Ranking system: pure match fixing

Competitive Discussion
1

All matches are fixed. There is no doubt about it. It is statistically impossible to first win 10 games in a row and then lose 40 games in a row, unless an algorithm behind the curtains is pulling the strings, making sure you end up in the worst teams possible. And in fact, the devs have admitted this is happening: you are given programmed losing and winning streaks (at least games for which your winning/losing chances have a very high or low percentage and thus basically a predetermined outcome).

I’m not sure what the devs were thinking, but they clearly have no idea how frustrating it is to win 2-4 gold medals time and time again, apparently being one of the best in your team, and still lose, lose, lose and lose nonetheless. Not once, not twice, not five times, not ten times, not twenty times, but like fourty (!) times in a row. Because apparently no matter what you do, the rest of your team the system gave you sucks so much that it basically becomes a 2 or 3 vs 6 fight. The two times I won during my last losing streak was because our team had a platinum smurf in it who was barely (!) able to make our team win (first time), and because the other team had 5 leavers (second time). I don’t want to win like that.

You can shout git gud a thousand times, but this has nothing to do with getting good. It has everything to do with an algorithm screwing up things on purpose. I’ve seen leavers, I’ve seen smurfs and I’ve seen super smurfs, and for some reason your system makes sure I end up with them. If the outcome is predetermined anyway, I could just as well have my cat play.

Of course, I’m human. I make plenty of mistakes. Sometimes I even totally screw up. But nothing, nothing in the world can justify losing streaks like these. We’re not talking about losing 50% of the matches. We’re talking about losing consistently every single match played. This is mathematically impossible. Either that, or I suck so immensely much that you should never have given me 1600 SR in the first place.

In that case however your system sucks too, as it is apparently not able to scale people well from the beginning. I started at 1500 last season, went down all the way to 500 (!), spent weeks on working myself up again to 1300 or so, then was placed 1500 this season, went up to 1600 and then all the way to 1200 again. So basically your system estimates I’m somewhere between 500 and 1500. If prices in our economy fluctuated like that, all hell would break loose.

The system is screwed. Hence I uninstalled the game - for once and for all. No one will care, but I just wanted to add my voice: don’t believe the lies about the game rewarding you for becoming good. It’s one big lottery, and on certain days the system decides it hates you and will make sure you cannot win not even if you made 200 kills in a match.

That, developers, is not called a game; it’s called a lottery. I’ve been gaming for 35 years or so, and never ever has a game make me feel so frustrated. I can handle being bad; I cannot handle unfairness. In fact, very few people can, and that is exactly why there are so many toxic people in this environment. You have created your own little monster.

I’ve put more than 100 hours in comp, so I do know a bit what I’m talking about. Don’t tell me I didn’t try. I gave your system a fair chance. You let me down.

A few tips to help you give those who decided to stay a more pleasant user experience:

  1. Punish leaving. And I mean really punish: 1 leave is an accident, 2 leaves is an accident, 3 leaves is a full-month ban.

  2. Punish smurfing. Someone who is structurally overperforming or underperforming on his account should get a permanent ban. I don’t care how much money he makes you.

  3. Stop cuddling and patronizing us. Give us the cold harsh SR we deserve and don’t try to influence it with programmed losing and winning streaks. We want neither. If my level is 1200 and stays that way for 2 months, I can live with it. The more you try to influence the scores, the worse your system will become.

In fact, it makes everything unreliable. People may judge this post based on the stats linked to my profile, but those are screwed up too due to the programmed losing or winning streaks you are giving players: don’t you understand that everything has become a lie? Now who will believe that plat players are actually good? Who will believe that bronze players are actually bad? No one will, because no one knows anymore.

  1. Log all chat and if possible audio. Permaban anyone showing toxic behavior if reported by a reliable source (what is reliable and not is easy to determine). Toxic behavior includes telling my 9-year old son, a 2000 SR Genji, to f off and shut up just because his voice sounds young. That has happened countless times. Your environment is the most toxic environment I’ve ever seen in a game. Your game is not kid-friendly, period. I couldn’t care less to be honest as I’m a dad who can communicate with his son and explain him most adults are more childish than he is, but don’t sell your game as kid-friendly. it absolutely is not.

It’s such a pity. The game is amazing. You have gold in your hands. And this what you do with it. What an utter waste.

2

I would like to see a quote on that. Pretty sure the devs have said that teams are composed in a way so that both teams have a 50% winning chance.

I haven’t heard of a 40 game losing streak beforehand, so that’s something unusual indeed. There are YouTubers out there who give advice and tips for free though; you just need to send them a recording of your game play.

3 Likes
3

The programmed winning and losing streaks, or at least the messing around with scores on purpose part, have been confirmed in an interview with PC Gamer: Google “Overwatch ranking meddles with your skill rating and your mind”

_During the first 40 to 50 matches, however, wins increase your skill rating by more, until you’re back up to your actual level. It’s a bit sneaky. “We do that to give you a sense that you are improving over time,” explained Mercer.

So there you have it: Overwatch is messing with you, making you feel like crap so that you feel better when you work your way back up to your previous rank.

As said, I absolutely have weak points, and sometimes I totally screw up. I’m definitely not a gold player. In such cases I apologize to the team and deserve the loss.

I really don’t mind practising to get better. The very fact that I put so many hours in comp hopefully shows this. But the system must be fair to make it worth my while. Currently it unfortunately is not.

4

I guess that’s an old quote referring to the previous placement system that placed you lower in SR initially, so that you would experience more wins as you get going. This quote is in line with that, because it says “…back to your actual level.” It’s pretty much the opposite of your argument, since it means that you would experience winning streaks.

However, placements don’t have that artificial SR drop anymore. That was removed some time ago. Your SR is always representing your actual skill level now.

5

That’s why no one ever makes permanent progress on the ladder, right? Oh wait…

3 Likes
6

Not everyone’s experience is the same. Your success does not deny my failure, neither does my failure deny your success. But can we agree that clearly somewhere something is wrong? Besides, the devs have admitted themselves they are fiddling with numbers to steer things their way. Fiddling with numbers is called match fixing, and in the end makes both your and my results invalid.

I’ve actually experienced winning streaks too. Shorter than losing streaks, but clearly wins that I actually did NOT deserve and were pure luck (see - I’m honest about that). For example because we got smurfs, or because the other team had 5 (!) leavers, as mentioned above. In both cases, I’d rather have lost.

1 Like
7

Can you explain what “their way” would be? Because the Scott Mercer quote implies that you get SR gains for free, so…

If something were clearly wrong, then I’m sure pro-players would be the first to complain. The only complaints from pro-players about the SR are rather specific though and it appears that they’re otherwise pretty much fine how SR is calculated.

1 Like
8

I don’t know what “their way” is, but as Blizzard is a company, it probably has to do with money. Fact is, I’m definitely not the only one who gets stuck in these loops, and people generally are not that inconsistent. The very fact that I’ve seen both 500 and 1600 SR up and down within a time span of say 2-3 weeks in itself is proof enough of the fallacy. That, plus the fact that wins and loses never come alone: you either win and win, or you lose, lose, lose and lose. Hardly do wins and losses alternate each other. Any statistician can tell you that this indicates that something is at work behind the screens influencing things. Try throwing six twenty times in a row with dice: it won’t happen.

That, plus the fact that apparently there is a hidden MMR score the devs dare not tell you about. Why not? Systems like these need total transparency to be believable; currently they are not. There’s only one plausible reason for this: someone is hiding something, and people mostly hide things because getting them into the known won’t make them very popular.

I can’t judge about gold or higher. All I’ve played is Silver and Bronze, and this is the situation.

1 Like
9

…and that’s the strongest argument against a system that forces losing streaks.

Blizzard makes money by keeping you playing the game and buying their merch. There’s literally no incentive for them in making people feel miserable. Quite the opposite: They gave you wins “for free” as per your quote (which is now removed; for making things fairer).

SR is essentially your MMR after 10+ games. My guess would be that it just isn’t very presentable to the user, so they show you a nice SR number instead. For example, MMR might be a floating point number.

I’ve just been promoted from Bronze to Silver. Guess what? Watched pro-players, watched YouTube videos on positioning/game-play, and read hero guides.

Fun fact: You can throw a six twenty times in a row and you’d still be dealing with a fair dice. It’s the distribution that counts as the number of samples taken approaches infinity. Streaks don’t imply that something is statistically fudged with.

2 Likes
10

I don’t know. Clearly smurf accounts make more money than a guy like me who only has one account. And I’ve seen many smurfs in bronze. A widow making 60 kills in Bronze is not a Bronze player. We all know this. Especially not if he brags about it in public chat. This is just one example. Not enough is done to combat this kind of stuff, as it’s clearly not in Blizzard’s interest.

That said, I’ve genuinely made 49 kills in one match with D.VA in Bronze, on point, and still lose the match. I can give you countless examples of this.

I’ve been watching pro-players too. My other son, who is high gold, has been sitting besides me watching me play, giving me tips. In the end he was just shaking his head saying: there’s nothing you can do about this, dad. It’s not you. It’s the game. A high plat player friend watched me play in QuickPlay when my SR was like 500 (!) and said there’s no way that number was right at that time. Now this is all anecdotical, so you’ll just have to trust me on this.

Thing is my stats are worth nothing at the moment, except the records, as the average numbers are highly skewed now due to the last incredible losing streak. Not that it matters anymore: I’ve quit and won’t come back.

Mathematically it’s of course the distribution that counts when the number of samples approaches infinite, but that still does not explain the high number of streaks, which have not only been observed by me. No matter how you regard the matter: the system is overcompensating and undercompensating on purpose because it got things wrong in the first place.

11

Okay okay let me explain what Mercer was talking about.

First know your SR is your “rank” against other players, while your hidden MMR is your “rating” of how good you are at the game. The matchmaker exclusively uses MMR.

Now they have simple system to ensure your SR is always representative of your MMR: bonus gains. If your SR is to low for some reason, you gain more SR, and lose less, until it catches up. This is most obvious for decayed players, who have their SR lowered a lot, but their MMR is the same (wouldn’t want GMs put against plats just because they didn’t play for a couple of weeks!).

So, this is the “trick” they decided to do. They lowered everyone’s SR after placements by a couple hundred. Your MMR was untouched. This way, everyone climbed SR at first. Obviously not everyone can win, the average win-rate will always be 50%. With this, EVERYONE could climb SR for a little bit. But players didn’t like it and they removed it.

Either way, it did not “match fix” in anyway. It didn’t even impact matchmaking. It was purely an aesthetic change to your visual rank.

They are that inconsistent. Every game ever has people go on big loss streaks. You can look at Dota 2 with visible flat MMR gains. Or even Starcraft 2 which is 1v1 (no team to blame). Both have people get on long streaks. You don’t always play at the same level, you tilt when you’ve lost, you get in your head etc. This is just how humans are. Ever hear of a pro sports player being in a slump?

Now assuming your streaks are accurate, you do seem to be unusually inconsistent. You should investigate why you are that way and look at the opportunity to fix it and improve yourself, instead of blaming “the system”. Consider taking breaks if you’ve lost twice, and look at some stuff on combating tilt…

www youtube com/watch?v=dUJySDueJI4

Finally, if you REALLY think it isn’t your fault, why don’t you record/stream the next time you get in this “match fixed” losing streak and show us. It would be easy to prove. Tons of people stream this game all day, but no one has ever recorded a “force loss streak”. Isn’t that a little suspicious?

3 Likes
12

Performance-based SR adjustments are actually fairly easy to “exploit”. You can still climb even with a slightly lower win-rate than 50%. Worked for me. Seems to work for a lot of other people. Sorry that it doesn’t work for you.

13

Thank you for your explanation about the MMR. As for inconsistency: getting 4 gold and 49 kills on point is not what I would call a slump at this level. Still, I got rewarded with a loss. Not once, but dozens of times. I didn’t get 49 kills every time of course, but as I wrote I was consistently getting 2 or more gold medals. Medals don’t say anything about my teamwork of course, but they say something: at least the game thinks I don’t suck that much.

Anyway, as said, the game has been deinstalled. Do with this feedback whatever you want. I won’t record another streak as I simply chose to spend my time in a more productive way on other pastures. You believe me or you don’t. If you think streaks like this are perfectly normal, then we’ll have to agree to disagree, I’m afraid.

But no matter what you say: a D.VA getting 4 gold medals and 49 kills and consistently numbers close to that does not deserve 500 SR. And I think that even if we totally leave out my performance, something should be done against leavers and smurfs, who have an enormous impact on your SR too at these levels.

That said, I do want to thank all of your for your time and feedback. I did meet two friends in Overwatch with whom I’ll explore World of Warcraft, and that’s worth something too.

14

I don’t necessarily agree its fixed, But I do find it weird that many if not most players dont have much of a win/loss rate above 50% even seasons when I started off so hot and were 10-20 wins above 50% there was always a streak to even it out.

I’ve ignored my SR, It is absurd to think you could determine how good or bad a player is based on the result of 6 random players winning or losing. There is so much flaw in that logic.

Medals mean nothing, If the whole game your team truly sucks and no one has more than 1 elim, then 2 elims is gold, and that isn’t good either. Don’t look at medals. Focus on your stats and getting better, not on SR. I know how good of a player I am, and its better than my SR thinks. Put me against anyone and I feel confident. In a team game like this, Solo Q you can not control 5 other players. if just one player is any bad, or has an off game that can cause the team to lose. Does that make you any worse of a player? Not at all, and its a joke that a game like this views it as such.

If a baseball player hits 3 home runs ever single game for the whole season and that’s all the runs the team scores, but his pitchers give up 4 or more runs every single game, then the team loses every single game, and the team would “suck” however under no circumstance would you ever call the player who hits 3 home runs every game a “bad” player. He would be considered the best player in the league if he was doing that.

The whole win = Skill raise, Loss = Skill drain system is so flawed, inaccurate, and pathetic its not even funny. Don’t focus on winning and losing, which if you solo Q is something you really can’t control, and just focus on putting up the stats and whatever happens, happens. Its less stressful that way and honestly the only way to logically play this game.

1 Like
15

A 40 loss streak is such an outlier, I’d wager it’s something far less sinister from Blizzard and could be something as simple as a change in your settings or internet connection.

16

Yeah…this is BS…if someone is on a winning streak, do you think they’ll team you up with others on a similar winning streak? No…they’ll pair you with Hanzo’s that need carrying and Sombra’s who don’t have a mic…if you win too much…THEY’RE GOING TO PENALIZE YOU IN THE WAY THEY PAIR YOU UP TRYING TO BRING YOU BACK DOWN TO WHAT THEY CONSIDER SHOULD BE YOUR RANK!

17

This happens to everyone, in every game and sport. If you play with a team, you are going to occasionally have good games and still lose. That’s life.

Now, if this happened on a streak it would be surprising. If you truly performed well and lost 5 times in a row (far less than 40) I’d be shocked and love to see it.

Leavers are banned very quickly, you can’t do much more with them without banning people who legitimately just got unlucky. As for smurfs, I agree with you on that. There seems to be little to actually stop people from smurfing. I think Blizzard should set something up to investigate and ban obvious smurf accounts.

I mean, you can go to masteroverwatch com and see tons of players with well above 50% win-rates. But you are right, most approach 50%, which just means the system is working. If a matchmaker worked well, everyone would quickly approach 50% as they get put into balanced games with/against players as good as them. That’s not a bad thing.

This is entirely wrong. Winning and losing is the only way to accurately measure skill, which is why every system with Elo and MMR is based on it.

If you are legitimately better than your rank, you are a “bias” that should give your team an advantage every game. This doesn’t mean you will every game, but you will win more than you lose, thus making you climb rank. Your teammates will, of course, be just as good as your enemies (on average) so your performance can influence your overall win-rate.

The fact the teammates are random is WHY it works. After 100 games, you might have played with 500 random teammates once each. In comparison, every single game you were on your team (obviously). So while you don’t completely control the outcome of a game, no one does, but you are ALWAYS a factor. After 100 games, you’ve impacted your rank 100 times more than any random player on your team.

There is overwhelming evidence, both in mathematical proofs and empirical results, that this approach works. You are simultaneously saying this isn’t a good way to rank players, and complain that so many people are close to 50%, which is a result of them being accurately ranked. Which is it?

I actually agree with you here. People get too wrapped up with their current SR. Focus on playing well and improving, and you’ll naturally climb. If you tilt over losing the occasional game that was out of your control, you’ll play worse and drop even more.

1 Like
18

I must be favored by blizzard since after 57 games i’m at 72% winrate this season with occasional 1-2 losses.

19

Agree to disagree then. I do not see how you can tell how good or bad a player is based on how lucky or unlucky he was with teammates he was given. Sometimes you get good teammates sometimes you get bad ones. Thats why I say put up the stats. If you consistently are putting up the same stats every game, High elims, low deaths, high OBJ time, those to me show how good a player is more than how many times his random teammates helped him win or lose.

1 Like
20

That’s why I am explaining it to you. Saying “agree to disagree” is just avoiding the fact that there a simple explanation that shows you were mistaken. Let me break it down

  • Your teammates are random, but on average, they are just as good as your enemies (and the MM explicitly tries to make them the same)
  • If you deserve to climb, then you are better than your teammates AND your enemies
  • Since your team has you, a better player, they are MORE LIKELY to win (not guaranteed)
  • As you win more OFTEN (not every game) you climb

Consider if a full 80% of your games were out of your hands due to “lucky/unlucky teammates”. The games were decided at the start, you couldn’t have done anything to change it (despite deserving a higher rank). 40% will be wins, 40% will be loses.

Then, the last 20% are close games, well balanced, tight. The type of games where one fight changes the outcome. Since you are a good, 4 gold, player who should be ranked higher, you should be able to tilt these well balance games in your teams favor. Make the difference, make them win. So you win these 20%.

Now you have a 60% win-rate, which will mean you steadily rise up the ranks as you play (until you cannot maintain that). This is even with a gross over-estimation that 80% of your games are out of your control.

This is the “intuitive” explanation, but this is an actual field of research in machine learning. If you’d like, I could point you to some of the literature on it, which leads to the same conclusion. Win/loss is by far the best way to rate players in any non-trivial game.

2 Likes