ESL - Elite Stats & Leagues web extension [beta]
- Ace of Swords
- Posts: 1493
- Joined: Thu 14 Mar, 2013 7:49 am
- Location: Terra
Re: ESL - Elite Stats & Leagues web extension [beta]
Seems like the match I've played a few days ago disappeared from the log, we didn't say Ranked but it wasn't a public match, it was an automatch one.
Map was outer reaches, name of the guy was Mr Cpt Diomedes.
Map was outer reaches, name of the guy was Mr Cpt Diomedes.
Re: ESL - Elite Stats & Leagues web extension [beta]
Yeah, I disappeared too
Re: ESL - Elite Stats & Leagues web extension [beta]
Ace of Swords wrote:Seems like the match I've played a few days ago disappeared from the log, we didn't say Ranked but it wasn't a public match, it was an automatch one.
Map was outer reaches, name of the guy was Mr Cpt Diomedes.
https://dawnofwar.info/esl/index.php?pa ... r=10008941
Re: ESL - Elite Stats & Leagues web extension [beta]
Forestradio wrote:on a bug-related note, it seems like all GK games are being counted as FC games, since i just played johngrammaticus and he was gk and in the log it shows it as fc...
Hakon says this is now fixed.
Ace of Swords wrote:Seems like the match I've played a few days ago disappeared from the log, we didn't say Ranked but it wasn't a public match, it was an automatch one. Map was outer reaches, name of the guy was Mr Cpt Diomedes.
It's still there: https://dawnofwar.info/esl/index.php?pa ... r=10008941
Re: ESL - Elite Stats & Leagues web extension [beta]
Not a bug with the ladder, but I'm wondering if a "rematch" phrase can be scripted in?
Just last night I had a huge lag spike (from an observer) and ended up losing my hero in the first minute of a game against Floid. It was a clear situation where a restart to the game without penalty would have been a good fit.
My thinking is, if it is possible to script a restart, to make it so that it has to happen in the first 2 minutes of the game and it has to be said in a single word by both parties, "rematch" or "restart" or w/e really.
Just last night I had a huge lag spike (from an observer) and ended up losing my hero in the first minute of a game against Floid. It was a clear situation where a restart to the game without penalty would have been a good fit.
My thinking is, if it is possible to script a restart, to make it so that it has to happen in the first 2 minutes of the game and it has to be said in a single word by both parties, "rematch" or "restart" or w/e really.
- Ace of Swords
- Posts: 1493
- Joined: Thu 14 Mar, 2013 7:49 am
- Location: Terra
Re: ESL - Elite Stats & Leagues web extension [beta]
Lulgrim wrote:Forestradio wrote:on a bug-related note, it seems like all GK games are being counted as FC games, since i just played johngrammaticus and he was gk and in the log it shows it as fc...
Hakon says this is now fixed.Ace of Swords wrote:Seems like the match I've played a few days ago disappeared from the log, we didn't say Ranked but it wasn't a public match, it was an automatch one. Map was outer reaches, name of the guy was Mr Cpt Diomedes.
It's still there: https://dawnofwar.info/esl/index.php?pa ... r=10008941
Alright thanks, I couldn't find it for some reasons, great job btw, love the ladder.
Re: ESL - Elite Stats & Leagues web extension [beta]
Tex wrote:Not a bug with the ladder, but I'm wondering if a "rematch" phrase can be scripted in?
We could, or we could just remove the game manually. It is technically simple to void it with a chat agreement though.
Re: ESL - Elite Stats & Leagues web extension [beta]
The rematch function sounds like it would be a great addition and would definitely reinforce legitimacy in the rankings.
Re: ESL - Elite Stats & Leagues web extension [beta]
In other news, it looks like the feature we had to tell apart automatch & custom games is in fact bugged since the backend rework that fucked up various things, so in the end we can't really separate them after all. This leaves two options:
- Rate all games. I don't personally like it, as I think it will skew the skill estimates, and it will also make people turn off the reporter, while I would like comprehensive stats for all game types across Elite.
- Use the chat agreement (or some other similiar technique, although more advanced options get a lot more complicated technically). This, of course, causes some inconvenience to the players.
Re: ESL - Elite Stats & Leagues web extension [beta]
I'm okay with all games being reported tbh...
The more games that are played, the more accurate the rating gets. If somebody idles with a super good rating at the top of the ladder, but only has 10 games played after everyone else has 100 or so, it should be pretty obvious that their rating is no longer credible. I'm not sure if some kind of decay form has been added, but for now I think we are all just super happy about having a working ladder again.
The more games that are played, the more accurate the rating gets. If somebody idles with a super good rating at the top of the ladder, but only has 10 games played after everyone else has 100 or so, it should be pretty obvious that their rating is no longer credible. I'm not sure if some kind of decay form has been added, but for now I think we are all just super happy about having a working ladder again.
- Adeptus Noobus
- Posts: 991
- Joined: Sat 15 Feb, 2014 12:47 pm
- Contact:
Re: ESL - Elite Stats & Leagues web extension [beta]
I would also not mind all games being ranked.
@Tex: The Glicko-1 system should account for inactivity by calculating the rating deviation (RD) value for each skill rating (SR) iirc. The first value shown is simply SR - RD. The calculated value for the SR is then displayed as SR ± RD.
@Tex: The Glicko-1 system should account for inactivity by calculating the rating deviation (RD) value for each skill rating (SR) iirc. The first value shown is simply SR - RD. The calculated value for the SR is then displayed as SR ± RD.
- Crewfinity
- Posts: 712
- Joined: Tue 03 Dec, 2013 2:06 am
Re: ESL - Elite Stats & Leagues web extension [beta]
I'm also in favor of all games being ranked. People have to choose to use the app anyway so having to say "ranked" in each game is kind of annoying, especially if you have to do it for automatch games as well
- Ace of Swords
- Posts: 1493
- Joined: Thu 14 Mar, 2013 7:49 am
- Location: Terra
Re: ESL - Elite Stats & Leagues web extension [beta]
Depends, probably the easier way is just to go with the "say Ranked" way, considering this if guy 1 has the program open and wants to to be ranked and guy 2 doesn't, assuming we go with fix N°1 (all games ranked) what happens?
- Forestradio
- Posts: 1157
- Joined: Sun 13 Oct, 2013 5:09 pm
Re: ESL - Elite Stats & Leagues web extension [beta]
would rather have all games considered ranked so that the esl thingy can get tested as much as possible
when 2.5 hits (Soon™) imo the ladder should be reset and not all games should be ranked except automatch and if agreed upon by the two players
will this be automatic with 2.5, without having to run the separate application?
when 2.5 hits (Soon™) imo the ladder should be reset and not all games should be ranked except automatch and if agreed upon by the two players
will this be automatic with 2.5, without having to run the separate application?
Re: ESL - Elite Stats & Leagues web extension [beta]
Would also support just making it easy and going all games ranked.
As for the problem of consent, maybe put in a chat password for discounting the game? Like, if you don't want it included all that needs to happen is for 1 person to say "unranked"?
EDIT: Also, could we possibly consider adding in a column for game version in the match log? For the sake of people trying to download replays and such.
As for the problem of consent, maybe put in a chat password for discounting the game? Like, if you don't want it included all that needs to happen is for 1 person to say "unranked"?
EDIT: Also, could we possibly consider adding in a column for game version in the match log? For the sake of people trying to download replays and such.
Re: ESL - Elite Stats & Leagues web extension [beta]
Tex wrote:I'm okay with all games being reported tbh... The more games that are played, the more accurate the rating gets.
The thing is, in this case all sorts of lul games, practice games with a new faction, trying out an insane build etc. are assumed to be serious and included in the skill estimation if any one participant (even observer) has the app running. The calculation should apply to competitive games by design.
I guess there's a third option, though – rating games where both sides have the app running. Although then you could RQ the app when losing to nullify the result… So not optimal either.
Tex wrote:If somebody idles with a super good rating at the top of the ladder, but only has 10 games played after everyone else has 100 or so, it should be pretty obvious that their rating is no longer credible. I'm not sure if some kind of decay form has been added, but for now I think we are all just super happy about having a working ladder again.
Players are ranked according to rating minus RD (67% statistical chance the player is worth that number), and RD increases slightly each day you're inactive. So yes, the "credible rating" decays.
There is also a limit of RD < 250 for appearing on the ladder at all; this could be lowered further if people still get too high with too few games. Or we could go with rating minus 2 × RD (95% chance of the player being worth that), which might be even too steep, but idk.
Re: ESL - Elite Stats & Leagues web extension [beta]
Forestradio wrote:would rather have all games considered ranked so that the esl thingy can get tested as much as possible
Yeh, that's what's happening now. A bunch of them are probably not intended as rated.
Forestradio wrote:when 2.5 hits (Soon™) imo the ladder should be reset and not all games should be ranked except automatch and if agreed upon by the two players
Unfortunately, like I said we can't determine automatch status because a Relic thingy is bugged.
Forestradio wrote:will this be automatic with 2.5, without having to run the separate application?
I think the best we can do is have the mod shortcut launch the app automatically, can't really inject it into DoW2.
Re: ESL - Elite Stats & Leagues web extension [beta]
Atlas wrote:Also, could we possibly consider adding in a column for game version in the match log? For the sake of people trying to download replays and such.
There is a mod version field on the match details page, it's just not getting logged for whatever reason atm. No space on the list page though, I'm already cutting names to a max length and hiding columns on narrow screens.
Re: ESL - Elite Stats & Leagues web extension [beta]
Parse error: syntax error, unexpected ';', expecting ')'
ladder seems slightly broken atm
ladder seems slightly broken atm
Re: ESL - Elite Stats & Leagues web extension [beta]
Tex wrote:ladder seems slightly broken atm
That's usually because I'm editing something (CBA having a test site). Always temporary.
In other news again, I added a new graph on the stats page that shows the win ratio (%) of each faction for different game length categories (5–10 minutes, 10–15 minutes etc.)
Re: ESL - Elite Stats & Leagues web extension [beta]
Lulgrim wrote:Players are ranked according to rating minus RD (67% statistical chance the player is worth that number), and RD increases slightly each day you're inactive. So yes, the "credible rating" decays.
There is also a limit of RD < 250 for appearing on the ladder at all; this could be lowered further if people still get too high with too few games. Or we could go with rating minus 2 × RD (95% chance of the player being worth that), which might be even too steep, but idk.
The RD seems like a really powerful stat. Leaning into it a bit more to see a "for reals" ladder makes a lotta sense.
Basically, the ladder looks awesome, but it feels cheapened by having people with less then 5 games ranked above people with over 50 Not sure the right solution, but this is AMAZING already. I think the ability to toggle between the ladder showing "all" (RD < 250) and "confident" (RD < 80) would be nice. Or maybe just a hard cap of not showing till some number of games has been played? My gut is that RD capping would be best, sicne it means inactive players will slowly disappear off the list.
As for opt in stuff, I'm in favor of logging all games where 1 or more people have the app up (though it'd be cool if obs couldn't count? Can the player who submitted be compared to the active players in the match to filter this out?) This is for a few reasons:
-Easy Adoption - Humans are lazy. The more games in this system, and the more people exposed to it, the more people will be interested. Make it easy.
-Gentlemans Rules - If a player is intending on "fooling around" and doesn't want the match to count, they can ask their opponent to leave the app off. It'll be real easy to notice if they didn't.
-This is no exact science - There are many different ways to "exploit" this system if you want, or unintentionally mess up data. So I'd error on the side of more data in the system, then being too cautious that it's the right data. If this was some big retail game I'd think the opposite, but this is a labour of love on a fan-made mod. Don't worry too much, just get some data!
Also, all the stats and graphs added are just delicious. https://dawnofwar.info/esl/index.php?page=stats
40% of matches on Calderis, heh. 75% Nid win ratio, no surprise Can those tooltips list both % and absolute counts? I'd be curious what number of the IG losses I've contributed to get them to 33% :/
- Crewfinity
- Posts: 712
- Joined: Tue 03 Dec, 2013 2:06 am
Re: ESL - Elite Stats & Leagues web extension [beta]
hahaha wow that faction popularity is so ork skewed. pretty sure that's mostly due to me and Tranca
this ladder is fucking awesome. loooooooving it
this ladder is fucking awesome. loooooooving it
Re: ESL - Elite Stats & Leagues web extension [beta]
Crewfinity wrote:hahaha wow that faction popularity is so ork skewed. pretty sure that's mostly due to me and Tranca
this ladder is fucking awesome. loooooooving it
lmao true! We probably have like 70% of the orc games on the ladder!
But, you and Tex are the only ones keeping the winrate above 50%! LOL
- Adeptus Noobus
- Posts: 991
- Joined: Sat 15 Feb, 2014 12:47 pm
- Contact:
Re: ESL - Elite Stats & Leagues web extension [beta]
Would it be possible to extract the faction win ratio per skill rating? It would give us a picture about which factions/heroes are played at a certain level of skill. It would also give us some insight wether certain factions provide an inherent advantage in 1v1s, regardless of skill. My hopes are that this could shed some light on the state of balance because statistically one faction should not significantly deviate from the others win/loss ratio under the hypothesis of a good state of balance. Obviously this graph will grant more and more insight as the number of games recorded increases. I hope I was able to make some sense
Re: ESL - Elite Stats & Leagues web extension [beta]
Very cool thought noobus. I just finished staring at the data for a few minutes and I noticed 3 things that may or may not be correlated by others...
1) It appears that the win rate per faction is most balanced at the 15-20 minutes mark. What I see here is that on average, a game played between players of equal skill level will take between 15-20 minutes. This data should really help people if they plan on running a "game clock" for timings, like some (all pros) SC2 players do.
2) It appears that tyranids are the most imba in T1... *cough* spore mines *cough*... Surviving T1 drastically reduces the tyranid win rate.
3) I'm not sure if anyone else can see this, but I feel like the stats are representing a severe lack of player skill in the IG department. I do not feel like IG are UP. I'm well aware of their good and bad matchups... maybe I just need to wait for more before calling this one...
1) It appears that the win rate per faction is most balanced at the 15-20 minutes mark. What I see here is that on average, a game played between players of equal skill level will take between 15-20 minutes. This data should really help people if they plan on running a "game clock" for timings, like some (all pros) SC2 players do.
2) It appears that tyranids are the most imba in T1... *cough* spore mines *cough*... Surviving T1 drastically reduces the tyranid win rate.
3) I'm not sure if anyone else can see this, but I feel like the stats are representing a severe lack of player skill in the IG department. I do not feel like IG are UP. I'm well aware of their good and bad matchups... maybe I just need to wait for more before calling this one...
Re: ESL - Elite Stats & Leagues web extension [beta]
It doesn't help that you mercilessly crushed my Inq with completely balanced 2x PM's all night Tex
To be honest, that's why I was asking to see absolute numbers in addition. Being able to see that only 10 matches with IG were played helps understand that the sample space is small. Nids also have such a small showing (7%) that it's hard to overly read into the other numbers. Any way to know what the total number of matches played is?
I think the things Tex and Noobus are talking about are nice, but I honestly would prefer they weren't added yet. The sample spaces we have are so small, that doing sub-set of sub-set comparisons are going to spit out some really goofy results. I'd urge focus on other features first.
The faction win by match length is weird to read. Nids winning 100% at sub5 mins means what exactly? Is it that every match with nids in it that ends in 5mins is a nid win?
Is GK working? Maybe I should play some GK tonight to get the numbers going. 1.3% played for the BEST faction in the game? Criminal!
To be honest, that's why I was asking to see absolute numbers in addition. Being able to see that only 10 matches with IG were played helps understand that the sample space is small. Nids also have such a small showing (7%) that it's hard to overly read into the other numbers. Any way to know what the total number of matches played is?
I think the things Tex and Noobus are talking about are nice, but I honestly would prefer they weren't added yet. The sample spaces we have are so small, that doing sub-set of sub-set comparisons are going to spit out some really goofy results. I'd urge focus on other features first.
The faction win by match length is weird to read. Nids winning 100% at sub5 mins means what exactly? Is it that every match with nids in it that ends in 5mins is a nid win?
Is GK working? Maybe I should play some GK tonight to get the numbers going. 1.3% played for the BEST faction in the game? Criminal!
- Crewfinity
- Posts: 712
- Joined: Tue 03 Dec, 2013 2:06 am
Re: ESL - Elite Stats & Leagues web extension [beta]
Cyris wrote:The faction win by match length is weird to read. Nids winning 100% at sub5 mins means what exactly? Is it that every match with nids in it that ends in 5mins is a nid win?
Is GK working? Maybe I should play some GK tonight to get the numbers going. 1.3% played for the BEST faction in the game? Criminal!
Meaning that some poor sods got their sluggas/shees/tics/everything wiped by spores in first engagement and then gave up
Can't really think of any other faction that has access to as much retreat killing capability for so cheap in first engagements.
It probably doesn't help that you, me, and forestradio have moved away from GK to orks/ig/orks haha
Re: ESL - Elite Stats & Leagues web extension [beta]
Cyris wrote:Can those tooltips list both % and absolute counts? I'd be curious what number of the IG losses I've contributed to get them to 33% :/
Yeah I'll try to fit them in.
Adeptus Noobus wrote:Would it be possible to extract the faction win ratio per skill rating?
You're talking about win ratios in skill rating brackets, e.g. who wins 1600–1800 games etc. As a byproduct of saving the onset rating/RD for display it is in fact easy to run that. However, also pointless at the moment as we don't have enough data – there's only a handful of balanced matchups. The results would be random. Could add it later though.
Cyris wrote:The faction win by match length is weird to read. Nids winning 100% at sub5 mins means what exactly? Is it that every match with nids in it that ends in 5mins is a nid win?
Well that's the definition of Nid win ratio in <5 minute games.
Re: ESL - Elite Stats & Leagues web extension [beta]
Maths: looks like 329 matches were played (source: I counted). Let's call it 330.
So, 1.3% GK means a grand total of 4 matches. Noooot exactly a good data set
IG at 12% means 40 matches. Not a large number, and very easily skewed by a bad streak or two. Like mine.
Then I think I figured it out
So, 1.3% GK means a grand total of 4 matches. Noooot exactly a good data set
IG at 12% means 40 matches. Not a large number, and very easily skewed by a bad streak or two. Like mine.
Lulgrim wrote:Cyris wrote:The faction win by match length is weird to read. Nids winning 100% at sub5 mins means what exactly? Is it that every match with nids in it that ends in 5mins is a nid win?
Well that's the definition of Nid win ratio in <5 minute games.
Then I think I figured it out
Re: ESL - Elite Stats & Leagues web extension [beta]
We need ALOT of more data before trying to base any balance evaluation based on ESL stats. But it's a start nontheless.
Return to “Community General Discussion”
Who is online
Users browsing this forum: No registered users and 23 guests