Tournament ELO

[n0brain]

the problem is that these are NOT to be compared to voobly or hd ratings. on the contrary i would make it even more different and do #1 as zero elo and everyone else going lower from there. elo lists with different datasets shouldnt give the false impression of comparability / similarity. a static elo list like this is only meant to show strength difference between players, not comparability with different datasets. if u want that than elo is just the wrong measurement formula to begin with.

 

[Michaerbse]

Hi @mungo__ ,

Two things about MSC:
1) ACCM vs. dogao was an AW, so the result should not be counted for the elo.
2) As far as I can tell the rating has ignored death match tournaments so far as death match is really a different discipline. However, in MSC game 5 is played as death match. Could g5 be ignored for the rating?

Keep up the good work!

 

[mungo__]

Hi everybody!

Some changes:

• A small section on the tournament pages list the sources used (mostly the aoezone threads)
• Some bugfixes, as the incorrect "Placings" on the player pages
• Thx @Michaerbse, fixed the AW. I am not 100% sure about the DM games. They should not count in general. However, this would really mess up the results.... Maybe we let them count as an exception, as its only 1/5 games? I am glad to hear other opinions on this.

As always, thanks for your participation!

Best
mungo

 

[chicklet]

Hi @mungo__

Is there a reason the rating jumps around (after 36)? This seems to occur multiple times.

I'd also like to comment on the use of elo decay.
In the voobly ladder it works fine - everyone has the potential to login and play some games to avoid decay - but tournament elo is different. Lots of tournaments are invite only or have a very limited qualification phase. This limits the ability for players to avoid decay for inactivity ; particularly players outside the top 8, 16 and 32. Eventually I can see the list producing big elo cliffs between those top 8, 16 and 32 players and the rest.

To illustrate my point - from Spring (currently ranked 50) down, all players have racked up - 20 points of decay. It is possible that non of these players have the opportunity to play in a competitive tournament over the next 18 months and so all of their elo decays to 1800.

I raise these issues because I appreciate your work.

 

[mungo__]

Hi @chicklet,

Is there a reason the rating jumps around (after 36)? This seems to occur multiple times.

Good spot! The reason is that we currently hide players with < 5 series or so. They occupy the numbers you dont see, and this happens especially often at the rating 1900, as this is the *starting elo* for most players. It's not pretty, but not exactly a bug. We should probably not count them in the overall ranking, they should have an "unknown" rank #.

I'd also like to comment on the use of elo decay.
[...]
Eventually I can see the list producing big elo cliffs between those top 8, 16 and 32 players and the rest.

Very good point. The decay is definetly an unresolved issue. I currently sympathise with starting the decay later, for example after 6 months of inactivity, and just marking the inactive players with a badge after, say, 3 months. This would make clear that for example Riut is probably not actually world rank #7, but also prevent players like Chris to remain high after years of inactivity...

 

[robdyrob]

On some chess servers, the elo change rate depends on the activity of the players using a somehow time-dependent k-factor. In my opinion, this would be better than a constant decay. If somebody is several month inactive and can't compete anymore, he simply loses a lot of points after a first round defeat in a tournament. It's probably more difficult to implement, though. Especially because there are extremely few evaluated games compared to chess.

 

[chicklet]

Hi @chicklet,


Good spot! The reason is that we currently hide players with < 5 series or so. They occupy the numbers you dont see, and this happens especially often at the rating 1900, as this is the *starting elo* for most players. It's not pretty, but not exactly a bug. We should probably not count them in the overall ranking, they should have an "unknown" rank #.

Very good point. The decay is definetly an unresolved issue. I currently sympathise with starting the decay later, for example after 6 months of inactivity, and just marking the inactive players with a badge after, say, 3 months. This would make clear that for example Riut is probably not actually world rank #7, but also prevent players like Chris to remain high after years of inactivity...

Hi @chicklet,


Good spot! The reason is that we currently hide players with < 5 series or so. They occupy the numbers you dont see, and this happens especially often at the rating 1900, as this is the *starting elo* for most players. It's not pretty, but not exactly a bug. We should probably not count them in the overall ranking, they should have an "unknown" rank #.



Very good point. The decay is definetly an unresolved issue. I currently sympathise with starting the decay later, for example after 6 months of inactivity, and just marking the inactive players with a badge after, say, 3 months. This would make clear that for example Riut is probably not actually world rank #7, but also prevent players like Chris to remain high after years of inactivity...

Sound like very sensible solutions to an awkward problem.

 

[IRC_tomate]

Hi, I would actually like to hear the feedback of the players included on the list, as I have seen recently this ELO could have been used as seeding for a tournament and avoid the qualification stage entirely (still no confirmation from the host), do actually pro players support the site?

As for my personal opinion, to include some of the tournaments you did feels horrible and unfair, just to name one: king of the realm

 

[snellefiets]

I think the site is great. I would be even better if there would be an option to filter on time. So for example if you filter on 2018-2019 you will see different names in the top 10 than filter on 2015-2017. If you could implement that the issue of no decay of ELO would also move to the background, since you could easily filter on "current" ELO (like last 24 months) and "historic" ELO (including all tournaments and results).

 

[Michaerbse]

Hi, I would actually like to hear the feedback of the players included on the list, as I have seen recently this ELO could have been used as seeding for a tournament and avoid the qualification stage entirely (still no confirmation from the host), do actually pro players support the site?

I don't think it should be used to avoid the qualification entirely as this would make it impossible for new players to ever qualify for a pro tournament. Non-pro tournaments are not listed so where to start?
I do hope that it's gonna be considered for seeding in the mid-term future though. Combining voobly 1v1 rating with tournament elo could already be an option.

As for my personal opinion, to include some of the tournaments you did feels horrible and unfair, just to name one: king of the realm

Agree about King of the Realms, not sure about others. Would you have more examples?

I think the site is great. I would be even better if there would be an option to filter on time. So for example if you filter on 2018-2019 you will see different names in the top 10 than filter on 2015-2017. If you could implement that the issue of no decay of ELO would also move to the background, since you could easily filter on "current" ELO (like last 24 months) and "historic" ELO (including all tournaments and results).

I believe mungo mentioned that this is currently not feasible as all numbers would have to be calculated on the fly, which causes performance issues (remember that each rating has an indirect influence on pretty much all others).
I agree that the decay discussion is an issue when trying to accurately describe the current top x players. Not sure whether it's a big deal for seeding though. Inactive players usually don't play tourneys by definition so their rating doesn't matter. In the worst case it would act like a "wildcard" for deserved players coming back like JorDan or RiuT...

 

[pumukel]

To be consistent with the 1v1s of the ECL LAN, all RM 1v1s from major TG tournaments of the last years are now added:

2011 World Clan League 7: https://dev.aoe-elo.com/tournament/100/World-Clan-League-7

JorDan_23 +10, Alive +9, Ruso +9, Tim +9, Surya +9 (who is this?)

2013 The Medieval Wars: https://dev.aoe-elo.com/tournament/101/The-Medieval-Wars

JorDan_23 +13, Chris +12, Tim +11

2014 Saiyanhood Legends United: https://dev.aoe-elo.com/tournament/102/Saiyanhood-Legends-United

Nicov +12, TheViper +10, Juan +9, Edie +9

2015 War is Coming: https://dev.aoe-elo.com/tournament/103/War-is-Coming

Yo +16, JorDan_23 +14, miguel +11

This caused some changes in the Hall of Fame, e.g. TheViper reached 2300:

 

[Influenza]

_Surya_ is slam i think

 

[pumukel]

_Surya_ is slam i think

Yes, a quick AoEZone search seems to confirm this. Thanks! I merged the data.

 

[siestes]

Every player who reached 2030 ELO or above on aoe-elo.com (a total of 46 players) (granted I didn't miss anybody, which is really possible)

Player	Date of ELO peak	ELO peak
TheViper	13 Apr 2019	2315
RiuT	07 Mar 2016	2208
Chris	01 May 2008	2207
TaToH	05 Jan 2020	2204
Hera	22 Mar 2020	2198
DauT	18 Apr 2010	2193
Yo	10 Aug 2019	2189
MbL	26 May 2019	2182
Liereyy	11 Feb 2019	2180
JorDan_23	27 Dec 2013	2177
Nicov	mar 2020	2123
(czech) Lojza	fev 2016	2104
Ding	apr 2009	2103
Tim	jul 2016	2101
BacT	mar 2019	2097
F1re	oct 2016	2091
Ruso	oct 2011	2087
TheMax	jun 2019	2086
St4rk	dec 2018	2078
Grunt	jan 2004	2071
slam	jan 2016	2071
Villese	mar 2020	2069
IORI	fev 2008	2066
dogao	mar 2020	2066
Edie	jul 2017	2059
Halen	jul 2003	2057
Capoch	fev 2004	2056
vivi	jun 2019	2051
yinghua	jun 2018	2050
Koven	apr 2002	2048
Lyx	jun 2019	2047
BlinG	mar 2010	2046
RuKie	sep 2005	2044
WraTh	avr 2009	2044
GeeK	aug 2005	2040
Vinchester	apr 2016	2038
Chunyu	jun 2005	2034
Hackl	jul 2005	2033
fengxin	nov 2013	2033
MoZory	dec 2001	2032
PeEl	jul 2004	2032
LaaaaaN	dec 2019	2032
saymyname	nov 2013	2031
aluba	jan 2002	2030
Juan	jan 2016	2030
ACCM	jul 2016	2030

 

[Crawsack]

@mungo__ Just wanted to say this website is awesome- thanks for putting it together!

 

[mungo__]

@mungo__ Just wanted to say this website is awesome- thanks for putting it together!

Thank you, that's great to hear!

 

[HappyLeaves]

The available database dump seems a bit out of date: updated: 2020-01-02.

 

[mungo__]

The available database dump seems a bit out of date: updated: 2020-01-02.

Hey @HappyLeaves, here you go:
Data from today as SQL | as CSV | as JSON

I also updated the links on the website.

Have fun!

 

Hey @HappyLeaves, here you go:
Data from today as SQL | as CSV | as JSON

I also updated the links on the website.

Have fun!

Hey there,
" An API for fetching data directly is planned for the future. "
are there any updates on an API to easily interact with and pull in your data?

 

[mungo__]

Hey there,
" An API for fetching data directly is planned for the future. "
are there any updates on an API to easily interact with and pull in your data?

Hey @simonsanone I wrote you a small API to fetch some data. It includes
- player list
- single player info
- tournament list
- single tournament info
It's available here: https://aoe-elo.com/api/ (I also linked it from the main page)

Let me know if you would like to have more data included.

Best
mungo

 

Hey @simonsanone I wrote you a small API to fetch some data. It includes
- player list
- single player info
- tournament list
- single tournament info
It's available here: https://aoe-elo.com/api/ (I also linked it from the main page)

Let me know if you would like to have more data included.

Best
mungo

POG <3

Thank you :-O

 

Let me know if you would like to have more data included.

Best
mungo

Hey mungo,
I was talking to one of the Liquipedia admins if there would be anything missing to pull in the data (that will be cached) and he said to me, that there would be a license needed that the data can be used by Liquipedia. (cite: "anything that allows us to use it would work, we can add a backlink")

There is this website: https://choosealicense.com/ that helps you decide for a license. Would be cool if you state what you have chosen on your https://aoe-elo.com/about page.

Thanks for the effort and cheers mate :-)

 

[chicklet]

Just a note that the flurry of recent tournaments, particularly those with players outside of the top 16, has had great consequences for improving the accuracy of the ratings.

AOE-Elo provides the rating of AOE players only based on their tournament games. We also collect tournament dates, results, team stats and plenty of other AOE data. Browse a bit!

AOE-Elo provides the rating of AOE players only based on their tournament games. We also collect tournament dates, results, team stats and plenty of other AOE data. Browse a bit!

aoe-elo.com

I think the top of the players leaderboard is looking in fine shape now (with a couple of exceptions for returning players, which should get ironed out if they remain active). Not sure about position 30 down, as my knowledge of the relative ability amongst this level isn't great, but they're getting more tournament games which is good news.

Congratulations @mungo__ and the team. I hope you get some recognition - it'd be great to see your site used for seedings in future tournaments.

 

[JoshuaR]

Yeah, at this point what makes the most sense to me for future tournament seeding is an admin seeding manually... but the manual seeding is literally only to give the admin the ability to fix any discrepancies from the tournament elo page (e.g. someone who doesn't have enough recent data).

 

[Überrushung]

Just a note that the flurry of recent tournaments, particularly those with players outside of the top 16, has had great consequences for improving the accuracy of the ratings.

AOE-Elo provides the rating of AOE players only based on their tournament games. We also collect tournament dates, results, team stats and plenty of other AOE data. Browse a bit!

AOE-Elo provides the rating of AOE players only based on their tournament games. We also collect tournament dates, results, team stats and plenty of other AOE data. Browse a bit!

aoe-elo.com

I think the top of the players leaderboard is looking in fine shape now (with a couple of exceptions for returning players, which should get ironed out if they remain active). Not sure about position 30 down, as my knowledge of the relative ability amongst this level isn't great, but they're getting more tournament games which is good news.

Funny that you say this now that TheViper lost the no. 1 spot for the first time in years.