This rating doesnt influence ur skill at all. If u own, u will own. Just dont keep it on mind=)
I didn't say that it influences my skill, I just think the rating will be more accurate and informative if games with stacks and solo games don't get mixed up. Since Valve didn't bother introducing a Single-Queue mode yet, a separate dotabuff rating would be the best alternative.
Sure, as a good player you can own in single and stack games but the likelihood that you own in a game without any friends against a well organised stack is much lower, even if your team has on average the same ELO.
I think the mechanism of match games has already considered the influence of W/L rate in stack games. You are more likely to play against another team if you are playing in a team, or you may meet higher-ELO players as your opponents. So it is not that easy to win a stack game.
I do not know Valve's system, it seems more likely, yes, but there are also a lot of games in which stacks play against random players. Even if the average ELO of the latter is higher (which I don't believe) good picks and teamwork are much more important in Dota and at least my W/L ratio is much higher in stacks.
I agree. I have big problems solo queuing, because I often have people in my team who just throw the game for me by picking all carries, taking all my farm, abandoning or stuff like that. Then people on the forums tell me that my winrate is only 50% whereas that other guy who got 90% winrate they think he was better than me, but he's just playing with a stack of 5 all the time and his allies just pick Nyx so the enemies can not counterpick his OD...
Por favor, se conecte para postar comentários.
Hey,
it would be useful if the new ranking was split in two parts: games with stacks and games without, as the winrate for the former is likely higher. Since dotabuff already tracks how many games are played with certain players, it would be easy to introduce a certain level after which a game is considered to be a stack-game (e.g. you played with 3 out of the 4 others at least 10 times before). This would make the rating much more accurate without adding more data than the win/loss rate.
Best regards