East Region Fan Poll

Started by pg04, July 05, 2007, 09:44:54 PM

Previous topic - Next topic

0 Members and 2 Guests are viewing this topic.


MRMIKESMITH

Quote from: HansenRatings on October 15, 2019, 12:13:16 PM
Quote from: Bombers798891 on October 15, 2019, 11:53:58 AM
Quote from: HansenRatings on October 15, 2019, 11:35:23 AM
Quote from: Bombers798891 on October 15, 2019, 11:25:36 AM
Quote from: HansenRatings on October 15, 2019, 11:10:55 AM

With that in mind, how should you compare a team from the UMAC, a very bad conference in the Midwest, to a team from the E8, a very good conference in the Northeast? Or a team from the ECFC to a team from the WIAC? They never play each other! 

Or do I devise a system that...only increases their standing relative to the rest of the country if they actually do better against the rest of the country?


These two statements seem to sort of run counter to each other. The only way for a team's standing to go up or down is if they...do something they're almost never given the chance to?

And just to clarify, I don't really have an issue with IC's ranking, or the East's in general. But I can't remember the last time IC even played a team from another region. Either Frostburg in 2014 or Moravian in 2013, I'd guess?

I used this analogy on that Twitter thread someone posted earlier, but I view my model like a heads-up poker tournament that started in 1997. Everyone had the same number of chips at the beginning, and then by playing against each other, some rooms (regions/conferences/whatever) of the casino have people with larger purses than other rooms. If I'm in the East region, I can increase my own personal pot by outperforming my opponents, but for my room's total purse to increase, someone in that room (it doesn't have to be me) has to get more chips from some other room. Then, when they come back to my room, there's more total chips for everyone.

Right. Except it's not heads up poker if you never play against someone whose chips you can take

But if I you and I are in the same room, and play against each other frequently, and you go play in a different room, I can take some of your winnings from you the next time we play!

Correct me if I'm wrong, if Team A loses to Team B 7 out of the last 10 years. However, Team A beats Team B in year 11 and both share similar record, Team B could possibly be still rank higher? Going a little deeper, if Team C beats Team B 10 out of 10, but in the current year have a worst record than both Team A and Team B, Team C still has a chance to be ranked higher. So essentially, as mentioned the ranking is basically a rolling total of data from 1997 and should not be used for current events, because current events are weighted very low.

HansenRatings

Quote from: MANDGSU on October 15, 2019, 01:49:48 PM
Quote from: HansenRatings on October 15, 2019, 12:13:16 PM
Quote from: Bombers798891 on October 15, 2019, 11:53:58 AM
Quote from: HansenRatings on October 15, 2019, 11:35:23 AM
Quote from: Bombers798891 on October 15, 2019, 11:25:36 AM
Quote from: HansenRatings on October 15, 2019, 11:10:55 AM

With that in mind, how should you compare a team from the UMAC, a very bad conference in the Midwest, to a team from the E8, a very good conference in the Northeast? Or a team from the ECFC to a team from the WIAC? They never play each other! 

Or do I devise a system that...only increases their standing relative to the rest of the country if they actually do better against the rest of the country?


These two statements seem to sort of run counter to each other. The only way for a team's standing to go up or down is if they...do something they're almost never given the chance to?

And just to clarify, I don't really have an issue with IC's ranking, or the East's in general. But I can't remember the last time IC even played a team from another region. Either Frostburg in 2014 or Moravian in 2013, I'd guess?

I used this analogy on that Twitter thread someone posted earlier, but I view my model like a heads-up poker tournament that started in 1997. Everyone had the same number of chips at the beginning, and then by playing against each other, some rooms (regions/conferences/whatever) of the casino have people with larger purses than other rooms. If I'm in the East region, I can increase my own personal pot by outperforming my opponents, but for my room's total purse to increase, someone in that room (it doesn't have to be me) has to get more chips from some other room. Then, when they come back to my room, there's more total chips for everyone.

Right. Except it's not heads up poker if you never play against someone whose chips you can take

But if I you and I are in the same room, and play against each other frequently, and you go play in a different room, I can take some of your winnings from you the next time we play!

Correct me if I'm wrong, if Team A loses to Team B 7 out of the last 10 years. However, Team A beats Team B in year 11 and both share similar record, Team B could possibly be still rank higher? Going a little deeper, if Team C beats Team B 10 out of 10, but in the current year have a worst record than both Team A and Team B, Team C still has a chance to be ranked higher. So essentially, as mentioned the ranking is basically a rolling total of data from 1997 and should not be used for current events, because current events are weighted very low.

That's kinda correct and kinda backwards, actually. That scenario can easily play out, depending on the other 8 games those teams played. If you're trying to build a predictive system, trying to rank teams strictly according to the rule of thumb "teams should be ranked ahead of teams they beat" is actually just about the worst system. In my (now defunct) prediction tracker, the "Nutshell Retrodictive Rankings" was designed with that premise, and it ranked dead last in predictions each season I tracked.

In my model, about half of a team's rating is directly attributable to their last 4 games, and then 1/4 to the 4 before that, and 1/8 to the the 4 before that, and on & on... So, a single random game from a previous season will make up about 1% to 2% of a team's current rating, but you add up a lot of those games over the course of a decade plus, and if an entire region consistently plays poorly against outside opponents over that whole span, that can add up to a lot of adjustments.
Follow me on Twitter. I post fun graphs sometimes. @LogHanRatings

Bartman

Quote from: Machiavelli on October 15, 2019, 01:18:15 PM

Hansen looking at Oline, as he reconsiders the D3 football ranking algorithm?
"I never graduated from Iowa, but I was only there for two terms - Truman's and Eisenhower's."
Alex Karras
"When it's third and ten, you can take the milk drinkers and I'll take the whiskey drinkers every time."
Max McGee

MRMIKESMITH

Quote from: HansenRatings on October 15, 2019, 02:11:05 PM
Quote from: MANDGSU on October 15, 2019, 01:49:48 PM
Quote from: HansenRatings on October 15, 2019, 12:13:16 PM
Quote from: Bombers798891 on October 15, 2019, 11:53:58 AM
Quote from: HansenRatings on October 15, 2019, 11:35:23 AM
Quote from: Bombers798891 on October 15, 2019, 11:25:36 AM
Quote from: HansenRatings on October 15, 2019, 11:10:55 AM

With that in mind, how should you compare a team from the UMAC, a very bad conference in the Midwest, to a team from the E8, a very good conference in the Northeast? Or a team from the ECFC to a team from the WIAC? They never play each other! 

Or do I devise a system that...only increases their standing relative to the rest of the country if they actually do better against the rest of the country?


These two statements seem to sort of run counter to each other. The only way for a team's standing to go up or down is if they...do something they're almost never given the chance to?

And just to clarify, I don't really have an issue with IC's ranking, or the East's in general. But I can't remember the last time IC even played a team from another region. Either Frostburg in 2014 or Moravian in 2013, I'd guess?

I used this analogy on that Twitter thread someone posted earlier, but I view my model like a heads-up poker tournament that started in 1997. Everyone had the same number of chips at the beginning, and then by playing against each other, some rooms (regions/conferences/whatever) of the casino have people with larger purses than other rooms. If I'm in the East region, I can increase my own personal pot by outperforming my opponents, but for my room's total purse to increase, someone in that room (it doesn't have to be me) has to get more chips from some other room. Then, when they come back to my room, there's more total chips for everyone.

Right. Except it's not heads up poker if you never play against someone whose chips you can take

But if I you and I are in the same room, and play against each other frequently, and you go play in a different room, I can take some of your winnings from you the next time we play!

Correct me if I'm wrong, if Team A loses to Team B 7 out of the last 10 years. However, Team A beats Team B in year 11 and both share similar record, Team B could possibly be still rank higher? Going a little deeper, if Team C beats Team B 10 out of 10, but in the current year have a worst record than both Team A and Team B, Team C still has a chance to be ranked higher. So essentially, as mentioned the ranking is basically a rolling total of data from 1997 and should not be used for current events, because current events are weighted very low.

That's kinda correct and kinda backwards, actually. That scenario can easily play out, depending on the other 8 games those teams played. If you're trying to build a predictive system, trying to rank teams strictly according to the rule of thumb "teams should be ranked ahead of teams they beat" is actually just about the worst system. In my (now defunct) prediction tracker, the "Nutshell Retrodictive Rankings" was designed with that premise, and it ranked dead last in predictions each season I tracked.

In my model, about half of a team's rating is directly attributable to their last 4 games, and then 1/4 to the 4 before that, and 1/8 to the the 4 before that, and on & on... So, a single random game from a previous season will make up about 1% to 2% of a team's current rating, but you add up a lot of those games over the course of a decade plus, and if an entire region consistently plays poorly against outside opponents over that whole span, that can add up to a lot of adjustments.

Ah I see. So going back to my example. Wesley has beaten Salisbury 7 out 10 and UWW has beaten Wesley 7 out of 7 and UWO has beaten UWW 2 out of 5, thus UW-O being ranked higher than Salisbury currently although current H2H says otherwise. I get the poker room theory as Wesley has been taking Salisbury poke chips over the years but given most to UWW, while UWO has taken some of UWW poke chips, thus accumulating more than Salisbury over that time as both teams have accumulated chips from their respective conference opponents.

Oline89

Quote from: MANDGSU on October 15, 2019, 03:19:14 PM
Quote from: HansenRatings on October 15, 2019, 02:11:05 PM
Quote from: MANDGSU on October 15, 2019, 01:49:48 PM
Quote from: HansenRatings on October 15, 2019, 12:13:16 PM
Quote from: Bombers798891 on October 15, 2019, 11:53:58 AM
Quote from: HansenRatings on October 15, 2019, 11:35:23 AM
Quote from: Bombers798891 on October 15, 2019, 11:25:36 AM
Quote from: HansenRatings on October 15, 2019, 11:10:55 AM

With that in mind, how should you compare a team from the UMAC, a very bad conference in the Midwest, to a team from the E8, a very good conference in the Northeast? Or a team from the ECFC to a team from the WIAC? They never play each other! 

Or do I devise a system that...only increases their standing relative to the rest of the country if they actually do better against the rest of the country?


These two statements seem to sort of run counter to each other. The only way for a team's standing to go up or down is if they...do something they're almost never given the chance to?

And just to clarify, I don't really have an issue with IC's ranking, or the East's in general. But I can't remember the last time IC even played a team from another region. Either Frostburg in 2014 or Moravian in 2013, I'd guess?

I used this analogy on that Twitter thread someone posted earlier, but I view my model like a heads-up poker tournament that started in 1997. Everyone had the same number of chips at the beginning, and then by playing against each other, some rooms (regions/conferences/whatever) of the casino have people with larger purses than other rooms. If I'm in the East region, I can increase my own personal pot by outperforming my opponents, but for my room's total purse to increase, someone in that room (it doesn't have to be me) has to get more chips from some other room. Then, when they come back to my room, there's more total chips for everyone.

Right. Except it's not heads up poker if you never play against someone whose chips you can take

But if I you and I are in the same room, and play against each other frequently, and you go play in a different room, I can take some of your winnings from you the next time we play!

Correct me if I'm wrong, if Team A loses to Team B 7 out of the last 10 years. However, Team A beats Team B in year 11 and both share similar record, Team B could possibly be still rank higher? Going a little deeper, if Team C beats Team B 10 out of 10, but in the current year have a worst record than both Team A and Team B, Team C still has a chance to be ranked higher. So essentially, as mentioned the ranking is basically a rolling total of data from 1997 and should not be used for current events, because current events are weighted very low.

That's kinda correct and kinda backwards, actually. That scenario can easily play out, depending on the other 8 games those teams played. If you're trying to build a predictive system, trying to rank teams strictly according to the rule of thumb "teams should be ranked ahead of teams they beat" is actually just about the worst system. In my (now defunct) prediction tracker, the "Nutshell Retrodictive Rankings" was designed with that premise, and it ranked dead last in predictions each season I tracked.

In my model, about half of a team's rating is directly attributable to their last 4 games, and then 1/4 to the 4 before that, and 1/8 to the the 4 before that, and on & on... So, a single random game from a previous season will make up about 1% to 2% of a team's current rating, but you add up a lot of those games over the course of a decade plus, and if an entire region consistently plays poorly against outside opponents over that whole span, that can add up to a lot of adjustments.

Ah I see. So going back to my example. Wesley has beaten Salisbury 7 out 10 and UWW has beaten Wesley 7 out of 7 and UWO has beaten UWW 2 out of 5, thus UW-O being ranked higher than Salisbury currently although current H2H says otherwise. I get the poker room theory as Wesley has been taking Salisbury poke chips over the years but given most to UWW, while UWO has taken some of UWW poke chips, thus accumulating more than Salisbury over that time as both teams have accumulated chips from their respective conference opponents.

That is where the debate ensues.  Is  a 2019 UWO team "better" than the Salisbury team?  Based on game theory (Hansen), the 2019 UWO team should win a head to head matchup. 

HansenRatings

Quote from: Oline89 on October 15, 2019, 03:24:45 PM
Quote from: MANDGSU on October 15, 2019, 03:19:14 PM
Ah I see. So going back to my example. Wesley has beaten Salisbury 7 out 10 and UWW has beaten Wesley 7 out of 7 and UWO has beaten UWW 2 out of 5, thus UW-O being ranked higher than Salisbury currently although current H2H says otherwise. I get the poker room theory as Wesley has been taking Salisbury poke chips over the years but given most to UWW, while UWO has taken some of UWW poke chips, thus accumulating more than Salisbury over that time as both teams have accumulated chips from their respective conference opponents.

That is where the debate ensues.  Is  a 2019 UWO team "better" than the Salisbury team?  Based on game theory (Hansen), the 2019 UWO team should win a head to head matchup. 

Here's the basic thought: Given the results (not just Win/Loss) of every game each team has played over the last decade plus, and giving optimal weight to recent results relative to past results, UWO is more likely to beat Salisbury than vice-versa. It's important to note that the better team doesn't always win. This is why I ended my first post on here saying my model probably isn't the best source if you're trying to rank resumes for this season. Projections are also probabilistic. If a team is given a 55% chance to win, it should lose 9 out of 20 matchups against that opponent.
Follow me on Twitter. I post fun graphs sometimes. @LogHanRatings

Doid23

Quote from: HansenRatings on October 15, 2019, 12:13:16 PM


But if I you and I are in the same room, and play against each other frequently, and you go play in a different room, I can take some of your winnings from you the next time we play!
Appreciate that you take the time to do this, and thanks for sharing some of the insight into your model, and your awesome explanations. +K

MRMIKESMITH

Rank      Team             Previous    Points   
1)     Ithaca  (6)            1         60   
2)     Salisbury              4         53
3)     Union                  5         46
4)     Cortland               2         44
5)     Delaware Valley        6         30
6t)     Hobart                 7         27
6t)     Wesley                 3         27           
8)     Brockport              NR        23
9)     WPI                    8         13 
10)     Alfred                 NR        4


Dropped Out: RPI

RV) Wilkes (3)

Key Game(s): Cortland #4 v. Alfred #10 , Ithaca #1 v. Hobart #6t, Delaware Valley #5 v. Wilkes #11

Voters: MANDGSU, RAMS1102, DUTCH BOY, Ufanbill, Oline89, Bartman

Ice Bear

Quote from: Machiavelli on October 15, 2019, 01:18:15 PM


"In my club I will splash the pot any ****ing time I please." Teddy KGB aka Mount Union
A long time fan of DIII Football!

rams1102

The only difference in my Poll and the new standings is that I had:

3. Cortland
4. Del Val
5. Union

Not sold on Union as #3. I know you can only play the cards dealt but the last 3 games of their schedule will really tell when they play Ithaca, Utica and RPI. In any event they are having a great season. I think at the end of the year there will be some disappointed Teams not making the Dance. There will definitely be a lot of great discussions to come.
It ain't over till it's over, and when you get to the fork in the road, take it.

UfanBill

Quote from: rams1102 on October 16, 2019, 01:14:06 PM
The only difference in my Poll and the new standings is that I had:

3. Cortland
4. Del Val
5. Union

Not sold on Union as #3. I know you can only play the cards dealt but the last 3 games of their schedule will really tell when they play Ithaca, Utica and RPI. In any event they are having a great season. I think at the end of the year there will be some disappointed Teams not making the Dance. There will definitely be a lot of great discussions to come.

This week I moved Union ahead of Cortland. Why? because I'm a Union fan and I am sold on them.  They are talented, well balanced and well coached and have met every challenge so far with relative ease and barring a trap game let down will be 7-0 when they meet their next big challenge, Ithaca week 8... Cortland...I have no issues with them. I believe they will win the E8 but I also believe the LL is stronger than the E8 this year. I ask why you have Del Val ranked ahead of Wesley even though the Wolverines won the head to head? Do you really still think the MAC and NJAC are stronger than the NY leagues this year? I don't.
"You don't stop playing because you got old, you got old because you stopped playing" 🏈🏀⚾🎿⛳

rams1102

Nothing wrong in being a homer especially when your team is undefeated. Wesley won I think in double OT and really should have lost IMHO. Also My Top 5 have 3 NY Teams and my top 10 have 6 NY Teams. Then the NJAC & MAC are 3 Teams. I can't understand why you think I feel they are stronget than the NY leagues. I don't. Thats why we can agree to disagree. :)
It ain't over till it's over, and when you get to the fork in the road, take it.

D O.C.

<< "In my club I will splash the pot any ****ing time I please." Teddy KGB aka Mount Union >>
😁😆😃😄

jmcozenlaw

Quote from: rams1102 on October 17, 2019, 01:47:37 PM
Nothing wrong in being a homer especially when your team is undefeated. Wesley won I think in double OT and really should have lost IMHO. Also My Top 5 have 3 NY Teams and my top 10 have 6 NY Teams. Then the NJAC & MAC are 3 Teams. I can't understand why you think I feel they are stronget than the NY leagues. I don't. Thats why we can agree to disagree. :)

It was actually 4 OT's. Each team could only muster 3 points in regulation time. DelVal's kicker missed a gimme at the end of the first half or I believe that game ends 6-3. The final score impacts both teams defensive stats (especially PPG)..................a ridiculous inclusion that some have tried to figure out a way to change. Imagine had this gone 8 OT's with a final of 52-50. Talk about screwed up season long defensive (and offensive) stats!!