  Though the class came up with many new and novel ideas...the ones that won dealt with new roads...even one around the island.  Why?.....bottom line....Money!  the US Highway funding that existed at the time, provided the key weighting factor in the exercise.  Not always rational, maybe unfair and so forth....but a reality.  The teacher was a visiting teacher that had worked in the real world and was political savy.
Look at the mess in big time college football and even in div 1 BB.  Yes MONEY is an unwanted blanket in college sports of student-athletics.
Besides, if the answer was here/nojIw....we will have little to post here on this board?

Luckily there is no money issue in D3 with regard to big money contacts etc., and I wasnt saying augie's work would be adopted by the NCAA, just that it would be publishable.  Also, my other comment was that if the NCAA had already done this analysis, why dont they just make it public?

[falcons2010]

Last set of rankings today?

[Ryan Scott (Hoops Fan)]

Last set of rankings today?

Nope, the NCAA apparently decided they didn't want to release them after all.

[falcons2010]

Another highly intelligent move by the NCAA. Lol, thats sarcasm by the way.

[pjunito]

A month ago, the committee chair was on d3 hoopsville and said the final regional rankings would be published.....  On Monday, he denies ever saying that and did an horrible job of explaining himself. I was waiting for him to say he was going to have tacos for dinner!

[falcons2010]

hahahah a nice reference to the east haven ct mayor pjunito! It makes no sense, but hey, lets just play the games now, throw them rankings out the window, as Byron Reeves did two yrs ago against Willaim Paterson