Top 25 talk

[Smitty Oom]

Issues I have with Massey and Bennett can be best explained this way:

Augustana's record on Massey's website has them with three losses. They are counting the game against Illinois on Augustana's resume. It was an exhibition game for Augustana. It is a quirk with lower divisions, but it happens (unlike Catholic who chose to play Maryland as a real contest). Yes, Illinois counts the game, by rule, as a real game, but Augustana didn't.

That game greatly impacts Augustana's resume with Massey and thus everyone they play. And they aren't the only one. The struggle with Massey to not count games is frustrating and in the Central part of the country a lot of these cross over games affect the system in my opinion. I know there are cross-over games, but I have not found any games against DII, DI, NAIA, etc. to have any impact on Division III... but Massey and Bennett is affected.

Sadly, DIII can't have their games isolated to just the division or games that actually count on a consistent basis and thus the rankings are affected.

I also went through both rankings this week with the 47 teams I was looking at for Top 25 reasons. I found some odd discrepencies across the board. I can write them up later, but I was shocked at how one team ranked high in both despite horrible SOS numbers (Bennett) while another with better numbers was ranked lower.

Another thing about Bennett, I don't like how the points for and points against seem to have a significant impact. The game is far too complicated to use points for and against as part of a barometer. John Carroll puts up a ton of points, while giving up a ton as well, does that mean they are better or worse than a team who doesn't need to score a lot because their defense is so stingy?

I too have found the WIAC and MIAC ranked highly by these systems, but the national championships isn't a good enough excuse to say it is right. Sure, the top teams have won. Great. No one is denying that. It is the number of teams so highly ranked in the MIAC, especially, that gets odd. I have stated the WIAC may be the best conference in the country this year with the most depth, so I am not surprised if they rank well. I have also said the MIAC is far more competitive than it has ever been. However, I don't feel more than three or four of the MIAC teams could compete well on a national level and yet the rankings make it seem like six or seven (off the cuff) could compete. That just doesn't add up to reality in Division III.

MIAC plays good teams, particularly in the WIAC, and does well enough to get some lift from it. If they were just getting hammered every year it would show. In the year when they were ranked 29th I imagine they did just that.

The WIAC is just a great league.

I'm not bothered about the D1 games because the fact that they count for D1 means that they're going to play to win, and the algo is not going to punish a D3 team for losing to a good D1. But if you can beat one, I'm sure it would reward that handsomely, and surely should. If they were exhibitions for D1 and counted for D3, you could have a problem with D1s not taking them seriously. But it's not, so that's not a problem.

Also this is a pretty new development in that regard having the games that count for one and not the other, and the WIAC being highly ranked is much much longer-term than that.

I'm hearing a lot of "well they just can't be that good, I don't think they are and so they aren't".

I am a MIAC guy, and I was shocked how high some of the teams were. I really like the Johnnies this year but I was stunned when I noticed they were no.1 according to Massey. Teams like Gustavus and St. Olaf are solid middle of the road teams but should not be considered in the top 50 teams in all of division 3. I personally think that although it is more competitive this year, the MIAC overall is down in talent when compared to the past couple years, especially 2015-2016 when the Tommies won the national championship. The top teams like John's and Augsburg are very solid, but the middle and lower tiers are (and should) be easier wins for the top tier this year.

Me, the MIAC does play a lot of games against the WIAC, but the majority of non0conference games are against UMAC teams. And frankly, those are not good games/teams for the most part. So I wouldn't say they necessarily have a tougher non-conference schedule than others.

[me]

Issues I have with Massey and Bennett can be best explained this way:

Augustana's record on Massey's website has them with three losses. They are counting the game against Illinois on Augustana's resume. It was an exhibition game for Augustana. It is a quirk with lower divisions, but it happens (unlike Catholic who chose to play Maryland as a real contest). Yes, Illinois counts the game, by rule, as a real game, but Augustana didn't.

That game greatly impacts Augustana's resume with Massey and thus everyone they play. And they aren't the only one. The struggle with Massey to not count games is frustrating and in the Central part of the country a lot of these cross over games affect the system in my opinion. I know there are cross-over games, but I have not found any games against DII, DI, NAIA, etc. to have any impact on Division III... but Massey and Bennett is affected.

Sadly, DIII can't have their games isolated to just the division or games that actually count on a consistent basis and thus the rankings are affected.

I also went through both rankings this week with the 47 teams I was looking at for Top 25 reasons. I found some odd discrepencies across the board. I can write them up later, but I was shocked at how one team ranked high in both despite horrible SOS numbers (Bennett) while another with better numbers was ranked lower.

Another thing about Bennett, I don't like how the points for and points against seem to have a significant impact. The game is far too complicated to use points for and against as part of a barometer. John Carroll puts up a ton of points, while giving up a ton as well, does that mean they are better or worse than a team who doesn't need to score a lot because their defense is so stingy?

I too have found the WIAC and MIAC ranked highly by these systems, but the national championships isn't a good enough excuse to say it is right. Sure, the top teams have won. Great. No one is denying that. It is the number of teams so highly ranked in the MIAC, especially, that gets odd. I have stated the WIAC may be the best conference in the country this year with the most depth, so I am not surprised if they rank well. I have also said the MIAC is far more competitive than it has ever been. However, I don't feel more than three or four of the MIAC teams could compete well on a national level and yet the rankings make it seem like six or seven (off the cuff) could compete. That just doesn't add up to reality in Division III.

MIAC plays good teams, particularly in the WIAC, and does well enough to get some lift from it. If they were just getting hammered every year it would show. In the year when they were ranked 29th I imagine they did just that.

The WIAC is just a great league.

I'm not bothered about the D1 games because the fact that they count for D1 means that they're going to play to win, and the algo is not going to punish a D3 team for losing to a good D1. But if you can beat one, I'm sure it would reward that handsomely, and surely should. If they were exhibitions for D1 and counted for D3, you could have a problem with D1s not taking them seriously. But it's not, so that's not a problem.

Also this is a pretty new development in that regard having the games that count for one and not the other, and the WIAC being highly ranked is much much longer-term than that.

I'm hearing a lot of "well they just can't be that good, I don't think they are and so they aren't".

I am a MIAC guy, and I was shocked how high some of the teams were. I really like the Johnnies this year but I was stunned when I noticed they were no.1 according to Massey. Teams like Gustavus and St. Olaf are solid middle of the road teams but should not be considered in the top 50 teams in all of division 3. I personally think that although it is more competitive this year, the MIAC overall is down in talent when compared to the past couple years, especially 2015-2016 when the Tommies won the national championship. The top teams like John's and Augsburg are very solid, but the middle and lower tiers are (and should) be easier wins for the top tier this year.

Me, the MIAC does play a lot of games against the WIAC, but the majority of non0conference games are against UMAC teams. And frankly, those are not good games/teams for the most part. So I wouldn't say they necessarily have a tougher non-conference schedule than others.

Why shouldn't they be considered in top 50? Half or more of D3 treats athletics as recreational and is hardly ever going to be good.

I guess we'll see how much depth there is in the league. It looks so far like a lot. Behind St. John's is 6 teams with either 2 or 3 league losses.

St. John's non-con win we know.
Augsburg beat Wartburg and Loras. Bethel beat UWL and Simpson. St. Olaf beat Stevens Point and Eau Claire (probably a big part of why people don't think they're very good). St. Thomas beat Dickinson. Carleton beat Platteville. Hamline beat Williams.

There are obvious reasons to think there are several capable MIAC teams this year.

And Bethel beats Augsburg on the road. And Hamline beats Carleton.

This is what happens in power conferences. They beat each other up. It's not a sign of weakness, it's a sign of strength.

[Darryl Nester]

How They Fared (So Far)

Top 25

Rank	Pts	Team	W-L	Results
#1	625	Whitman	14-0	def. #3 Whitworth, 91-75; 01/12 vs. George Fox
#2	530	Augustana	11-3	LOST to Wheaton (Ill.), 81-86; 01/13 vs. Carroll
#3	512	Whitworth	12-2	LOST at #1 Whitman, 75-91; 01/13 vs. George Fox
#4	479	Williams	11-2	01/12 vs. #44 Tufts; 01/13 vs. Bates
#5	473	UW-Whitewater	12-2	won at UW-Eau Claire, 78-65; 01/13 vs. UW-Stout
#6	461	Washington U.	10-2	01/12 vs. Brandeis; 01/14 vs. New York University
#7	459	UW-River Falls	11-2	def. #27 UW-Platteville, 91-84; 01/13 at #13 UW-Oshkosh
#8	458	York (Pa.)	14-0	def. Penn State-Harrisburg, 88-78; 01/13 vs. T#33 Christopher Newport
#9	452	Swarthmore	12-1	won at McDaniel, 80-49; 01/11 vs. Johns Hopkins; 01/13 at #22 Franklin and Marshall
#10	359	Wittenberg	14-0	won at Wabash, 72-58; 01/13 at T#35 Wooster
#11	352	MIT	13-1	def. Emerson, 83-69; 01/13 at Wheaton (Mass.)
#12	349	Lycoming	15-1	def. #38 Albright, 86-85; 01/13 at Arcadia
#13	345	UW-Oshkosh	12-2	def. UW-Stout, 89-66; 01/13 vs. #7 UW-River Falls
#14	306	Wesleyan	10-2	01/12 at #18 Hamilton; 01/13 at Amherst
#15	287	St. John's	11-1	won at Carleton, 69-68; 01/13 at T#39 Augsburg
#16	258	Middlebury	9-3	def. Morrisville State, 85-64; 01/12 vs. Bates; 01/13 vs. #44 Tufts
#17	249	Ohio Wesleyan	10-4	LOST to Hiram, 79-88; 01/13 at Allegheny
#18	219	Hamilton	12-0	01/12 vs. #14 Wesleyan; 01/13 vs. Connecticut College
#19	115	Marietta	11-3	won at Mount Union, 78-57; 01/13 at Heidelberg
#20	111	Emory	10-2	01/12 vs. Carnegie Mellon; 01/14 vs. Case Western Reserve
#21	84	Wartburg	11-3	def. Luther, 73-51; 01/13 at Simpson
#22	77	Franklin and Marshall	11-2	LOST at Muhlenberg, 80-82; 01/11 at Washington College; 01/13 vs. #9 Swarthmore
#23	70	John Carroll	12-2	def. Otterbein, 122-66; 01/13 at Wilmington
#24	59	Eastern Connecticut	12-2	def. Mass-Dartmouth, 101-77; 01/13 at Southern Maine
#25	56	Baldwin Wallace	11-3	LOST to Ohio Northern, 73-80; 01/13 at Otterbein

Others receiving votes

Rank	Pts	Team	W-L	Results
#26	55	New Jersey City	12-2	def. Stockton, 81-67; 01/13 at T#35 Ramapo
#27	48	UW-Platteville	12-2	LOST at #7 UW-River Falls, 84-91; 01/13 vs. UW-La Crosse
#28	46	St. Norbert	11-2	01/12 at Illinois College; 01/13 at Monmouth
#29	40	Juniata	13-1	def. Goucher, 64-53; 01/13 at Drew
#30	39	Skidmore	8-3	LOST at Vassar, 59-67; 01/12 at Rochester Tech; 01/13 at Hobart
#31	25	Rochester	9-3	01/12 vs. Case Western Reserve; 01/14 vs. Carnegie Mellon
#32	23	Illinois Wesleyan	11-3	01/13 at North Park
T#33	13	Christopher Newport	11-3	won at Southern Virginia, 83-58; 01/13 at #8 York (Pa.)
T#33	13	Nichols	10-2	won at Curry, 100-68; 01/11 vs. Salve Regina; 01/13 at Wentworth
T#35	11	Ramapo	11-4	won at Rowan, 86-76; 01/13 vs. #26 New Jersey City
T#35	11	Wooster	11-3	def. Allegheny, 79-58; 01/13 vs. #10 Wittenberg
#37	10	Salem State	11-3	def. Fitchburg State, 80-63; 01/13 at Framingham State
#38	9	Albright	11-3	LOST at #12 Lycoming, 85-86; 01/13 vs. Messiah
T#39	7	Augsburg	11-3	won at St. Thomas, 62-52; LOST to Bethel, 67-78; 01/13 vs. #15 St. John's
T#39	7	Buena Vista	12-2	won at Dubuque, 96-87; 01/13 vs. Loras
T#39	7	Emory and Henry	13-1	def. Roanoke, 75-73; 01/13 at Hampden-Sydney
T#39	7	Hanover	10-4	def. Earlham, 90-54; 01/13 at Anderson
#43	4	Nebraska Wesleyan	12-2	LOST at Loras, 87-97 OT; 01/13 vs. Dubuque
#44	3	Tufts	11-3	01/12 at #4 Williams; 01/13 at #16 Middlebury
#45	2	Lebanon Valley	11-5	LOST to Arcadia, 82-94

[Pat Coleman]

D. I do feel that the one year a middling MIAC team beat a MIAC team that beat the undefeated defending national champs got peoples panties in a bunch. But anomalous results effect every rating system, human based or computer based. Remember, it's just one data point out of several interlocking data points in a big data universe. Trust data.

You've mentioned this twice but just to clarify -- I have had this opinion of Massey's ratings effectiveness at the D-III basketball level for longer than that.

Trust data, sure. Trust the formula, though ...

If you don't trust the Massey formula, I'd venture to guess you don't trust much of anything. The guy does quality ratings of sports around the world with the same basic formula.

But that doesn't mean it works for Division III basketball.

[FCGrizzliesGrad]

D. I do feel that the one year a middling MIAC team beat a MIAC team that beat the undefeated defending national champs got peoples panties in a bunch. But anomalous results effect every rating system, human based or computer based. Remember, it's just one data point out of several interlocking data points in a big data universe. Trust data.

You've mentioned this twice but just to clarify -- I have had this opinion of Massey's ratings effectiveness at the D-III basketball level for longer than that.

Trust data, sure. Trust the formula, though ...

If you don't trust the Massey formula, I'd venture to guess you don't trust much of anything. The guy does quality ratings of sports around the world with the same basic formula.

But that doesn't mean it works for Division III basketball.

It's obviously better than just random chance, but it's certainly not perfect either. It's somewhere in the middle meaning it can be useful as a tool but not as the only tool.
Just looking over Saturday's games, out of the 184 predicted winners, only 131 were correct or a 71.2% accuracy rate. Not exactly spectacular.

[me]

D. I do feel that the one year a middling MIAC team beat a MIAC team that beat the undefeated defending national champs got peoples panties in a bunch. But anomalous results effect every rating system, human based or computer based. Remember, it's just one data point out of several interlocking data points in a big data universe. Trust data.

You've mentioned this twice but just to clarify -- I have had this opinion of Massey's ratings effectiveness at the D-III basketball level for longer than that.

Trust data, sure. Trust the formula, though ...

If you don't trust the Massey formula, I'd venture to guess you don't trust much of anything. The guy does quality ratings of sports around the world with the same basic formula.

But that doesn't mean it works for Division III basketball.

Doesn't mean it doesn't, and it's much more likely that it does. It's not like it's rigged to work for D1 or something.

[me]

D. I do feel that the one year a middling MIAC team beat a MIAC team that beat the undefeated defending national champs got peoples panties in a bunch. But anomalous results effect every rating system, human based or computer based. Remember, it's just one data point out of several interlocking data points in a big data universe. Trust data.

You've mentioned this twice but just to clarify -- I have had this opinion of Massey's ratings effectiveness at the D-III basketball level for longer than that.

Trust data, sure. Trust the formula, though ...

If you don't trust the Massey formula, I'd venture to guess you don't trust much of anything. The guy does quality ratings of sports around the world with the same basic formula.

But that doesn't mean it works for Division III basketball.

It's obviously better than just random chance, but it's certainly not perfect either. It's somewhere in the middle meaning it can be useful as a tool but not as the only tool.
Just looking over Saturday's games, out of the 184 predicted winners, only 131 were correct or a 71.2% accuracy rate. Not exactly spectacular.

LOL wow. nothing like a totally out of context number to prove a point, I guess.

In the interest of trying to be helpful rather than just critical, one way to evaluate predictions might be to group all the games where the ratings said the prob was 60-40 or less for either team. And then see what the results were. Same for 70-30, etc.

Another way would be compare the ratings prediction accuracy to other methods.

I have a feeling 71% for the whole country is just fine.

[Greek Tragedy]

TGHIJGSTO!!!

[FCGrizzliesGrad]

D. I do feel that the one year a middling MIAC team beat a MIAC team that beat the undefeated defending national champs got peoples panties in a bunch. But anomalous results effect every rating system, human based or computer based. Remember, it's just one data point out of several interlocking data points in a big data universe. Trust data.

You've mentioned this twice but just to clarify -- I have had this opinion of Massey's ratings effectiveness at the D-III basketball level for longer than that.

Trust data, sure. Trust the formula, though ...

If you don't trust the Massey formula, I'd venture to guess you don't trust much of anything. The guy does quality ratings of sports around the world with the same basic formula.

But that doesn't mean it works for Division III basketball.

It's obviously better than just random chance, but it's certainly not perfect either. It's somewhere in the middle meaning it can be useful as a tool but not as the only tool.
Just looking over Saturday's games, out of the 184 predicted winners, only 131 were correct or a 71.2% accuracy rate. Not exactly spectacular.

LOL wow. nothing like a totally out of context number to prove a point, I guess.

Ok, how large of a sample size would you like then?
Sunday 9 of 11, Saturday 131 of 184, Friday 29 of 37, Thursday 10 of 11, Wednesday 87 of 124, Tuesday 14 of 22, Monday 2 of 2

So to start January off it's 282 of 391... a whopping 72.1%.

I guess the question is, do you consider that predictive rate to be indicative of a model that is any better than what a D3 voter or follower could muster? In my opinion it's not but I'd like to hear your thought.

[me]

I have no idea why but the message wouldn't send without deleting the other stuff.

We're not even to the point where sample size matters, because no one has any idea if 72% is actually good or not. I don't.

I actually just added a bit to my previous post before seeing you posted this. Basically if the probability of a given team winning was 55%, then it's pretty hard to knock the computer for getting 5 or 6 out of 10 those right over a significant period. Bigger problem would be if it was getting 3 or 8 of 10.

Just quickly counting up the 90%+'s, 37 of 39 of those came in, for 94.8%. That is pretty spot on assuming that the distribution between 90-100 is uniform.

[FCGrizzliesGrad]

In the interest of trying to be helpful rather than just critical, one way to evaluate predictions might be to group all the games where the ratings said the prob was 60-40 or less for either team. And then see what the results were. Same for 70-30, etc.

Another way would be compare the ratings prediction accuracy to other methods.

I have a feeling 71% for the whole country is just fine.

Here's the winning percentage for all of the games Jan 1-7. The 50/50 games did have a chosen winner for each (likely a 50.2%/49.8% type of split) so I've listed whether it was the favorite or underdog

4%-1; 8%-1; 10%-1;
90%-5; 91%-10; 92%-8; 93%-5; 94%-5; 95%-7; 96%-8; 97%-6; 98%-8; 99%-9; 100%-5

11%-1; 12%-2; 13%-1; 15%-1; 16%-1; 17%-3; 18%-1; 19%-1; 20%-1
80%-8; 81%-8; 82%-2; 83%-3; 84%-5; 85%-9; 86%-9; 87%-4; 88%-5; 89%-6

21%-3; 22%-4; 23%-4; 25%-4; 26%-3; 27%-1; 28%- 3; 29%-1; 30%-4
70%-4; 71%-5; 72%-13; 73%-6; 74%-8; 75%-6; 76%-9; 77%-2; 78%-7; 79%-7

31%-2; 32%-3; 33%-2; 34%-3; 35%-4; 36%-3; 37%-2; 38%-3; 40%-1
60%-5; 61%-3; 62%-1; 63%-8; 64%-5; 65%-4; 66%-4; 67%-3; 68%-7; 69%-3

41%-5; 42%-1; 43%-6; 44%-8; 45%-4; 46%-1; 47%-4; 48%-2; 49%-4; 50%(underdog)-7
50%(favorite)-4; 51%-5; 52%-4; 53%-2; 54%-2; 55%-3; 56%-6; 57%-5; 58%-3; 59%-5

90+ favorites win 76 of 79... 96.2%
80-89 favorites win 59 of 71... 83.1%
70-79 favorites win 67 of 94... 71.3%
60-69 favorites win 43 of 66... 65.2%
50-59 favorites win 39 of 81... 48.1%

76+ favorites win 160 of 186...86.0%
50-75 favorites win 124 of 205... 60.5%

[me]

So all of that seems pretty close to what you'd expect, given the 1 week sample.

I hope you didn't count all that up by hand and then tabulate it by 1% buckets. The 10% summaries would have been more than sufficent I think, and then just assume there aren't a lot more 89%s than 81%s, and figure for 80-89 you should see ~85%. Looking at just 1 week you're probably not going to see a number that precise anyway.

Looks like the 70s group was a little low, but that could just be something unusual about this week (either randomness, or something else that isn't random but isn't immediately evident, maybe coming off holidays has an effect on favorites - I don't know). But overall I think this looks pretty much as it should.

[y_jack_lok]

Whatever FCGrGr did to figure all that out is impressive. +1K.

[smedindy]

I guess the best thing is to compare that rate to KenPom.

One thing that Pat, Dave, etc. should consider is that expected results don't move the needle. So the inclusion of the Augie - Illinois game probably doesn't matter much, because it was expected.

If, say, a UMAC team would upset a top MIAC team, then....that would have a some effect.

[Dave 'd-mac' McHugh]

I guess the best thing is to compare that rate to KenPom.

One thing that Pat, Dave, etc. should consider is that expected results don't move the needle. So the inclusion of the Augie - Illinois game probably doesn't matter much, because it was expected.

If, say, a UMAC team would upset a top MIAC team, then....that would have a some effect.

My only curiosity is... despite comments that it doesn't matter much... is how much does it inflate (or not) their resume and thus the resume of those they play (i.e. like it would in the SOS math for OWP and OOWP; though, SOS is only DIII teams - so not a direct comparison).

Here is the deal: the comments made were voters need to look at Massey and Bennett because teams highly ranked there are not highly ranked in our poll and visa versa. The premise being those rankings are telling us things that we voters are not paying attention to or are not willing to factor in, thus we aren't correctly ranking teams.

My contention is I am not sure whether to believe the numbers especially when I find fault in some of the information. While I am ranking Augustana high and Massey is as well, I used Augustana as an example because it jumped right out at me. At the time I looked at them, Massey had Augustana at three losses when in reality it was two and I knew the rankings were factoring in the Illinois game. IF the system is factoring in Illinois, we do not know how much influence it is or isn't giving it. Thus, if they are doing that for Augustana, then I have to assume they are doing it for others AND it would be affecting opponents of those teams. That starts to raise questions in my head about the rankings I am looking at.

Is Augustana really that good in the ranking's eyes if they didn't include Illinois? How much would they go down or up in the rankings if the game wasn't included (and how much would it affect their opponents; remember, Augustana's schedule is also affected by being in contact with at least the Big Ten). What about other teams whose games are being considered and shouldn't? Which direction would they move and what numbers would go up or down accordingly?

With that in mind... why should voters be implored to use a ranking system that doesn't seem to be accurately ranking the teams when it is factoring games that aren't relevant or count.

Again... I am curious about this. I am trying not to attack the rankings. I am trying to better understand them if I am going to use (or trust) them any more than I am now.

If one wants me to use those as measuring sticks and I see a flaw I can't understand its impact, isn't a fair to question how accurate really is the ranking? Maybe I am over-selling Augustana myself. The lost again. I need to make an adjustment. However, if I should look at Massey and they keep them high or drops them a bit (what if) based on not necessarily accurate information, how am I to factor that in?

I hope that makes sense. I am just trying to grasp the concept of being told these rankings should be utilized as reliable information when I see a unreliability issue and the impact of that issue can't be truly explained or understood.