Great Lakes Region

Started by sac, February 21, 2007, 06:46:48 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

KnightSlappy

Quote from: Dave 'd-mac' McHugh on February 10, 2013, 10:10:59 PM
KnightSlappy... I have an answer to why your calculations are slightly off... I will be posting a blog on the topic hopefully later tonight - worse case tomorrow morning.

Looking forward to it. Is it the calculation method or data efficacy?

Dave 'd-mac' McHugh

Calculation method... I am finishing it up now, so it should be posted before lunch :).
Host of Hoopsville. USBWA Executive Board member. Broadcast Director for D3sports.com. Broadcaster for NCAA.com & several colleges. PA Announcer for Gophers & Brigade. Follow me on Twitter: @davemchugh or @d3hoopsville.

Dave 'd-mac' McHugh

Host of Hoopsville. USBWA Executive Board member. Broadcast Director for D3sports.com. Broadcaster for NCAA.com & several colleges. PA Announcer for Gophers & Brigade. Follow me on Twitter: @davemchugh or @d3hoopsville.

Pat Coleman

And as I read this, I see it as a return to sensible math, rather than the average of averages that has been written in the handbook for years but not performed until recently.

Rely on you guys now to tell us if the numbers line up yet.
Publisher. Questions? Check our FAQ for D3f, D3h.
Quote from: old 40 on September 25, 2007, 08:23:57 PMLet's discuss (sports) in a positive way, sometimes kidding each other with no disrespect.

KnightSlappy

#1159
Quote from: Pat Coleman on February 11, 2013, 12:09:43 PM
And as I read this, I see it as a return to sensible math, rather than the average of averages that has been written in the handbook for years but not performed until recently.

Rely on you guys now to tell us if the numbers line up yet.

I get the numbers to line up, but I think the average of the averages (old way) is the right way to go. Summing the numbers puts a bigger emphasis on how many games the other team has played. Taking Thomas More for an example: their SOS would be the same if they added a home game versus either a 15-0 team or an 18-3 team.

Interesting byproduct of the "new" way: The old way would have adjusted Thomas More's SOS down by about .010, the new way adjusts it up by .010.

Dave 'd-mac' McHugh

Yeah... I was thinking the same thing about how many games, but that is also where the flaw is. If a team has not played a lot of games and only lost a few, the percentage is going to be out of whack high. At least this formula breaks down each part of the data instead of just taking the average.
Host of Hoopsville. USBWA Executive Board member. Broadcast Director for D3sports.com. Broadcaster for NCAA.com & several colleges. PA Announcer for Gophers & Brigade. Follow me on Twitter: @davemchugh or @d3hoopsville.

Pat Coleman

Maybe I'm crazy, but I think average of averages only works when everyone has played the same number of games. Isn't someone's .600 worth more over 25 games than over 15 or 20? Seems like .600 over 25 games should have 25 games of weight.
Publisher. Questions? Check our FAQ for D3f, D3h.
Quote from: old 40 on September 25, 2007, 08:23:57 PMLet's discuss (sports) in a positive way, sometimes kidding each other with no disrespect.

KnightSlappy

#1162
Quote from: Pat Coleman on February 11, 2013, 12:31:38 PM
Maybe I'm crazy, but I think average of averages only works when everyone has played the same number of games. Isn't someone's .600 worth more over 25 games than over 15 or 20? Seems like .600 over 25 games should have 25 games of weight.

I think that's correct to an extent, but is 25 games really all that different than 24 or 23 (especially when teams from the same conference can have the same schedule and have a different number of regional games)? Here's what the new method says about opponents:

9-9 (home)
15-0 (road)
8-2 (home)
0-15 (home)

is easier than

8-8 (home)
16-0 (road)
8-2 (home)
0-16 (home)

Old method said these were identical schedules. In both cases the opponents played a combined 58 games.

ziggy

Quote from: Pat Coleman on February 11, 2013, 12:31:38 PM
Maybe I'm crazy, but I think average of averages only works when everyone has played the same number of games. Isn't someone's .600 worth more over 25 games than over 15 or 20? Seems like .600 over 25 games should have 25 games of weight.

More games only increases the certainty that the .600 team is in fact a .600 team, if that makes sense. More games doesn't change the difficulty in facing that opponent.

Dave 'd-mac' McHugh

Quote from: ziggy on February 11, 2013, 12:35:33 PM
Quote from: Pat Coleman on February 11, 2013, 12:31:38 PM
Maybe I'm crazy, but I think average of averages only works when everyone has played the same number of games. Isn't someone's .600 worth more over 25 games than over 15 or 20? Seems like .600 over 25 games should have 25 games of weight.

More games only increases the certainty that the .600 team is in fact a .600 team, if that makes sense. More games doesn't change the difficulty in facing that opponent.

Yeah - that is where I was headed in my thinking. The NCAA wants to prove that if you are say a .750 team its because you are 15-5 and not 9-2 in your regional games. Under the averages system those two records would be considered even; under the new system the team with more wins/games gives you a bigger boost to your SOS.
Host of Hoopsville. USBWA Executive Board member. Broadcast Director for D3sports.com. Broadcaster for NCAA.com & several colleges. PA Announcer for Gophers & Brigade. Follow me on Twitter: @davemchugh or @d3hoopsville.

KnightSlappy

Quote from: ziggy on February 11, 2013, 12:35:33 PM
Quote from: Pat Coleman on February 11, 2013, 12:31:38 PM
Maybe I'm crazy, but I think average of averages only works when everyone has played the same number of games. Isn't someone's .600 worth more over 25 games than over 15 or 20? Seems like .600 over 25 games should have 25 games of weight.

More games only increases the certainty that the .600 team is in fact a .600 team, if that makes sense. More games doesn't change the difficulty in facing that opponent.

Right, if two teams had, say, a .450 SOS heading into the final game and both were home against .500 teams, the team facing the 14-14 team would end up with a better SOS than the team that team that played the 12-12 team.

Conversely, if two teams had a .550 SOS going in to the final game, the team that played the .500 teams with the fewest total games would come out looking better.

Is this what we want?

Pat Coleman

Actually, yes, I think so.

And of course, you must recognize that the mathematical difference between the two teams' final SOS in your hypothetical is really quite minimal.
Publisher. Questions? Check our FAQ for D3f, D3h.
Quote from: old 40 on September 25, 2007, 08:23:57 PMLet's discuss (sports) in a positive way, sometimes kidding each other with no disrespect.

Dave 'd-mac' McHugh

It is about the number of games being played... or not played. That is where they were having the trouble in Division II... there were a number of provisional teams not playing a lot of games in Division II as they worked their way into the division. That resulted in teams with less games given the same benefit as the teams with more games. It is certainly a bigger factor there than it is here, but we do see it in Division III at times as well (Hope and Calvin come to mind considering sometimes the number of NAIA teams they play).

I don't have a problem that your SOS is boosted by a team that is playing more Division III teams in their region than a team that is playing less... I know it depends on the other coach's scheduling, but if we are already basing the WP on the regional games, than why shouldn't the SOS be based on those numbers and not the average as well.

And while I know this doesn't matter right now, next year we will have more games in the data flow for this, so it may become a smaller factor. And just remember, this is not a Division III decision, this is the Championships Committee decision that makes decisions for all sports in all divisions like keeping the final regional rankings private, despite cries from the Division III men's committee for one to make them public.
Host of Hoopsville. USBWA Executive Board member. Broadcast Director for D3sports.com. Broadcaster for NCAA.com & several colleges. PA Announcer for Gophers & Brigade. Follow me on Twitter: @davemchugh or @d3hoopsville.

KnightSlappy

So basically playing Finlandia at home is the way to go. You'll get a win (not affected by weighting) and their relative few number of games will make it so their negative SOS impact doesn't hurt so much.

ziggy

Quote from: Dave 'd-mac' McHugh on February 11, 2013, 12:50:02 PM
It is about the number of games being played... or not played. That is where they were having the trouble in Division II... there were a number of provisional teams not playing a lot of games in Division II as they worked their way into the division. That resulted in teams with less games given the same benefit as the teams with more games. It is certainly a bigger factor there than it is here, but we do see it in Division III at times as well (Hope and Calvin come to mind considering sometimes the number of NAIA teams they play).

I don't have a problem that your SOS is boosted by a team that is playing more Division III teams in their region than a team that is playing less... I know it depends on the other coach's scheduling, but if we are already basing the WP on the regional games, than why shouldn't the SOS be based on those numbers and not the average as well.

And while I know this doesn't matter right now, next year we will have more games in the data flow for this, so it may become a smaller factor. And just remember, this is not a Division III decision, this is the Championships Committee decision that makes decisions for all sports in all divisions like keeping the final regional rankings private, despite cries from the Division III men's committee for one to make them public.

Dave, great work getting this info, by the way.

Given the way it came about, it seems the NCAA has attempted to fix a very specific issue in a way that (in my view) is more problematic.

The SOS is about gauging the strength of opponents (and opponents' opponents) as a way to give context to winning percentage. Like I said before, the number of games an opponent has played has nothing to do with their strength and therefore should have no impact on a team's SOS. Instituting or changing a minimum game requirement is a direct solution to the problem and allows the SOS calculation to remain in place. After all, the problem wasn't the SOS calc to begin with. It was a scheduling issue.