Toward Better Matchmaking

"Globuhl;c-1693079" wrote:
The problem with that kind of approach is that you need some rounds for the ranking system to kick in. Why not considering only the 20 or 30 stronger characters? For most, those are the characters you'll end up using anyway. And later on, once some rounds are behind us, yes, the algorithm should definitely also taking our history of victories and defeats into account.

I didn't want to get too technical about how Elo ratings work, but there is a parameter that will determine how quickly ratings "adjust" to new information. You can use this to help the ranking system get calibrated faster. In my experience, it doesn't take long for these ratings to start being quite accurate.

Matching based on strongest characters is a good idea too. However, there are some kinks one would need to work out. Is it strongest based on GP of those X number of characters? If so, then someone's G11 5 star Traya might not make the cut while their G12 7 star Savage would. Yet I think we would all rather have the Traya than the Savage!

Which brings up another issue: character quality. GP undervalues characters that are still viable at lower stars and/or gear, like Traya, Nest, Nihilus, etc. who have strong mechanics. Just focusing on the top X characters in terms of GP would miss that. You would probably want something like a weighted average of the X strongest characters' GP as well as the player's total GP as two separate components. Better yet, some kind of "character strength index" that accounts for the strength of certain mechanics, powerful leader abilities, etc. but that's probably not be feasible.