I posted this on aoezone too, but thought that the official forums might benefit from the example… apologies if you read it twice…
So, A bit of a long scenario incoming, I apologise in advance. The tl;dr is that I think that the new system is going to be awful for casual players, especially those that play with a variety of teammates of a slightly higher skill level on a regular basis, who will lose most of their games.
I’ll start with what I consider “fair”. For me, fairness is that you go into every game with a 50% chance of winning. After maybe an early adjustment period, a “fair” ranking system should try to set every match up with equal teams. Therefore, if we have a “fair” system, every player should go into every game with a 50% chance of beating their opponent and over time, everyone should have a 50% win rate.
So, let’s have a scenario where there is a group of 6 players, each of them has a slightly different Elo, player A is rated 1500 1v1, player B is 1400 1v1, down to player F at 1000 1v1.
Assuming that Team Elo = 1v1 Elo (the exact number really doesn’t matter), as a team of 6, their average Elo is 1250, if they all play. Let’s say they meet every Friday night and play 5 games - not everyone can make it every week, so the teams vary, you could have A+B playing some 2v2s or maybe A, D, and F playing a 3v3. Under the current system, team Elo should roughly reflect individual record, so A and B may be rated at 1400 and 1500, and would be teamed against another team with an average of 1450, which would be a roughly fair game. If A and F play a 2v2, the average is 1250, giving roughly a fair game. Taking the argument that A may carry some games and get doubled in others, given the large skill gap, they should still win somewhere around 50% of matches.
As time goes on, the new system would reward A very few points for a win and he would lose lots of points for a loss. Player F would gain lots for a win, but only lose a couple of points for a loss. As a consequence, within a few months, all of the players have a rating of 1250, because despite F’s lower skill level and A’s higher one, they have played enough games to be pulled towards each other. The next Friday, only A and B can make it and play some 2v2s. They stomp all 5 games as their “fair” Elo is somewhere between 1400 and 1500, not 1250. let’s say they gain 80 points for winning 5 games at 16pt a time, they are now rated 1330. The next week, B and F make it. their “fair” rating is somewhere between 1400 and 1000, maybe at 1200, but the system thinks they are rated at 1290 (halfway between 1330 and 1250.) So they lose all 5 games. B loses maybe 90pts and F loses 70pts, putting them at 1220 and 1180. F then plays some games with D - their “fair rating” is 1100, but they are matched against a 1215 team. F loses again and is now rated at ~1170. Wanting to improve at team games, F decides to play some 2v2s online with no partner, F is still rated at 1170, even though his ranking should be 1000. He gets hammered 6 times before rage quitting, leaving his ranking at 1080.
Next Friday rolls around and A, B and F turn up for a few games. A is still rated at 1330, B at 1220 and F at 1080, average is 1210, but the “fair” average is 1300. As a result, they win 5 games. F gets about 120pts because the system favours him and ends up back over 1200.
F will generally hover at 1200, A will generally hover at 1300, but the team average will stay close to 1250, but what matters is F’s journey above. F played 18 games - he went on a 12 match losing streak before winning 6 in a row. this pattern repeats itself regularly and F has a win percentage of somewhere around 35%, the same thing happens with A, but the other way around. He wins 65% of his games, but stays at ~1300.
F gets demotivated by losing 2/3rds of his games and quits the group. G comes along, who is rated 900 1v1… Rinse and repeat.
The above numbers may seem far-fetched, but they are a pretty accurate reflection of the group I play with - just before this change, we were rated in a nice spread from 1200 team Elo to 2000 team Elo, and our 1v1s (when we play them) are all in the range 1100-1450, which isn’t a massive spread. After this change, I fully expect for the weaker members of our team to lose ~70% of their team games and get frustrated, so hence I don’t think it’s fair. Compare that with the system as it was yesterday and I think all of us were sitting nicely at a recent win rate of ~50% (feel free to look me up on your favourite stats website). We lost some and got sad, we won some and felt like gods, we looked at the fools spending 18h a day gaming the system so they could have a 4500 rating for… what? We were having fun. I worry that we may have a bit less fun now, which makes me sad and triggered this rant.
To predict the counter arguments:
- Yes, I, a 1450 could decide to play with “better” players, but I know these guys IRL and we get on and it’s nice to have a little chat over d/c and not just play games, it’s about having fun with your friends
- Yes, it’s not as simple as 1500+1000=1250+1250, but it’s a reasonable start, and better than just relying on how good F’s rating currently is in team games because he won a few last week. The bigger the gap between A and F the less they should probably play 2v2s, but we’ve actually had pretty reasonable games recently.
- Yes, it is annoying that there is a way to game the system. I get that people are resigning until they were ranked elo 200 to then team up with a guy ranked 4500 and beat up on people at 2250 because both of them are actually 2500 players, but seriously, they are wasting a lot of their own time and the whole “don’t resign for 5mins” blocker did at least make that less easy for them to do.
- Yes, it’s annoying when you get stomped by a Pro/Smurf/haxer, but this change will do nothing to change the fact that pros don’t really play ranked TG or that people like to make smurf accounts, both of those phenomena will carry on. In fact, We’ve discussed me making a smurf account so I can play games with “F” from a lower rating than when I play games with “B and C” - to make it “fairer”
rant over.
Please feel free to console me, argue back or just disagree
)
I think they’d have to calculate the winners Elo gain and then split that across the losers rather than calculating the losers individually.