hardest game of your life
“Expected, 19%”
easiest win of your life
“Uphill battle, 21%”
Most one sided unfair game of your life
“Reversal, -21%”
matchmaker is drunk man
31 Likes
They use rng as their mm. No compromises.
11 Likes
I think the matchmaker “estimates” where it thinks you should be and puts you in fixed matches until you get there up or down
1 Like
Yep. ±10 kilometers miles.
4 Likes
Welp there is one thing MMR cant predict, the players. Just because a group of people mashed together might be better on paper if that genji is across map spamming need healing then it can result in tilting etc etc.
Also delete sombra, turrets, junk tire from game. Ty.
2 Likes
so that’s why it puts 20% cahnce to win? because it can’t predict?
Because people are not robots, they may have an estimate but what if they are having a bad day, someone just yelled at them irl, etc.
2 Likes
I’m fairly confident they’ve implemented algorithmic astrology to get those %-age numbers.
1 Like
maybe if it was a 0-3 then the game shouldn’t use “reversal” since there was literally 0 chance that it was correct in that matchup
3 Likes
To be frank, I have no idea why they added these categories. It was clear as day to everyone involved that their mm was garbage before and adding an explanation why it is so couldn’t have possibly solved the issue. It has just made it an even bigger clown show than it already was. But then again - they weren’t exactly a paragon of competence for 5 or so years.
8 Likes
This account was gm2 seasons 3/4. Didn’t play, decayed plus the recent reset had me in a diamond 1 vs masters 2 comp game lol. Felt bad for my team, didn’t feel too winnable.
1 Like
Even the lead of matchmaking said it’s impossible to predict match outcomes yet they do it anyway
‘’Most of the time though, the difficulty of a match is dictated by things that we can’t see from just looking at the MMRs, things like players choosing the right/wrong hero, or getting momentum early, etc. Very hard to predict!’’
Quoted directly from his twitter
1 Like
just had a Route 66 match.
4:01 on our clock, they got held at first the entire time.
“Uphill Battle”
if I just lost like that and saw “reversal” on the end screen I would have contemplated offing myself
2 Likes
doesnt matter, it matches one side to have a harder time to win, surpirse surprise it’s not gonna win, wow, such hard prediction, how did the system know?
2 Likes
Aside from them being extremely inaccurate, it’s dumb how these modifiers only change your amount gained by 1-2%
Personally… I’d guess that even trying to implement this system would have major inaccuracies even if you tried really hard to make it as accurate as possible by considering all of the variables.
Take two players that are in the same rank. They are not only going to still be differently skilled despite being in the same rank and division, but one could thrive a lot more than the other on certain maps or could thrive more than the other when they play in a certain comp or against a certain comp, etc.
I think people even think “If I’m Gold and hypothetically I were to always get into only Gold lobbies, I should automatically expect to have similar outcomes as any of my other matches.”
There are so many variables, then add in the fact that you are playing with 9 other players that also each have their own variables.
Now try to make a system that detects your chances of winning simply based off all the players in the lobby’s SR.
Like I highly doubt they designed the system to be like “well this player is 2500 SR, but they generally struggle on this map, so punch that into the calculation” along with every other possible variable and then do that for every player in the lobby to dictate which team is favored.
Pretty sure it just takes the average SR of both teams and if one team is higher, then it states to favor that team to win and by how much depending on how large the gap is in the two team’s average SR.
That isn’t an accurate way to determine the favored team. In fact, it’s pretty inefficient and poor way as it overlooks a lot of other determining factors.
They probably shouldn’t have added the feature in… at least not like how it’s presented.
That’s not exactly what it’s doing.
OW elo is probably similar (but different) than how chess elo works.
You aren’t going to go up against another chess player that has the exact same elo as you. You’re going to play against players that are a little lower or a little higher than you.
If you get matched against a chess player that is higher, then hypothetically on paper, the game is naturally favored to the higher elo player. So if you lose against this player, you don’t lose as much elo as if you lost to a player that is the same or lower than your elo (this is designed so that if you were to keep playing against higher elo players than yourself and losing, you don’t just plummet down… because then you would be playing with players that are much lower elo than you would most likely stomp them).
It’s also vice versa. If you beat a chess player that is lower elo than you, you get less elo awarded to you. This is also to prevent you from playing a bunch of lower elo players and shooting up way beyond your skill and having to play against players that will murder you.
So in OW, what the system is trying to do (poorly imo) is determine which team is more skilled based off their SR and by how much. Once that’s determined, it just states that one team is favored to win, because they have a higher average SR.
That’s always going to be the case, because matching every player in a game exactly with the same SR across the board for both teams is too difficult (queue times could be hours if not longer).
So instead it “tries” to match the teams close to the same average SR rating, but because one team will always be at least slightly higher than the other, it then “favors” the higher rated team. It’s not actually implementing any favoritism, it’s just guessing what it thinks the likely outcome would be based on the SR ratings of the teams.
But one should note that there are so many variables outside of just a players SR, that this isn’t super accurate. Just because one team is considered to be more likely to win, doesn’t mean they are gong to win.
Even in Chess (which is only 1v1 scenario) there have been times where the top 5 players have outcomes where guy holding the top #4 spot, can’t beat guy in number top number 3#, but top number #5 can beat number #3, but can’t beat #4 or anyone else above #3 and #4 can sometimes beat #2.
3 Likes
it certainly has been interesting to see now that we actually get told if the enemy or our team has higher mmr now, how little correlation there is.
however, it is important for the integrity of sr and mmr that the system rewards us if we win a match against higher mmr and punishes us if we lose to a team that had lower mmr.
it doesnt really matter if the high mmr match is actually hard or the low mmr match is easy. what is important that just like in the chess elo system, the system always treats everybody according to their mmr.
i would die to take a look at the actual data though. like, is there any sufficient correlation between estimated win chance and actual win chance. how often do favoured teams win vs lose? if favoured teams for example lose more often than win, then somethis has to be severely wrong. the law of large numbers dictates so.
the modifiers change the sr gain/loss according to how unequal the match was (according to mmr difference).
basically if a full bronze team went up against a full gm team and won, the modifier would be completely of the charts.
so, low modifier numbers actually are a good thing in that context, meaning the mmr difference is rarely big.
if the matchmaker would be perfect or had an infinite player pool, it would never match unequally and so there would never be modifiers.
1 Like
At least the percentage changes are small. They seem kind of pointless since they’re usually at 1% each. I’d like to see more modifier info post-match and maybe more substantial gains/losses because of them, but only if the info is strong.
1 Like
The “reversal” mechanic is completely broken, i have 400 comp games this season and so far it it has only applied when i get completely stomped and the enemy team full holds me on first point.
They sure pulled a reversal on us.
1 Like