Honestly, until you can figure out a way to balance for teamwork and cooperation, there's really no way for ELO or any other matchmaking scheme to properly balance teams and create a fun and competitive environment on a consistent basis.
The result of the current attempts based almost exclusively on wins/losses, swings wildly between extreme balance issues (a team with 5 Assault-class mechs vs a team with 2, or a team with 3 Raven 3-Ls, vs a team with three non-ECM Lights, and sometimes no ECM mechs at all), and a significant percentage of the wins end up as shutouts/near-shutouts (7-1, 8-0, etc).
Really, the
first thing that could aid this would be to incorporate capturing and analyzing the scale of the wins & losses and not just whether they were wins or losses.
- Instead of a binary Win/Loss, actually capture how many kills & deaths each team had, assign all possible outcomes a point scale from least (0-8) to greatest (8-0), and factor that into player ELO as well. It will give a much more informative and accurate assessment of the quality of the match and the magnitude of the victory or loss. (To take this a step further and add another level of information quality, also capture whether the player was alive or dead at the end of the match.)
- Example: Two players each play 50 matches, of which both had 25 wins and 25 losses. Currently, their ELO would be identical if I understand it correctly. However if Player A won their 25 matches 8-0, while Player B's 25 wins were mostly 8-7 or 8-6, and they were among the dead on their team almost every time, that's a hugely significant difference in skill, and I don't believe ELO accounts for that right now and it may go a long way in explaining some of the imbalances we experience today.
- If Player B's team is carrying him/her to the wins they have, when Player B's ELO rises and the matchmaker starts assigning them as a higher ELO member of a team, they are unable to contribute enough skill to carry their team to victory and instead we see the constant flux and imbalances as a result, where the ELO thought Player B was more skilled than they really were simply because they had wins, when in reality they were carried by their teammates to those wins.
Second, start tracking battle values for the 'mechs that were fielded and run that data against the wins/losses for the games where BV is captured, to see if any patterns emerge.
- Factor BV(battle value) for 'mechs into the matchmaking algorithm to help better balance the firepower of teams
In addition to the battle value, a
third factor is certain special equipment that can significantly alter the balance if only one team's mechs have it.
- ECM is a good current example of equipment that should be accounted for in a useful matchmaking/balancer function. If in time, ECM becomes less dominant, it could simply be removed from the equation.
Edited by jay35, 15 March 2013 - 12:15 PM.