Just for a change a thread not regarding a ban, unfair unban or whatever
I recently played some Btank games after quite a while and noticed that most of the players have <1500 elo. Even the top200 consists of 75% players with <1500 elo. One problem might be the very small player pool nowadays. I remember (years ago) you needed at least 1600 elo to be in the bottem of top10. Top3 had up to 1700+ elo.
In theory, the following calculation should be "0" for a good elo system.
sum((Elo of every single player)-1500) = 0
That means, if someone is above 1500 (for example 1550) you need the counterpart with 1450 elo because elo can only fluctate.
Problem is now, if someone leaves a game and his remaining team still wins 4v5 he doesnt gain elo while all five players of the opponent team lose elo. This drains elo out of the system and over time the average of all players drops below 1500 when it should always be 1500.
I also noticed that a team doesnt gain the same elo in summery that the other team loses, but I dont have any data to determine if thats a zero-sum game over the time.
Finally my suggestion:
Rework the formula in that way that it is settled before the game starts how much elo the whole team can win/lose in summary.
For example:
Team1 has 50.5% win chance
Team2 has 49.5% win chance
average elo gain per player in a 50:50 game is 10.
Team1: 10* (49.5/50.5) * 5players = 49
Team2: 10* (50.5/49.5) * 5players = 51
If team 1 wins they can gain as a team 49 elo while team 2 loses 49.
If team 2 wins they can gain as a team 51 elo while team 1 loses 51.
If team 1 managed to win the game 5v5 the players share the 49 elo
If team 1 managed to win the game 4v5 the elo of 49 will be distributed between 4 players now.
After the game ended and the elo win for the team is fix the actuall distribution among the players can be caltulated with parts of the old formula. If you did well on the game you earn a bigger part of the cake.
Why am I suggestion this?
Simply because the ranking, with all due respect, sucks for the current situation. A lot of good players have <1500 elo. New players with fresh 1500 elo unballance games. You can reach top10 within 1 day. It might be frustrating if you play decent while having <1500 elo. You are still over the average elo (of maybe 1450? idk), but it feels low. In addition it might be better to halve the Elo you can win/lose per game to reduce the high volatility due to winning or losing streaks.
I recently played some Btank games after quite a while and noticed that most of the players have <1500 elo. Even the top200 consists of 75% players with <1500 elo. One problem might be the very small player pool nowadays. I remember (years ago) you needed at least 1600 elo to be in the bottem of top10. Top3 had up to 1700+ elo.
In theory, the following calculation should be "0" for a good elo system.
sum((Elo of every single player)-1500) = 0
That means, if someone is above 1500 (for example 1550) you need the counterpart with 1450 elo because elo can only fluctate.
Problem is now, if someone leaves a game and his remaining team still wins 4v5 he doesnt gain elo while all five players of the opponent team lose elo. This drains elo out of the system and over time the average of all players drops below 1500 when it should always be 1500.
I also noticed that a team doesnt gain the same elo in summery that the other team loses, but I dont have any data to determine if thats a zero-sum game over the time.
Finally my suggestion:
Rework the formula in that way that it is settled before the game starts how much elo the whole team can win/lose in summary.
For example:
Team1 has 50.5% win chance
Team2 has 49.5% win chance
average elo gain per player in a 50:50 game is 10.
Team1: 10* (49.5/50.5) * 5players = 49
Team2: 10* (50.5/49.5) * 5players = 51
If team 1 wins they can gain as a team 49 elo while team 2 loses 49.
If team 2 wins they can gain as a team 51 elo while team 1 loses 51.
If team 1 managed to win the game 5v5 the players share the 49 elo
If team 1 managed to win the game 4v5 the elo of 49 will be distributed between 4 players now.
After the game ended and the elo win for the team is fix the actuall distribution among the players can be caltulated with parts of the old formula. If you did well on the game you earn a bigger part of the cake.
Why am I suggestion this?
Simply because the ranking, with all due respect, sucks for the current situation. A lot of good players have <1500 elo. New players with fresh 1500 elo unballance games. You can reach top10 within 1 day. It might be frustrating if you play decent while having <1500 elo. You are still over the average elo (of maybe 1450? idk), but it feels low. In addition it might be better to halve the Elo you can win/lose per game to reduce the high volatility due to winning or losing streaks.