EPTI copy and pasted between the site and the friend add list TT__TT Is there anything you have to do with gateways?
BGH auto balance -> http://bghmmr.com/ - Page 4
| Forum Index > BW General |
|
Jealous
10076 Posts
I copy and pasted between the site and the friend add list TT__TT Is there anything you have to do with gateways? | ||
|
D3AD-R3TARD
Poland271 Posts
On May 14 2020 01:13 Jealous wrote: I copy and pasted between the site and the friend add list TT__TT Is there anything you have to do with gateways? Battle tags from bghmmr.eu: D3ADR3TARD#2308 and BghLeagueBot#2231 should work. The region is of course Europe. You can also PM me and I'll send friend request to you. | ||
Liquid`Drone
Norway28503 Posts
On May 14 2020 01:13 Jealous wrote: I copy and pasted between the site and the friend add list TT__TT Is there anything you have to do with gateways? you prolly have to log on to europe in the starcraft launcher, not just on battle.net | ||
|
Jealous
10076 Posts
On May 14 2020 10:10 Liquid`Drone wrote: you prolly have to log on to europe in the starcraft launcher, not just on battle.net Just tried that, said it was already pending TT__TT I'll just PM them. PS. This system is so stupid. | ||
|
Jealous
10076 Posts
| ||
|
mrc
Poland72 Posts
First of all I'd like to thank D3AD-R3TARD for such great work with this thing. Really impressive! It's a great idea, great execution. It's sort of reviviying the EU bnet community. I wish it would run 24/7. Some things to consider: How to deal with laggers? How to deal with really bad players and griefers? How to deal with persistent leavers/dodgers (i.e. "x: Sorry but 8 tr sux/3 protoss sux/ally noob. x has left the game.") Maybe it could be done by a !ban vote option (for example requires 5 votes)? | ||
|
MeSaber
Sweden1228 Posts
Probably need to make a combination of win/loss ratio + apm, if you want somewhat even games. MMR alone doesnt say anything. win/loss alone is a bad indicator when people want a loss account. If it however is teamed with apm it might prove to be more useful to see if someone is faking it. However if he both uses low apm and loss account it wont help either. Edit: Other data that may want to be involved into showing someones skill is resource collection. Did he gather much resources? Same goes for this, if its a zerg player with superb muta control he doesnt need much resources to be useful/win, it would however show on APM, as muta micro cant go without high apm. Edit2: about loss account: You could make it so that the team who won even though a loss account left before win, he would still be eligible for the win instead of a loss (the team won and he was in the team, no matter if he died 5 mins in and game went for 50 mins, he still get the win). This would add false positives to a bad player if he actually died 5 mins in for real, however you could make it so if he stayed for more than 75% of the games duration he still gets win even if he lost. Whats important here is that good players cant disguise as bad players, then it will automatically be an uneven game as a good player can easily play like 3 if he wanted to. | ||
|
Jealous
10076 Posts
On May 19 2020 02:37 MeSaber wrote: A good player doesnt need high APM to be good. Its a bad indicator of skill, especially in a BGH game that you play for fun and dont really try to play good but more unorthodox. Probably need to make a combination of win/loss ratio + apm, if you want somewhat even games. MMR alone doesnt say anything. win/loss alone is a bad indicator when people want a loss account. If it however is teamed with apm it might prove to be more useful to see if someone is faking it. However if he both uses low apm and loss account it wont help either. Edit: Other data that may want to be involved into showing someones skill is resource collection. Did he gather much resources? Same goes for this, if its a zerg player with superb muta control he doesnt need much resources to be useful/win, it would however show on APM, as muta micro cant go without high apm. Edit2: about loss account: You could make it so that the team who won even though a loss account left before win, he would still be eligible for the win instead of a loss (the team won and he was in the team, no matter if he died 5 mins in and game went for 50 mins, he still get the win). This would add false positives to a bad player if he actually died 5 mins in for real, however you could make it so if he stayed for more than 75% of the games duration he still gets win even if he lost. Whats important here is that good players cant disguise as bad players, then it will automatically be an uneven game as a good player can easily play like 3 if he wanted to. Have there been loss accounts/smurfs already or...? | ||
|
D3AD-R3TARD
Poland271 Posts
On May 15 2020 10:48 Jealous wrote: Just want to make sure - I have to be logged into Europe server through the Blizzard Launcher to see when games are hosted/bot is online, yes? Yes, games are hosted on Europe server only. They appear as normal Top vs Bottom games. On May 18 2020 22:34 mrc wrote: Hello, First of all I'd like to thank D3AD-R3TARD for such great work with this thing. Really impressive! It's a great idea, great execution. It's sort of reviviying the EU bnet community. I wish it would run 24/7. Some things to consider: How to deal with laggers? How to deal with really bad players and griefers? How to deal with persistent leavers/dodgers (i.e. "x: Sorry but 8 tr sux/3 protoss sux/ally noob. x has left the game.") Maybe it could be done by a !ban vote option (for example requires 5 votes)? Thanks. As for the problems you mentioned: Difficulty with banning lagers is that is not clear who lags. I am considering banning people who are red before game. But that will not fix the problem. Often lagers are displayed as yellow or green and only in game it is clear that they lag. However, I you leave early in game it doesn't count so and for now it's the only way to avoid games with lag. I don't have problem with bad players. Usually their MMR reflects their skill and the team who gets them also gets the best player so usually it's even. As for toxic players I also thought about voting/banning players but I had another idea. I was thinking about command that would allow the best player in each team to ban one other player from the team. I think that would be fast and simple solution and would motivate players not to leave early or sit in base for the whole game. What do you guys think? On May 19 2020 02:37 MeSaber wrote: A good player doesnt need high APM to be good. Its a bad indicator of skill, especially in a BGH game that you play for fun and dont really try to play good but more unorthodox. Probably need to make a combination of win/loss ratio + apm, if you want somewhat even games. MMR alone doesnt say anything. win/loss alone is a bad indicator when people want a loss account. If it however is teamed with apm it might prove to be more useful to see if someone is faking it. However if he both uses low apm and loss account it wont help either. Edit: Other data that may want to be involved into showing someones skill is resource collection. Did he gather much resources? Same goes for this, if its a zerg player with superb muta control he doesnt need much resources to be useful/win, it would however show on APM, as muta micro cant go without high apm. Edit2: about loss account: You could make it so that the team who won even though a loss account left before win, he would still be eligible for the win instead of a loss (the team won and he was in the team, no matter if he died 5 mins in and game went for 50 mins, he still get the win). This would add false positives to a bad player if he actually died 5 mins in for real, however you could make it so if he stayed for more than 75% of the games duration he still gets win even if he lost. Whats important here is that good players cant disguise as bad players, then it will automatically be an uneven game as a good player can easily play like 3 if he wanted to. The system right now uses win ratio from custom games + APM and later on results from balanced games. All this info combined gives MMR score. As for resources, replays don't contain this info. There is a score at the end of the game but for now I don't use this. Reading things like muta micro from replay file is pretty challenging task but I do plan to use game info from replays to estimate player performance in the game. As for leaving before the game ends, it doesn't change anything. Wins and losses are for team only. So if your team won you get the points no matter when you left. But yes people start new accounts and appear as noobs to the system. I see this a lot and also don't like this. I thought about games for league members only (at least 4 games) but there are so | ||
|
Jealous
10076 Posts
On May 19 2020 05:21 D3AD-R3TARD wrote: Yes, games are hosted on Europe server only. They appear as normal Top vs Bottom games. Just going on Europe server in-game is not enough to see friends - you have to log in through the Blizzard app to Europe server, and then in-game you go on Europe server. That is how i was able to see your friend add, and now get notified about games. | ||
|
mrc
Poland72 Posts
As for toxic players I also thought about voting/banning players but I had another idea. I was thinking about command that would allow the best player in each team to ban one other player from the team. I think that would be fast and simple solution and would motivate players not to leave early or sit in base for the whole game. What do you guys think? I could be leaving game before it starts when I dont like my team composition, or leaving on game start. if I were the top ranked player in the game I could be simply banning whoever I don't like playing against or with. this will make my mmr likely higher and my voting power even stronger. therefore, I think democratic voting would be more fair. another option that comes to mind would be to punish leavers with a rejoin cooldown. btw. what happens when I leave the game early and my teammates continue playing and lose? | ||
|
kurrrak
Poland98 Posts
| ||
|
Jealous
10076 Posts
 I joined for about 30 minutes before giving up. | ||
|
D3AD-R3TARD
Poland271 Posts
On May 20 2020 23:31 mrc wrote: On May 19 2020 02:37 MeSaber wrote: I could be leaving game before it starts when I dont like my team composition, or leaving on game start. if I were the top ranked player in the game I could be simply banning whoever I don't like playing against or with. this will make my mmr likely higher and my voting power even stronger. therefore, I think democratic voting would be more fair. another option that comes to mind would be to punish leavers with a rejoin cooldown. btw. what happens when I leave the game early and my teammates continue playing and lose? Yes, that option could make very good players ban noobs too much. But I thought about only banning one person from your team. So one ban per game. On May 24 2020 04:14 kurrrak wrote: I've changed my mind about APM, don't count it into ranking. Often times I'm taken as a team leader just because I have high APM, even despite win/loose ration being 1/2. It makes games really frustrating since I can't be bothered to say to rest of noobs from my team what to do because I'm no strategist. So they all sit in base whole game, waiting for orders. 90% of games looks like that. If you played more than 4 games and I suppose you did, your MMR is based more on results than APM. APM is only used in first 4 games to find good starting point for MMR and for this it works more often than it doesn't. As for 50-50 win rate: all players should have around 50-50 win rate but different MMR. If you have MMR of 3000 and win rate 50/50 that means you win half of games where you play with noobs against average players. So it's different from 50/50 win rate of someone with MMR of 400 who wins half of the games where he plays with pros against average players. I currently don't have as much time as I would like to work on this MMR system but I plan to look for better formula and recalculate all scores. On May 24 2020 07:06 Jealous wrote: Btw the other day I finally managed to join some games but the bot was always in the player 1 slot, never moved to obs, and no games started I joined for about 30 minutes before giving up.Yeah, bot hangs sometimes. If it doesn't go to observer slot within 10 seconds of starting new game, it is frozen and just ignore those games. | ||
|
mrc
Poland72 Posts
 | ||
|
Timebon3s
Norway585 Posts
Almost always close even games | ||
|
D3AD-R3TARD
Poland271 Posts
On June 07 2020 01:36 mrc wrote: how about auto-ban for people with1 mmr? As I said earlier noobs are really not the problem. If the system knows someone is a noob there is still a chance to balance the teams and have a game. I would rather ban unranked players (less than 4 balanced matches). Most of the unbalanced matches I've played are not balanced because of some new player who is either much better or much worse than expected. And you can't really asses MMR out of stats alone. So this problem will happen over and over. I'm thinking of a command that will allow any ranked player to ban all unranked players. I also thought about banning people with little stats. Like people with 10 matches. Not only they are unranked so there is very little knowledge of their skill in BGH and APM but also there is very little information in the stats. With those players it is very difficult to guess their MMR. | ||
|
Jealous
10076 Posts
On June 07 2020 03:46 D3AD-R3TARD wrote: As I said earlier noobs are really not the problem. If the system knows someone is a noob there is still a chance to balance the teams and have a game. I would rather ban unranked players (less than 4 balanced matches). Most of the unbalanced matches I've played are not balanced because of some new player who is either much better or much worse than expected. And you can't really asses MMR out of stats alone. So this problem will happen over and over. I'm thinking of a command that will allow any ranked player to ban all unranked players. I also thought about banning people with little stats. Like people with 10 matches. Not only they are unranked so there is very little knowledge of their skill in BGH and APM but also there is very little information in the stats. With those players it is very difficult to guess their MMR. For me this would be a problem, because I am someone who has basically not played on Europe at all on my current accounts (have been making new accounts and deleting old ones due to BSL and other events/tournaments), so I have like 10 games on my account. Also, really struggling to get into a BGH bot game, partially because of time zone difference as well as high demand (except when the bot is not working). So, it feels like if players can ban based on these factors, it's like a window of people joining this "ladder"/community is "closing" if that makes sense. | ||
|
Crazy Citizen
4 Posts
Firstly, thank you for making this bot Is there a way to report team-killers? I had a match where a player killed his 2 allies and then, either, left the game or allied the enemy. The team killer and the 3 enemies all got +MMR https://imgur.com/a/TbBF8a1 I have the replay file too, if needs-be. "Finlandno1Zerg" was the offending player. Essentially, a player can wait a few minutes, avoid the enemy and 4v2, then the 4 players all gain MMR. Also if anyone can explain the psychology of trolls like this, I would be very interested ^^ | ||
|
Jealous
10076 Posts
On June 09 2020 01:24 Crazy Citizen wrote: Also if anyone can explain the psychology of trolls like this, I would be very interested ^^ Why play fair and risk losing if you can just backstab and win? If your team is noobs then why waste time trying to help them win if you can just kill them? More fun than carrying noobs right? Something like that. | ||
| ||
