New ranking formula
- Jackson Harrison
- Senior Member
- Posts: 6573
- Joined: Sat Dec 31, 2022 17:25
- Location: England
- Contact:
New ranking formula
I have spent the last few weeks coming up with a new ranking formula and after research and exploring different systems I propose this https://teal-quintina-79.tiiny.site
Constructive feedback will be appreciated.
Constructive feedback will be appreciated.
-
- Senior Member
- Posts: 1725
- Joined: Thu Oct 18, 2007 02:22
It's simply Elo rating, nothing fancy... No need for so many importance factors, it already takes into account the strength of the teams anyway.Jackson Harrison wrote: ↑Thu Jun 15, 2023 17:55 I have spent the last few weeks coming up with a new ranking formula and after research and exploring different systems I propose this https://teal-quintina-79.tiiny.site
Constructive feedback will be appreciated.
And what you call F is called K factor,and you used too big number
- Jackson Harrison
- Senior Member
- Posts: 6573
- Joined: Sat Dec 31, 2022 17:25
- Location: England
- Contact:
It's based on ELO but modified to fit the UEFA CC. I wanted to weigh each round of each competition. What modifications would you suggest? I think an ELO rating like this is good and can be easily monitored and changes after each match but not by too much. If you loose to a minnow you loose points and the minnow gains them, this doesn't happen under the current system, for example currently Celtic would get the same points for beating Real Madrid or Ferencvaros in the group stage which is unfair IMO.amirbachar wrote: ↑Thu Jun 15, 2023 19:30It's simply Elo rating, nothing fancy... No need for so many importance factors, it already takes into account the strength of the teams anyway.Jackson Harrison wrote: ↑Thu Jun 15, 2023 17:55 I have spent the last few weeks coming up with a new ranking formula and after research and exploring different systems I propose this https://teal-quintina-79.tiiny.site
Constructive feedback will be appreciated.
And what you call F is called K factor,and you used too big number
-
- Senior Member
- Posts: 1725
- Joined: Thu Oct 18, 2007 02:22
Of course it's not like the current rankings... The is no reason for so many weights. Also Elo ratings for clubs exist on many websites, for exampleJackson Harrison wrote: ↑Thu Jun 15, 2023 19:47It's based on ELO but modified to fit the UEFA CC. I wanted to weigh each round of each competition. What modifications would you suggest? I think an ELO rating like this is good and can be easily monitored and changes after each match but not by too much. If you loose to a minnow you loose points and the minnow gains them, this doesn't happen under the current system, for example currently Celtic would get the same points for beating Real Madrid or Ferencvaros in the group stage which is unfair IMO.amirbachar wrote: ↑Thu Jun 15, 2023 19:30It's simply Elo rating, nothing fancy... No need for so many importance factors, it already takes into account the strength of the teams anyway.Jackson Harrison wrote: ↑Thu Jun 15, 2023 17:55 I have spent the last few weeks coming up with a new ranking formula and after research and exploring different systems I propose this https://teal-quintina-79.tiiny.site
Constructive feedback will be appreciated.
And what you call F is called K factor,and you used too big number
http://clubelo.com/
-
- Posts: 251
- Joined: Fri Jul 28, 2017 09:08
Since a few months we also have Opta Rankings which are also based on Elo system . The main advantage is that they literally cover every european domestic competion while clubelo does not.amirbachar wrote: ↑Sat Jun 17, 2023 16:21Of course it's not like the current rankings... The is no reason for so many weights. Also Elo ratings for clubs exist on many websites, for exampleJackson Harrison wrote: ↑Thu Jun 15, 2023 19:47It's based on ELO but modified to fit the UEFA CC. I wanted to weigh each round of each competition. What modifications would you suggest? I think an ELO rating like this is good and can be easily monitored and changes after each match but not by too much. If you loose to a minnow you loose points and the minnow gains them, this doesn't happen under the current system, for example currently Celtic would get the same points for beating Real Madrid or Ferencvaros in the group stage which is unfair IMO.amirbachar wrote: ↑Thu Jun 15, 2023 19:30
It's simply Elo rating, nothing fancy... No need for so many importance factors, it already takes into account the strength of the teams anyway.
And what you call F is called K factor,and you used too big number
http://clubelo.com/
Having studied and analysed clubelo ratings extensively I would say they are more accurate than Opta, but they are only limited to the 28/55 european countries.
Elo ratings of all remaining countries are far from being accurate and lot of teams playing in european qualifiers this season do not even have a clubelo rating.
Another issue which is not discussed enough is the inter-league adjustments. When two teams from different countries face each other, how do we adjust their ratings and also those of all other teams from that country. Also considering how many few matches teams from Andorra, San Marino, etc are playing in a season.
Last but not the least: We must be careful with using weights (and FIFA should too!). The outcome of a match is a combination of actual strenght difference but also randomness. And a very high weight would give too much importance to random events
-
- Senior Member
- Posts: 1725
- Joined: Thu Oct 18, 2007 02:22
Yeah, I know Opta exist, don't think they publish the full rankings, right?elkjiaer is back wrote: ↑Tue Jun 20, 2023 10:51Since a few months we also have Opta Rankings which are also based on Elo system . The main advantage is that they literally cover every european domestic competion while clubelo does not.amirbachar wrote: ↑Sat Jun 17, 2023 16:21Of course it's not like the current rankings... The is no reason for so many weights. Also Elo ratings for clubs exist on many websites, for exampleJackson Harrison wrote: ↑Thu Jun 15, 2023 19:47
It's based on ELO but modified to fit the UEFA CC. I wanted to weigh each round of each competition. What modifications would you suggest? I think an ELO rating like this is good and can be easily monitored and changes after each match but not by too much. If you loose to a minnow you loose points and the minnow gains them, this doesn't happen under the current system, for example currently Celtic would get the same points for beating Real Madrid or Ferencvaros in the group stage which is unfair IMO.
http://clubelo.com/
Having studied and analysed clubelo ratings extensively I would say they are more accurate than Opta, but they are only limited to the 28/55 european countries.
Elo ratings of all remaining countries are far from being accurate and lot of teams playing in european qualifiers this season do not even have a clubelo rating.
Another issue which is not discussed enough is the inter-league adjustments. When two teams from different countries face each other, how do we adjust their ratings and also those of all other teams from that country. Also considering how many few matches teams from Andorra, San Marino, etc are playing in a season.
Last but not the least: We must be careful with using weights (and FIFA should too!). The outcome of a match is a combination of actual strenght difference but also randomness. And a very high weight would give too much importance to random events
Both clubelo and Opta do inter-lesgue adjusment as they should, you are correct that it's important for accurate rankings with small amount if international matches.
The best public rankings are 538 BTW, they are partially Elo based, but also xG, xT and budget based. All of these together make rankings that are almost competitive with betting sites.
There is also this website with all the leagues in it:
https://www.euroclubindex.com/
Last edited by amirbachar on Tue Jun 20, 2023 17:57, edited 1 time in total.
-
- Senior Member
- Posts: 1725
- Joined: Thu Oct 18, 2007 02:22
And yeah, the weight should be based on a research of what would produce the best performance of the rankings. That's what clubelo did.elkjiaer is back wrote: ↑Tue Jun 20, 2023 10:51Since a few months we also have Opta Rankings which are also based on Elo system . The main advantage is that they literally cover every european domestic competion while clubelo does not.amirbachar wrote: ↑Sat Jun 17, 2023 16:21Of course it's not like the current rankings... The is no reason for so many weights. Also Elo ratings for clubs exist on many websites, for exampleJackson Harrison wrote: ↑Thu Jun 15, 2023 19:47
It's based on ELO but modified to fit the UEFA CC. I wanted to weigh each round of each competition. What modifications would you suggest? I think an ELO rating like this is good and can be easily monitored and changes after each match but not by too much. If you loose to a minnow you loose points and the minnow gains them, this doesn't happen under the current system, for example currently Celtic would get the same points for beating Real Madrid or Ferencvaros in the group stage which is unfair IMO.
http://clubelo.com/
Having studied and analysed clubelo ratings extensively I would say they are more accurate than Opta, but they are only limited to the 28/55 european countries.
Elo ratings of all remaining countries are far from being accurate and lot of teams playing in european qualifiers this season do not even have a clubelo rating.
Another issue which is not discussed enough is the inter-league adjustments. When two teams from different countries face each other, how do we adjust their ratings and also those of all other teams from that country. Also considering how many few matches teams from Andorra, San Marino, etc are playing in a season.
Last but not the least: We must be careful with using weights (and FIFA should too!). The outcome of a match is a combination of actual strenght difference but also randomness. And a very high weight would give too much importance to random events
For FIFA for example, the weights are way too small. It means the rankings would soon be basically long term rankings rather than current strength.
-
- Posts: 251
- Joined: Fri Jul 28, 2017 09:08
For Opta rankings you can find them here:amirbachar wrote: ↑Tue Jun 20, 2023 17:44Yeah, I know Opta exist, don't think they publish the full rankings, right?elkjiaer is back wrote: ↑Tue Jun 20, 2023 10:51Since a few months we also have Opta Rankings which are also based on Elo system . The main advantage is that they literally cover every european domestic competion while clubelo does not.amirbachar wrote: ↑Sat Jun 17, 2023 16:21
Of course it's not like the current rankings... The is no reason for so many weights. Also Elo ratings for clubs exist on many websites, for example
http://clubelo.com/
Having studied and analysed clubelo ratings extensively I would say they are more accurate than Opta, but they are only limited to the 28/55 european countries.
Elo ratings of all remaining countries are far from being accurate and lot of teams playing in european qualifiers this season do not even have a clubelo rating.
Another issue which is not discussed enough is the inter-league adjustments. When two teams from different countries face each other, how do we adjust their ratings and also those of all other teams from that country. Also considering how many few matches teams from Andorra, San Marino, etc are playing in a season.
Last but not the least: We must be careful with using weights (and FIFA should too!). The outcome of a match is a combination of actual strenght difference but also randomness. And a very high weight would give too much importance to random events
Both clubelo and Opta do inter-lesgue adjusment as they should, you are correct that it's important for accurate rankings with small amount if international matches.
The best public rankings are 538 BTW, they are partially Elo based, but also xG, xT and budget based. All of these together make rankings that are almost competitive with betting sites.
There is also this website with all the leagues in it:
https://www.euroclubindex.com/
https://theanalyst.com/eu/2023/06/who-a ... -rankings/
Unfortunately they are not active anymore.amirbachar wrote: ↑Tue Jun 20, 2023 17:44 The best public rankings are 538 BTW, they are partially Elo based, but also xG, xT and budget based. All of these together make rankings that are almost competitive with betting sites.
-
- Senior Member
- Posts: 1725
- Joined: Thu Oct 18, 2007 02:22
Ogiman wrote: ↑Thu Jun 22, 2023 18:29Unfortunately they are not active anymore.amirbachar wrote: ↑Tue Jun 20, 2023 17:44 The best public rankings are 538 BTW, they are partially Elo based, but also xG, xT and budget based. All of these together make rankings that are almost competitive with betting sites.
Of course they are:
https://projects.fivethirtyeight.com/so ... -rankings/
If you open this page via mobile, at the top of it you can see this announcement:amirbachar wrote: ↑Thu Jun 22, 2023 20:55Ogiman wrote: ↑Thu Jun 22, 2023 18:29Unfortunately they are not active anymore.amirbachar wrote: ↑Tue Jun 20, 2023 17:44 The best public rankings are 538 BTW, they are partially Elo based, but also xG, xT and budget based. All of these together make rankings that are almost competitive with betting sites.
Of course they are:
https://projects.fivethirtyeight.com/so ... -rankings/
"This page is frozen as of June 13, 2023, and will no longer be updated."
-
- Senior Member
- Posts: 1725
- Joined: Thu Oct 18, 2007 02:22
Oh wow, it really is the end of an era. Can't they at least open source it? That's a shame we will no longer have it.Ogiman wrote: ↑Fri Jun 23, 2023 10:36If you open this page via mobile, at the top of it you can see this announcement:amirbachar wrote: ↑Thu Jun 22, 2023 20:55
Of course they are:
https://projects.fivethirtyeight.com/so ... -rankings/
"This page is frozen as of June 13, 2023, and will no longer be updated."
-
- Posts: 371
- Joined: Thu May 16, 2019 00:45
Noooo! I really enjoyed following their predictions. Much more accurate than euroclubindex (perhaps less than clubelo but clubelo stopped doing league and cup predictions).amirbachar wrote: ↑Sat Jun 24, 2023 02:11Oh wow, it really is the end of an era. Can't they at least open source it? That's a shame we will no longer have it.Ogiman wrote: ↑Fri Jun 23, 2023 10:36If you open this page via mobile, at the top of it you can see this announcement:amirbachar wrote: ↑Thu Jun 22, 2023 20:55
Of course they are:
https://projects.fivethirtyeight.com/so ... -rankings/
"This page is frozen as of June 13, 2023, and will no longer be updated."
Quite sad about this news