[...]I started to get data about the misses and ineffective hits of me and my girlfriend while using the best cannon of the game, BL-10. It wasn't easy to get the data as the game has no combat log and the developers explicitly told that they never put in any. I had to make marks on a paper during battles, so probably made mistakes. This data is not at all accurate, but the result is so big that the signal-noise ratio is probably good enough: On battles that we won 13% of the hits were ineffective. On battles that we lost, 29% of the hits were ineffective.
Now my tinfoil hat explanation: the matchmaker decides who needs a win to not fall too far from the average and who needs a defeat to not elevate too far from it and assemble the teams accordingly. The "winner" team gets a bonus to its penetration chance, the "loser" team gets a penalty. While you can make a miracle and with awesome play you can win even these battles, or lose a "won" battle to griefers or 6+ AFK-ers, but the bonus/penalty on average works, making sure that everyone is just slightly away from average. After the victory is decided, the players do the rest. As I already found, on the winner side the kills are much more distributed, half of the winner team scores at least one kill. So while in every battle the good ones "pwn", the "good ones" change from battle to battle.[...]