View Full Version : [SUGGESTION] Hybridized Matchmaking

02-12-2017, 01:50 PM
I might be wrong about this, but as far as it feels to me:

Matchmaking Rating (MMR) seems to be based on the overall performance as a player rather than specific for each hero

Which I can understand to a certain degree. After all most of the mechanics are still shared among different heroes and one can assume that a player who has already a few heroes in upper prestige levels would still be far stronger than a new player even when picking a hero he never touched before.

However a player also will not going to be nearly as good with a new hero as with a hero that he spend dozens of hours playing with already. That leads to the problem that after playing a single hero a lot, the MMR will assume you to be just as strong as that particular hero and it makes it intimidating to try any new one because you will get slaughtered at first, not being as strong on the new one as on the one you know and were matched for.

This happened to me in older test versions after playing Kensei up to prestige 3. Trying any other hero was just a horrible experience and it's not like you can learn actual fighting just from bots. At best you can memorize most of the mechanics while being somewhat bored, but playing against players is different still.

On the other hand. You don't want a veteran to be basically smurfing once for each hero, because he has already got down the mechanics so well, he will obliterate every new player even when touching a hero for the first time.

So here is an idea assuming your MMR System is an ELO-type of system.

Every Hero should have his own MMR for every player and if you play that hero the MMR is being modified according to the normal rules. But even apart from the hero that you play the MMR of all your other heroes is also influenced by win and loss, just to a much smaller degree.

Such a system would still assume that you are getting better overall if you play one hero a lot, but would actually account for not being as good on other heroes. You could pick up a new one and might have a decent chance to be properly rated from the very get go. It is just a matter of scaling it correctly.

What do you think? How would you prefer matchmaking rating to be determined?

Lord Katerchen

02-12-2017, 02:31 PM
I agree that it would be more encouraging for the players to try the whole variaty of classes when they would be matched either with players of a overall "lower tear" or with other players trying a new class after mastering one specific class. The chance of creating equal teams on a skill base would be much higher.

The question is if players with a high overall skill which always pick the same class would play unranked matches since those players get easily attracted by ranked modes.

But well my theory about filtering this is just working if those kind of players will play ranked mode i know that. It's the same kind of "System" we see in a lot of other games.
Like trying new decks in for example Hearthstone. There we have unranked.
Or trying a new hero in a Moba we also have the "Normal" mode.

02-12-2017, 03:41 PM
I agree but this would require major changes to the match flow, to prevent people from switching heroes after MM. So I don't see it happening any time soon.

02-12-2017, 07:34 PM
Well, in case that we have ranked and unranked modes that helps the problem a little bit because you can at least seperate your 'serious competetive play' and 'other play' rating. But even in 'unranked' modes of most games there is an MMR intended to make the matches as close as possible. So it would only partly alleviate the problem.

Good point. I didn't think about it. But switching to a per-hero MMR would not make this problem any worse or any better. Either direction results in your current MMR not to be representative of your actual strength in case of a hero switch.