But you know, %66 winrate for 2000 games is actually insane. But it still needs a rework.
I am not very familiar with the hero score system. I only used my previous 3 games for estimate, so my idea is still very premature.
To understand this rating system better, we need to know the algorithm.
Овај коментар је измењен
Mr Grim.
What you forget here is MMR. And how mmr changes over time.
To clearify.
Noob plays in 1k bracket gets 100% winrate 100 games. has X kda
Pro plays in 5k bracket gets 100% winrate 100 games. has X kda.
Pro plays in 5k bracket gets 100% winrate 200 games (but with higher KDA) in a gradually higher getting bracket.
Овај коментар је измењен
Пријавите се да бисте поставили коментар.
Currently, number of games is valued too highly for Hero Score
Example:
Lost 1 game, dropped 5 points
Won 2 games, gained 3+5 points
Basically, play a hero 2000 times with a minimum win ratio of 66% and decent KDA, and you'll likely be on the top 100.
On a side note:
According to statistics, 2000 sample size is enough for an accurate estimate.
Players with 2000 games and 4000 games should have the same hero score if their KDA and win rate are identical.
[IMG]http://i.imgur.com/dSzVhLk.jpg[/IMG]
Does this look fair?
# of games give too many points to hero score?
There are flaws with every system; let's try to build the best rating system together.