23 replies to this topic

[Elessar]

Kaeseblock, on 26 February 2013 - 07:38 AM, said:

...

Btw, @ PGI:
I would like to see at least my own ELO rating.

This would be nice ...
while I forn myself have begun to take statistics of my battles (including Score) and therefore am able to give an educated guess anout my average score for each weight class, it would be nice if such thing would already be included in the stats page of my profile

[buttmonkey]

im not sure what to think of elo, its a good idea but after playing my fav mech all maxed out on upgrades and xp, playing a new mech makes the games much more difficult. maybe its just coincidence

[Rumrunner2]

Not so skilled solo players benefit from ELO because they dont have to face the topdogs anymore.

But skilled players and teamplayers suffering very often bad imbalance.

[ohtochooseaname]

Maybe it's just me, but it seems like the MM is taking one person, matching them up with someone else, then putting them on the same team instead of on opposing teams. This would generally result in even numbers of classes on one team (so, 2, 4, 6, or 8 assaults). There would obviously be exceptions, but I wonder what the aggregate data shows. I ran a Monte Carlo analysis on the likelihood that there will be an even number of a certain class of`Mechs assuming a random uniform distribution and that zero Mechs of a certain class doesn't count as an even number. It turns out, there is only a 44.5% chance that there will be an even number of 'Mechs of a certain class. This means odd numbers are ~25% more likely than even. Random distribution is a pretty liberal assumption, but I think it indicates that there may be a bug in the MM algorithms. I'll have to keep a class tally in the future to see.