How works the ratings and matchmaking ? Is it viable?

First thing is I love the auto matchmaking idea. I think it often makes things way more easy and fast than the old system for find good games. But it sometimes goes wrong…

Questions for devs : Is it a on-going or a future priority to upgrade the quality of the matchmaking and rating system ? Is it impossible ? Do you find it ok despite the story I relate here (maybe you could tell me where im wrong ?

With this post I hope to convince ppl that the match making system is not perfect and need ameliorations.
I try to be the most objective as possible in every word I use here. So don’t see in this post a frustrated player who just want to complain after some losses.

It was a bit late on the night (1am UTC), so maybe there were not a lot of players.
I’ve selected all setting for treaty and supremacy (1V1 to 4V4).
Despite the late hour, a sup 3V3 teamgame launched after a very short time of research(3-4 min). The game ends very quickly. Trying to not be frustrated, I launch a new one. Same thing : “a sup 3V3 teamgame launched after a very short time of research(3-4 min). The game ends very quickly.”

After the second game, I decide to write this post.

I checked the ratings of every player of this last game:
_opponents were ranked 59th, 55th and 32nd in “sup team” game mode and they all have around 2070-2100 “Elo” points.
_ My first mate was 3 205th and had 1185 “Elo” points. My second mate never plaid a single game on his account (not even on another game mode) so he had 1 000 points as each new player (btw he lost 0 Elo points after the game cause he was still on 1 000 points). And I was 56th with 2071 points.

Ofc the game was a no match. I dont think it makes any fun for the winner team, my “1185 Elo” teammate, my “1st game plaid” one, or me.

About ratings gain after the game :
I’m not sure but I guess opponents had a +0.
My mate who never plaid a game got a -0
I got a -12

the story about the first team game I plaid that night was exactly the same : me and two “lows” (very bad from my appreciation - explorer who dies after one min of game against the first treasure, vills idle for 20 sec after recolting the crates, very unusual deck, etc…-) but I can’t tell their ratings cause I didn’t checked. But the thing is I lost exactly same amount of points than the other game : -12.

From experience, if 2 players of a same team win a team game, the one player with the lowest ranked of the 2 will gain more points. And at the opposite, when they lose, the one player with the lowest ranked of the 2 will lose less points.

The issues are for me: 1) the matchmaking should not consider such disparities in ratings for pairing people (even if my 2 teammates and the 3 opponents were waiting for longer than me) ->because the game has no sens and is frustrating for every one.
2) I think the game was objectively unwinnable from our side (Hope you can agree from the ratings of every player ? ) -so the issue is the bit the same than 1) -
3) the formula for points gained and loss is weird when there are big disparities between players. IF my opponents got +0 : The ratings gained and loss are not a 0 sum system : (+0)+(+0)+(+0)+(-0)+(-12)+(-x)<0. Which is weird but can fine if the formula is accurate in games with less disparities between players and the system doesn’t create games where the formula isn’t accurate.
IF my oppenents got +1 or more : its too much reward for the difficulty
In both cases, the -12 for the high player is a bit rough when u know that a fair game win/lose gives around +14/-14 (from my experience)

I end the post with some subjectives and personal thoughts :
_This can happen to every one and I think its very negative for the game. Even more if ppl from the losing team are flaming each others.
_ I think there are similar issues with 1V1 queue where ppl with too big disparities get paired, maybe just make ppl wait a bit longer ? (and make a pop up / notification when a game is found when u were using other applications from ur computer while waiting a queue)
_ Ratings gains are inaccurate and I think it makes ppl less going for team game if the dont have mates they queued with (at least it will be true for me for current state). It might kill the matchmaking system for team games ?
_it doesn’t help friends with big disparities in level to play sometimes together because it will boost the lower friend and derank the higher friend (for instance :currently lower friend will gain +18/-9 and higher friend +9/-18 - or in the exemple from this topic : +0/-12 and like +40/-0. Maybe it is a bit too much difference ??

_ideas ? :
-if adding friends on the game was possible you could queue/ or play normal games with ppl u met with the “good games” made by the match making to avoid this issue.
-(I guess opponents I speak of in this topic queued together, else making more fair teams would have been very easy) The matchmaking could tend to make games a bit in favor of random teams (I mean where players didn’t queue together and get paired in the same team by the match making system) by , for exemple, substract 30 points from ppl who queue alone compare to teammates who queued together (maybe u have stats from both aoe2 and aoe3 between the % wining rate of team of players who queued together ?)
-In League of Legend, to avoid this issue (not best answer), ppl can’t “duo queue” over a certain Elo (around master or something)

Ty for reading. Please feel free to comment :slight_smile:

1 Like