Guild Wars 2

It’s not easy to isolate the impact of the matchmaking algorithm on the outcome of games but I designed a simple experiment to provide a measure for the bias introduced in favor or against an individual player.

I played my best in 100 ranked games (just conquest) and entered the results in a spreadsheet. The plain average win ratio was 0.21 (21 wins out of a hundred). As there are lots of matches with largely unbalanced teams, it seemed appropriate to give more weight to matches with close outcomes. To do this, I simply divided the score of the losing team by the score of the winning team and squared the result. For example a close win of 500 to 450 would give a weight of 0.81 while a crushing defeat of 500 to 50 would give a weight of 0.01, a logical choice as individual prowess is more likely to make a difference in close games. With these weights factored in, the average ratio became 0.251.

With an unbiased algorithm, I would expect no big difference between the weighted and unweighted ratios in the long run, but over a hundred games, 1 in 5 actual wins compared to 1 in 4 potential wins seems like a very big bias.

https://forum-en.gw2archive.eu/forum/game/pvp/END-ALL-B-ALL-answer-4-MMR-you-re-looking-4/first#post6089979

I have practiced watched videos and improved done all you have suggested since the season started but unfortunately cant carry whole teams and with MMR tanked I cannot show my improvement – and oh yes – it is a team game no matter how good an individual is!

I don’t agree with discounting the blow outs. If all other players are really good, you could be such a liability that they get blown out. Unless the other team is awful, having an afk/non-contributed causes a blow out loss.

Also, I’ve been in plenty of matches where there were two opinions on how to play the map, and the team would have won if they would have committed to one of them. This is probably more common see the lower levels, but it essentially gives the other team a free win.

Another option for the results of the OP is lack of skill. I would suggest the following further calculation:
Go through your data until you find a win streak (5 wins should be enough). Redo the calculation after the win streak. Your results should be not statistically different from 0.5.