Dote Night: MMR Problems And A Potential Solution

Part of a miscellany of serious thoughts, animal gifs, and anecdotage from the realm of MOBAs/hero brawlers/lane-pushers/ARTS/tactical wizard-em-ups. One day Pip might even tell you the story of how she bumped into Na’Vi’s Dendi at a dessert buffet cart.

Above is a picture of two bears. I have been watching a bearcam while I try and write about matchmaking rating (MMR) in Dota 2 [official site]. Eventually I started imagining the bear you can just about see in the background shouting at the one on top of the little waterfall “WHAT’S UR MMR, SCRUB?” as they both fail to catch any fish.

I guess that’s as good a jumping off point as any: I don’t associate MMR with anything good. I associate it with miserable gaming experiences and being used to undermine people in discussions.

Before I get any further, I want to be clear that I’m talking about *visible* solo MMR – the number Valve shows you after you’ve played a certain number of ranked calibration matches. I’m not talking about the invisible ratings that they use to put you into groups in non-ranked play.

I’ve written about MMR before so here’s the theory as I outlined it then:

The point of matchmaking as Valve describes it is to set up games of Dota such that each team has as close to a 50% chance of winning as possible, that a five stack of players (likely on external voice comms) isn’t matched against five individual players speaking different languages, and that the spread of skill is similar on each side. Essentially matches are supposed to be balanced and fun.

One of the main factors it takes into account for all this is the MMR. There are five different MMRs at the moment and they depend on the type of game you’re queuing for as well as who you’re with. You’ll have a number assigned for solo unranked games, party unranked games, solo ranked, party ranked and team games.

When people ask about your MMR it’s the solo ranked number that they mean. That’s also the one people are talking about when they mention the “Road to 6K” challenge. In order to change that number you need to win matches in solo ranked mode. For the general MMR area I’m in right now a win gives everyone on your team roughly 25 points and a loss takes away roughly 25 points. The idea is that you move up or down until your number stabilises because you’re winning as much as you are losing around that particular number rating.

Having a high MMR (6K is considered high, hence that challenge) is therefore correlated with being “objectively” good at the game. It’s treated as a badge of honour and, often, authority.

There are various strategies people suggest for improving your MMR:

Play a carry and then if you’re good enough you’ll be able to brute force a victory.
Play a support and then you’ll be able to control the pace of the game and give your carries the best shot at snowballing.
Play something independent so you can do well if your team is shit.
Play something team-focused so you can save their sorry asses when they screw up.

And so it continues with some options making more sense than others…

Thing is, the strategies are different but the mentality informing them is often pretty similar. By that I mean they tend to assume that you are a) being placed below your REAL skill level and b) that your teammates are liabilities. They’re not people, they’re weird, unpredictable NPCs to be corralled into formation as best you (the blameless hero of the piece who is being hard done by on the MMR front) can.

In my personal experience ranked matches are far more hostile than unranked. If you’ve found otherwise please do say – I’d be interested to hear your experiences. But in my time with the mode the players seem more prone to blaming others and less inclined to help one another. I’m convinced it’s because of those MMR numbers and that aforementioned mentality. You want your hit of nice green 25s – you deserve them – and every mistake a teammate makes gets in the way. This isn’t the case in every match, but it’s been a common experience for me and one I ultimately stepped away from. Unfortunately not before I’d decimated my own MMR while trying to improve it ;)

“Does that number matter?” is a question I’ve asked myself a few times now that I have the damn thing.

It does and it doesn’t.

It doesn’t because I know when I’m getting better at Dota and when I’m stagnating. I know when I’ve got the hang of a new hero or worked out a new skill build or pulled off a tricky set of actions or found a fantastic ward spot so we can catch chumps unaware. I truly don’t need a number for any that.

It does because I now have that number – for better or worse – and that number means something in the wider community. Sometimes when you play, your MMR is a factor – I’m playing an industry tournament at the moment and they asked for everyone’s MMR so they could try and work out the bracketing for games. When emailing mine over I felt horribly self-conscious because I know it’s about half of one of my teammate’s. It has never gotten in the way of us playing together and we have very different in-game roles and styles of aggression which complement one another. Generally I don’t think about our MMR, we just play together as a team and accomplish cool things but at that moment I was reminded about these numbers and it felt bad.

I felt self-conscious again when an MMR of around my level was being spoken about as “low” while on a Skype call despite (I hope) that I’m considered an equal to the people I was with.

I see Redditors and commenters asking each other or asking authors for their MMR as if it impacts their ability to make observations about the game. Making observations about how crowdfunding works? “What’s your MMR?”

At the moment ranked solo queue just seems such an unpalatable option. It’s partly because the matches simply aren’t fun what with the stress and the 1vs4 mentality you’ll see in teammates (or yourself). It’s also because you never get to recalibrate. I did my calibration games aeons ago. The first number I was working with was when I knew far less about the game. Now I know far more about the game but I’ve been through phases of trying to improve my MMR, wiping hundreds of points off it as I go. Regardless of how accurate it is, it’s felt like a weight for a long time. It’s a problem lurking, a broken thing needing to be fixed rather than anything positive.

What I’d love is for those solo MMR levels to be wiped after every International. You have a season lasting a year, you calibrate and then you can try to fiddle with it all you like. People can ride the Road to 6K or Meander Down The Lane to 3K or Fall Drunkenly Into The Ditch of Double Digit MMR but then once TI’s confetti is swept away you start again. Solo ranked never risks becoming this intractable problem because there’s always a point at which you get a fresh start.

Maybe your MMR is lower than average. That’s pretty common, actually, because that’s how averages work. It should be okay. It *should* just mean that you’re matched with people of around the same level so you can get a relatively balanced gaming experience. But giving people that number overtly means it’s used differently. For some it’s a way to gauge their own progress. It can be a challenge or a puzzle to solve. That’s cool. For others its something they feel bad about and which other people use as a weapon. It ends up being correlated with all manner of unrelated knowledge or skill. It can make you feel like you’re stuck.

Allowing for a reset would help alleviate some of the negatives without ruining the positives.


  1. jonbro says:

    (whats my mmr? worse than yours, I promise it).

    Seasons are a totally reasonable response to this. Leagues and ladders are in a bunch of other games, and they work totally well, but you still retain a little bit of your ladder position when you start the next year. As far as I can tell, the only thing that the calibration does is not show your mmr, and allow it to swing much more widely (maybe like 100-300 points per game I am guessing).

    I have been playing more lol recently because I find the ranked just so full of assholes, and without a clear feedback method on the reporting system.

    • nearly says:

      I don’t really like the idea of ladders and seasons. I don’t know what I dislike about them, but that’s been one of the reasons I’ve kept away from LoL and HotS even as I’ve put quite a few hours into Dota 2 ranked play. Maybe part of it is remembering resets on Halo 2 and having to spend a week or so getting stomped by much better players while the rankings got where they needed to be and the matches got a bit more even. Maybe that’s not even true and just something I imagined.

      Still, I’m not sure what really changes given what I’ve seen from the League community in response to rankings. People that want to embark on the “Road to 6K Challenge” will do so: seems unfair to force everyone to try and claw their way up when some of us are quite content to sit where we are and play an even match without the “I must rise!” mentality.

      On the report feedback front, I rarely throw a report to the wind. Most any time I report someone, I’ll get another free report because Valve ended up taking action against that person. Maybe you’re being a bit too liberal with reports?

  2. Steven Hutton says:

    Some other Moba’s do allow a rank reset at the start of a new season. Honestly, it makes the problem worse as much as anything. Now people who place low feel hard done by once a year. “My placement matches went badly and now I’m stuck in MMR hell!”

    “I was totally diamond league last season now I can’t get above bronze because of the damn reset.”

    Also, it means once per year there’s a week during which the match making is really terrible because everyone’s MMR has been condensed into a narrow band.

    There is a fix for this though and it’s called don’t tell people their MMR. If people know the exact number it becomes a kind of points that they feel ownership over. Something they can lose.

    So hide that number. Instead break bands of MMR up into leagues. Then break the leagues down into divisions and break the devisions down into groups of 100 players. Now the worst number you can have is 100. And the best is 1. If you want to track your progress you have a metric. Are you getting better? Well, are you progressing through the leagues? But you don’t know your exact ranking against every other Dota player in the entire world. You just have a small group of people at around about your level to compare yourself to.

    This is the way blizzard handle their shit in SC and it works really well.

    • DedlySpyder says:

      I don’t think it would be quite that bad for Dota though. From what I understand, the 10 calibration games set you against people based on your unranked MMR (the hidden number). So, on the reset week, you would still get paired against equally skilled people, in theory.

      The problem with it not resetting, is that if your hidden MMR goes up (because you calibrated when you first started, for example) your ranked stays the same. So, people who calibrated right away at like 1k are stuck there until they grind themselves up, no matter what they are now.

    • Banyan says:

      I believe MMR used to be hidden and was made public because players were certain that they were losing games because Valve was maliciously putting them in games with noobs. Now that MMR is visible, those players now assume that all the other players are keeping them from their true greatness. On balance, it was better before, but I only play unranked so it’s all academic to me.

      I had this weird moment earlier this year when a team member recognized another player’s voice as being a Dota streamer on Twitch. I sucked at that match but I have a sneaking suspicion that I may be “not awful” at the game if I’m being placed into games with recognized players. And suspecting that you don’t suck is a hell of a lot better mentally than obsessing about how to suck less.

  3. alphagator says:

    Seasons are every few months in Awesomenauts–which I really, really wish this site would examine again as it’s grown a great deal since the last time it’s been mentioned here–and it works a treat. There’s one night of chaos when matchmaking is more or less confused at season reset, but apart from that it has exactly the effects you mentioned.

    Seasons also are interesting because they provide discrete deadlines for whatever goals you might want to achieve (total matches won, games with a particular character, improving your win % or ranking or whatever). It provides a different pacing to skill attainment when there is a time limit.

  4. Sonny Bonds - Lytton PD says:

    I feel similar MMR blues. I calibrated like 1000 matches ago and I generally play support. I calibrated around 1900 and plummeted to 1400ish really quickly. I’m around 1550 or something now and its just brutal because your teammates are either inexperienced, terrible, or both.

    Then again maybe I get exactly what I deserve and I totally suck at Dota??

    • Raiyne says:

      Playing support is just brutal at low-level MMR, because your hero doesn’t scale into the late game, you only have a limited timeframe to exert your skill, and even then, you’re depending on your (shitty) teammates to carry well into the late game. At sub 3k MMR, you can pretty much get by without a dedicated ‘support hero’, just pick something that can have impact throughout the game and you’ll climb easy peasy.

    • Vandelay says:

      Right there with you. I rarely play solo (ranked or unranked,) so haven’t got a solo MMR yet. I did get my team MMR with one friendfriend though, which after a terrible bunch of placement matches (I believe we won zero, although I was playing it like I play unranked and just picking heroes at random rather than ones I am actually good with,) I got about 1600-1700. As happened to you I quickly dropped down to about 1400 after a never ending stream of awful games filled with vile people.

      The thing is, the friend I was playing with had previously done his team MMR with a friend of his who plays a lot and has a very high MMR. He got about 2100. He reported that playing ranked was a much better experience then normal play, with nicer players and closer games. Being in games with players that were about 500 lower MMR was complete opposite. We ended constantly with our carries having double figure deaths within the first 25 minutes and the bile within the team quickly reaching a level where co-operation between us would become impossible.

      Eventually we struck upon a formula. He played Nature’s Prophet and I would play a tanky character, like Wraith King or Slardar (particular if the enemies had cloak.) He would push lanes and I would distract the enemy long enough for him to take down towers and barracks. This worked unbelievably well. I think we got a win streak of 8 before it failed once, and some of those games we won with some truly poor teammates. I still only have an MMR of about 1550, but the quality of players we are getting definitely seemed to improve, at least in attitude if not in skill.

      Unfortunately, it got dull playing the same characters over and over again, so we stopped. I generally lean towards support characters more, which have a terrible time in low MMR games. A team made up of Phantom Lancer, Phantom Assassin, Sniper, Viper and Riki bizarrely seem to be the most effective.

  5. Dukey says:

    It took me ages to see two bears.

  6. Synesthesia says:

    seasons sounds like a very legit idea. I would love to see some form of implementation of it.
    I, also, cannot play ranked. The sheer toxicity and the overall un-funness of the whole ordeal is very offputting.

  7. Oreos says:

    If people didn’t say what’s your mmr when they are arguing, they would say how many wins do you have or whats your win rate or how many hours you have. People will always try to find the upper hand in an argument.

    Having played for 5 years and at 5.5k mmr. I believe Dota plays like this. 25% of games you will win, because you can’t loose because you have a really good player. Another 25% of games you will loose because of your team, or disconnects, or bad picks or whatever. But that leaves 50% that are up in the air. Anything could happen, when you loose those games there is no one to blame but yourself. You didn’t play a perfect game, no one does. Everyone makes mistakes, no matter how small, they can all affect the outcome of the game.

    In those games, sometimes one teamfight can change the way the game is going to go. And if you loose one because a teammate missed a huge ult or something, many will instantly blame the “retard earthshaker” so they feel better. There is a flipside though, the arrow you missed top, the earthshaker remembers that and thinks the same about you. Sometimes the games you feel you where wronged by your team in clutch moments, were those 50% where one thing can decide a glorious victory or a horrible defeat. Learn to love your team, because they are people too and they may even feel that same un warrented hatred for you. the game game is already hard enough at 5v5, dont make it 4v1.

    • Oreos says:

      My god my grammar is trash. I meant 9v1. :)

    • Oreos says:

      My god i suck at proofreading. I meant 9v1.

    • Kitsunin says:

      I think you’re right, in that for most game you could have turned them into a win had you played at a professional level of skill. The issue is just that, too often the odds are against you to a point where simply playing very well isn’t quite enough, because your small mistakes will be compounded by the mistakes of your team. Even though you could have won by playing better, it’s still discouraging that you’re being put in situations which demand an unrealistic amount of skill to succeed at because of the other people on your own team. And this happens constantly at low mmr, it’s pretty much the only reason you ever lose, because you’re always going to have a few people who actually belong at the low mmr while you don’t.

      And that’s what makes low mmr not fun, it doesn’t feel like you’re playing a tough game against competent opponents, it too-often feels like you’re playing well against semi-incompetent opponents. You’re losing through your own fault, sure, but it’s because you have people hanging onto your ankles and weighing you down too much for you to handle, rather than because your opponents are giving you a strong challenge. It’s just far less fun this way.

      • sonson says:

        You sound very sure of the fact that it’s other people’s fault. If you were as good as you say you were, you would be able to win in spite of your team. Better players make less mistakes. You might make less than your team, but you’re still obviously making enough that it counts at the end. Players out of place in their MMR bracket can repeatedly carry a team singlehanded-It’s been done on lots of occasions. You might be-maybe are-better than the people you play with in absolute terms at some level. Just not good enough to be as good as you need to be to play with the players at the next level. If you keep improving, you will get there. If you keep thinking you’re the forsaken chosen one, a passenger to circumstance in spite of your brilliance, you won’t.

        • Kitsunin says:

          Back when I did play Dota I can say with confidence that I was frequently playing the best out of my own team, but still not well enough to turn stomps into wins, typically. I was decent/good but not excellent, and I got started with a desperately low mmr, I was always aware that I made mistakes, but I was also aware that they were pretty small mistakes, and usually to do with motor skills rather than overall strategy. Yes, I know a pro can carry the worst of games, as could too I were I better, and I said that.

          The feeling was that of playing with an anchor tied to your leg, where playing HoTS as I do nowadays, and which I could swear has far better matchmaking (but who knows really, anecdotes and all that) typically feels like struggling against a superior foe. All I’m arguing is that it’s much more fun to lose because your skill wasn’t equal to that of your enemies, rather than because your skill wasn’t superior enough.

          • sonson says:

            You weren’t as good as you think you were, it’s that simple. Put it this way- if there are 4 idiots in your team, and you who you believe is a fair bit better than them, then there will either be also 4 idiots on the other team, and one person who is better than them, or maybe not even at all. Maybe you are the only good player in the match and the rest are thoughtless scrubs. You still aren’t good enough to tip the balance against 5 other scrubs, or overcome the other good player on the other team attempting to do the same.

            If they are as bad as you make out, constantly making mistakes, and you only make a few, then a team making always mistakes versus your team with not as many mistakes, should win out often enough that you climb the rankings. You won’t win every match sure, but you should win more than you lose, if what you say is correct, if you are better than the tier you were ranked with.

            People don’t climb solo MMR on the basis of playing as a team, but on the basis of how their contribution effects their teams ability to win over games. If you were as out of place in your MMR as you claim then your mere presence should have created a consisted imbalance in your teams favour, especially since you claim you the people you played with were basically useless, but it didn’t. Hence, you were where you belonged.

            I played some low priority with a friend who was stuck in there for a laugh last week, all random. We ended up as a a team of 5 supports, characters who the 2 of us had never played before, versus a team of four carriers and supports, we trounced them, why? Because bad players, the sort you describe are easy to beat. Bad players don’t know when to take towers, when to farm, when to rotate, when to gank, when to defend, when to initaite etc. If you know when to do *one* of these things reliably and consistently chances are you’ll win over a team who never know when to do it. if you have an otherwise equal game but you reliably rotate to turn an equal match up into an uneven one, or turn an attempt at a tower push into a completed one, like anyone competent should be doing, over the course of the game your team will snowball exponentially. Even just dewarding in a low tier will probably be enough to give you the edge.

            A guy here
            link to

            climbed from 2900 to 5400 in 150 games. Getting from 2900-3500 took 24 games, all of which he won. He also makes the point that even though he had silly win rates (like 65% plus) you never see anyone with that win rate, or anything close to it, usually. Why? Because people are where they belong.

            The article also lists someone who at 3.5 k took an account at 4.5 and whose win rate went down to 45%. The other guy also points out that even though he had silly win rate (like 65% plus) you never see anyone with that win rate, or anything close to it, usually. Why? Because people are where they belong.

          • Kitsunin says:

            My team wasn’t braindead or anything, I was just good enough to be topping our scoreboard frequently, winning my lane, never feeding, and such. I was sometimes frustrated by teammates that were just kind of wandering around aimlessly and dying repeatedly, but mostly it was annoying how we were getting heavily outfarmed even though it was difficult to identify clear mistakes I was making.

            I’m not saying I wasn’t raising through the MMR — I was, and my winrate was something about 60% by the time I decided solo Dota isn’t fun for me. I believe when I first saw my MMR, it was at about 2,000. After spending a couple weeks playing like crazy, some fifty hours, I was up to 2,500. I’m not saying there wasn’t advancement, but I’m saying it was way too slow, and in the meantime the average player I was playing with felt like an anchor, rather than a challenger.

          • Kitsunin says:

            *50 games, not hours. I didn’t have that kind of time.

          • sonson says:

            If your team mates were average, then so were you. If you were consistently getting outfarmed then there are things you can do about that- learn how to farm better yourself, or focus on how to stopping them farming, or translate an early lane victory into an early rax.

            Winning your lane means nothing. It’ just short hand for getting more exp and gold than your match up in the lane. It’s what you do with that foundation that counts. You clearly weren’t able to use that gold and experience consistently enough to reliably turn it into a *team* advantage, hence why you were ranked as you were.

            If you were as superior as you claim to have been, 50 games is handily enough to climb quickly. There are lots of examples of it. If you were climbing slowly, it suggests that you had the ability to improve your game and were doing so, which I suspect is the case- but you wern’t as decisive a team mate as you thought you were.

          • Kitsunin says:

            Well, it’s hard not to get outfarmed by someone who has ten kills against your teammates, y’know? I know I could have been better, but I also know I was decently below my mmr at the time. I’d been playing unranked for almost 200 hours before I started ranked, and the big thing is that the quality of matchmaking suddenly took a massive dive. Maybe it was bad luck, 50 isn’t a great sample size, but it sure went from typically being challenging games with skilled players on either side to crapshoots where the big question was “Which team’s gonna feed?”

            Does it count for something that my hotslogs mmr is platinum rank? I know how to MOBA, and I have enough perspective to say that the low-mmr Dota experience sucked.

          • sonson says:

            It’s not uncommon that one lane is one sided. From your own experience, in fact. You said that you yourself often won your lane but lost the game. In this experience, someone else won their lane, and they worked out how to translate that into victory. In this example, you lost because your team fed-but in your other one, even though you win your lane (and very probably because the other team fed) you still lose. That should tell you something. Early kills don’t count for shit unless people know how to use the gold, and lots of people don’t. A good hard carry with 1-4 is perfectly capable of having more farm at 20 minutes than the Pudge at 10-2 if he plays well. If you can stack in a jungle and kll 70% of a creep wave reliably in lower tiers you will have a massive gold advantage irrespective of what’s going on elsewhere. If you get gold quick you get items quicker and you can start to get double and triple kills and change the momentum.

            I was in a game last night where we were ahead 33-6, but we were all early gankers and they were all carries. We didn’t push quick enough and we lost, because come half an hour even though Sniper probably had half my farm his ability growth and potential to explode was far more valuable than what I had on my fed YOLO axe.

            But also- they were sensible, ran to their base and weathered the storm and slowly fed on creep waves. If they had kept trying to win the jungle back we would have continued to knock them into next week. But without anything to gank, our ganking line up was useless, and it was gg. We couldn’t work out how to get round the impasse and lost the long game.

            Rather than declare that to be unfair or my team noobs, I accepted that I should have picked differently, or thought outside of the box. If I’d played really well we would have won. I played well at the start, but then tailed off, was out of position alot, made some bad calls, was farming when i should have been pushing, meandered around waiting for fights when I should have been farming etc which is also when the momentum changed. It wasn’t solely my fault- but we would have done better if I had played better, that’s inarguable.

            TLDR: you can easily out farm the enemy in most instances, even from behind, if your skilled enough simply by being good at the basics. If you’re not skilled enough to reliably outfarm the enemy, then you’re where you bleong. Anyone can win when they’re on top- climbing mmr is about doing that consistently and also being able to outskill the opponent from being behind as well. If you can’t do that, it doesn’t make you bad- it just makes you like everyone else in your tier. If you have to rely on your team to keep you on top -which is the vast majority of players- *you are no better than them*.

          • Kitsunin says:

            I didn’t really care about winning or having a high mmr anyway. My issue was simply that matchmaking wasn’t consistently putting me with or against players of close skill. I can accept that there are a million things I could have done better, but what matters is that I played as well as I played, and it didn’t feel like the other people were at the same level. Not consistently so. Because I was never trying to be the best Dota player, I just wanted fair matches so I didn’t study how to carry my team, I assumed playing decently and being a temporarily above-par 20% of my team would be enough to get placed in consistently fair matches. It wasn’t, not on a reasonable timeframe at least.

          • sonson says:

            *You didn’t play well*, that’s the point. If you weren’t reliably winning, the part you played wasn’t reliably good enough.You probably think that most times you won, it was down to you, and most times you were lost it was down to your team. I.E your responsibility at a team member only counts or that your play only “works” when you’re winning.

            It’s called Defence of the Ancients, i.e you have to kill the ancient to win. It’s not called “Win my lane and don’t feed often”, which is what you seem to think equates to victory, but it isn’t. That’s a good start, but it’s not the game, and you obviosuly didn’t know how to translate it into vcitory more often than not, which is what good players do.

            For all the obvious mistakes you spotted in respect to your team mates, they will have noticed the same with you. For all the bad stuff you noticed, they will have done other stuff well that you simply missed or ignored. From the sounds of it rather than noticing you’re lane winning they will have been thinking “Why is this guy who stormed his lane not pushing, why are his items so low, why is he not ganking other lanes or roshing, or farming for a decisive item”, all of which are legitimate questions.

            You weren’t good enough to kill enough ancients, and thus you were where you belonged.

          • Kitsunin says:

            No, I don’t think I won because of me. The only basis for my belief that I was in the wrong mmr, was the fact that the games I was being dropped in were consistently unbalanced. When I played unranked this was not the case, I assume because it had much longer to calibrate my mmr. That’s really all there is to it.

          • jrodman says:

            It’s hard to internalize that there are many aspects of the game. You will tend to play better at the aspects that you value and understand, and not as well at the aspects that you do not value or do not understand.

            Thus, you probably were doing better than average(for your games) at the points you believed to be important, but yet worse than average at the aprts you do not believe to be important, but yet remain important.

            In general, the way to improve is to believe that you are making errors and to try to figure out what they are. The mentality that says that you’re playing better than your teammates generally but yet not improving on MMR is what holds a player back.

          • Kitsunin says:

            I just don’t understand why the games felt so damn unbalanced when I started playing ranked. I only ever had an issue with my mmr insofar as it relates to the fun of the game and the frustration dealing with unbalanced matches.

  8. neoncat says:

    Thoughts on ranking / rating from other games:

    == World of Tanks ==
    World of Tanks doesn’t have a real built-in rating system, but there are several widely-used plugins that calculate ratings based on accumulated match statistics. It does generally correlate to skill, with the caveat that it’s easy to inflate your rating if you play certain tanks which maximize the most significant factors (e.g., damage output).

    Anyways, one thing that’s always boggled me about MOBAs is the idea of playing long matches with random players – they’re, on average, less competent because the highly-skilled players seem to gravitate towards modes other than random, such as tournaments / strongholds / clan wars where you choose your team. At least in WOT, it’s just a 10ish minute match with large teams, so the one guy idling or driving off a cliff or team-killing doesn’t ruin the entire match.

    I couldn’t imagine wanting to play for 45 minutes with just four other random players, even if they’re approximately at the same skill level as myself (and as mentioned above, WOT has no in-game rating system for matchmaking).

    == Card Hunter ==
    … maybe I’ll add something here later …

    • nearly says:

      As far as WoT goes, people do use WN8 ratings as a somewhat reliable indicator of skill, as well as Win Percentage. When the invariable “why do people think Win Rate actually relates to skill?” question comes up, the logic is usually that it does accurately reflect skill because, tracking for as many games as the “average” player will rack up quickly, it eliminates the effect of other variables like teammates or enemies (and over a large enough sample size, vehicle). While MMR in Dota 2 abstracts this a little by tying it to a score, the shifts in score will reflect your win rate which will reflect your performance.

      Of course, then you’re stuck having to respond to a discussion of whether or not matches are balanced or not and that’s a totally different ballgame. I have a pretty decent Win Percentage for someone playing WoT as long as I have, but I’ve never felt that I tanked my stats (no pun intended) on a losing streak. In Dota 2, on the other hand, I’ve lost several hundred MMR in a horrific spiral of negativity and unpleasantness. While some of those games were entirely winnable, the amount of people that intentionally fed themselves or courier after not getting to go to the lane they went to makes it really hard for me to say that the MMR I’m slowly grinding up through/from is where I really belong. I don’t necessarily have any interest in climbing matchmaking, per se, but I don’t feel that I “fairly” lost as many points as I did and do feel that the tier I’m at sometimes plays more like attempting to herd cats than playing the game I was a few months ago.

      • neoncat says:

        Yeah, WN8 is more of a holistic measure, though still manipulable if you really want. (e.g., that’s mostly the difference between the “really really good” and “unicum” tiers)

        As for Card Hunter, which uses a fairly standard ELO rating, I definitely feel like overall ranking is mostly up to chance. I’ve dropped multiple hundreds of ELO on a bad day, when the dice roll poorly or I get matched against counter-builds. And then gained it back in another day, when I top-deck the ultimate combo first round every time. It sucks.

        All the top players know that the ELO rating is dumb as anything more than a vague indicator of skill in such an environment, but it does a fairly good job of separating veteran from inexperienced players. Still, for me, it becomes just another thing to get angsty about when the rng is hatin’ and everything is going poorly.

  9. Fenix says:

    What I have found to be the biggest obstacle for me to take mmr seriously is how many people make it pointless by playing on smurf accounts. I wish it could be stopped.

  10. El_MUERkO says:

    The MMR system is broken, abused and terrible, there have been a million suggested fixes to the system but none are implemented. Maybe there’s some weird psychology from Valve going on here and they have some math to prove it helps the community but I honestly think it costs them players. I know several friends who gave up on DOTA because of it.

    • Oreos says:

      You can’t fix what isn’t broken. What are your issues with the system?

      • Kitsunin says:

        Probably that you can end up dumped at a wholly unsuitable mmr after your qualifiers, and from there you’re guaranteed a horrible experience for a long-ass time because the system isn’t nearly volatile enough after the qualifying games.

        • sonson says:

          This has been debunked though. If you’re really significantly better than your team to the point that you are out of place its possible to pull them with you in your wake, several players have proved it. If you’re not-you can’t. But that’s why you’re there in the first place.

          I’ve played a not unsubstantial number of games where even as support I’ve bee able to orchestrate a victory in spite of serial feeders and under farmed carries. But I’ve played far more games where I haven’t been able to do this. I’m obviously not terrible if I can carry a team on occasion as a support- but I’m also on a 51% win rate like most people, suggesting that I win as many times as I lose, regardless of how “good” I might think I am, I’m not so good that I can overturn the need for other players on most occasions, I need them as much as they need me. Really good players can consistently make up for a bad player or two. I can occasionally. There’s a big difference.

          • Kitsunin says:

            What has been debunked? Because yes, you can get a bad start after qualifiers, and I said little beyond that.

            I’m just saying that it isn’t volatile enough, it takes too many games to change your rating and that makes it very tedious.

          • jrodman says:

            Since the MMR range is maybe 4000-ish, (say 1500-5500) that means it could theoretically take an unbroken 160 win streak to go from the bottom to the top. That’s certainly longer than in say Starcraft 2, where rising to grandmaster from a scrub ranking can be done in around 50 games. Not to mention that Starcraft 2 matches are shorter.

            Still, that’s not a completely silly number of games, and it’s very unlikely that anyone ever is this far miscalibrated, so more likely numbers of wins for correction are values like 15 or 20. Of cousre at that level of disparity, there will inevitably be a number of losses mixed in, so we can increase that to some value like 30 or 40, but still not crazy.

            Yes, it can take more time than seems fun to correct a wayward MMR, in the case that, for example, you stopped playing rank for a long time and are coming back to it far exceeding your rating. But it’s not unworkable.

      • El_MUERkO says:

        They are too numerous to go into detail, if you want detail search ‘MMR’ on the Dota 2 Reddit. If you believe it’s not broken then why are some of the longest threads on the Reddit and Dev forums all regarding the MMR system? Why is Smurfing, Boosting, Account Selling such a problem?

        My experience is maybe one in ten games are between two reasonably balanced teams and some of the reasons for imbalance are easy to fix. I assume they’re not fixed as Valve want to keep the search timer as short as possible or are sadists. But what’s worse; waiting four minutes for a fun game or two minutes for a game that wasted thirty/forty minutes of your life and made you wish you could explode heads with your mind?

        • sonson says:

          “If you believe it’s not broken then why are some of the longest threads on the Reddit and Dev forums all regarding the MMR system?”

          Because people don’t like to be made to feel mediocre.

          There are numerous instances of pros taking low MMR accounts and jumping up the ladder in no time at all

          There are zero instances of 2-3 k players buying 4+ accounts and holding their own there for any length of time

          Most people who complain about their rankings don’t stack, they don’t block, they don’ understand creep equilibrium, they don’t smoke, they don’t ward pull camps, they don’t check oppositions items until late game, they forgot/mistake hot keys and to use consumables, they hoard gold and lose it when they die, they have sub 15 cs at 5 mins in lane, they don’t orb walk.

          All of these things can be practiced solo and learned and require no team interaction at all and would improve most people’s games considerably, and are essential to *every* role. But people basically just play as they always have done, they don’t work on things, and they assume they don’t rise because of a conspiracy or broken system rather than because they’re average players who will never improve because you don’t improve without mastering the basics.

        • jrodman says:

          It’s easily possible to explain all that behavior by the idea that people are far too emotionally invested in their number, and also have fun playing against people who are much weaker than them. Since those two things are demonstrated fact, we don’t really have to imagine some flaw in the MMR system to explain those behaviors.

  11. XX says:

    Solo queue games are a different game mode to teamplay, there are ways of playing that will be successful in solo that won’t have the same efficacy against coordinated opposition, and likewise being used to playing as a team against teams may mean you fail to properly exploit/account for the lack of coordination in solo.

    I don’t think its an entirely meaningless number for teamplay, it indicates a certain amount of mechanical skill, game knowledge and versatility, but obviously there is a lot going into your teams MMR that isn’t represented in your solo MMR. In fact, I’d argue that for most teams these unmeasurables are also immeasurably important, not least because teams that are incapable of not self destructing after every loss…well they aren’t teams then are they? Seems to me that a good way to avoid inter-team drama is to pay little heed to the differences in numbers – you don’t need a number to know if you enjoy playing with somebody else or not and that’s all that _should_ matter.

  12. namad says:

    I’m not sure the author knows what calibration matches are. How they work…. The suggestion of re-calibrating every year wouldn’t do anything for players who play large numbers of games… it might effect players who only play small numbers of games, but if anything it would make their experience worse not better.

  13. Moose says:

    Having seasons would be great, like most I calibrated fairly early on and have continued to play unranked to the point where I’ve probably improved a bit since then.

    As a mainly support player though I’d love season resets so that I could try and find my rank in other aspects of the game. My carry play isn’t great but I’d love to, one month, start from scratch and try and rank solely as a carry or mid or whatever, similar to how in hearthstone one month I might just play ranked with one particular deck or class and see how far I can take it.

    Obviously the current ranked setup is absolutely not somewhere you should be trying new things (which in itself is an issue because the whole thing stagnates and people only play what they are comfortable with) but that means there’s no real way to test your ability in different roles.

  14. popej says:

    Sigh, I can’t keep up with the cool kids anymore. This has nothing to do with measles, mumps and rubella!

  15. Sarracenae says:

    Generally i’m quite happy to just play unranked. However i do wish they would change the pick system for unranked all pick to be more like the ranked version. Picking and counterpicking in the last second is really annoying in unranked, particularly when you are trying to pick a hero that fits the team then half the team changes who they are hovering over at the last second and you end up in a team with no synergy.

    Clearly this is probably more of a problem at my scrub lvl than at high levels but still, i really see no reason for the last pick bums rush.

  16. trn says:

    ‘Thing is, the strategies are different but the mentality informing them is often pretty similar. By that I mean they tend to assume that you are a) being placed below your REAL skill level and b) that your teammates are liabilities.’

    I can’t speak for Dota 2, but in Heroes of the Storm, where MMR is kept hidden, this often holds true.

    While many people game online, gaming is still, mostly, a solitary hobby. In our homes, playing on our own, we are ‘the best’. We win games, wangle our way onto leaderboards, rake-in trophies and Steam achievements and generally have our onanistic tendencies validated. When we venture online, however, we do not win every time, we are no longer the champions of our gaming space. People say its the anonymity of online gaming that encourages e-peen measuring, blame culture and anti-social behaviour – but I think it is also a natural consequence of finding ourselves doing something badly in public we usually do very well in private. Some don’t handle that very well.

  17. PikaBot says:

    I’ve always found unranked games to be much more toxic than ranked ones. The people who play ranked – that is, who’ve played enough of the game to get access to ranked – have an investment in the game, in the community, and in winning that match. If they shoot off their mouths and get globally muted, it’s a real problem for them. Meanwhile, unranked games are full of people with no such investment, who have absolutely nothing stopping them from ripping into the other players.

  18. sonson says:

    I refuse to do the MMR challenge to find my thing. I like DOTA for it’s variety, team play and the never ending challenge. I’ve discovered since playing it that I like winning far more than I would like to admit, but still, the core things I like, I can get with friends and in solo que enough to not have to commit to anything ranked.

    It’s difficult in the sense that there are often players who do make your position unwinnable-but it’s also undeniable that if all you do in DOTA is commit to getting better regardless of your team and the result, every match, you will get better and consequently more likely to be able to compensate for the weaknesses of others, just like in any team situation. Someone who thinks they’re the best at DOTA will never be able to compete with someone lat the same level ooking to improve.

    In general it’s so well balanced -the vast majority of people have win rate of 50% – that it’s foolish to ever expect to win. This is a game where a team of 5 could beat another team of 5 43 -2 and then the exact same team of players could lost to the same team solely on the basis of draft. There are *so many* variables.

    Some of the best advice I ever read was don’t go in with the intention to win- just to play well. The result is outside your control, but your own performance isn’t, and if you play well you’ve done your bit to moving your team to victory, are more likely to be able to dicate games as far as one player can etc.

  19. BooleanBob says:

    For better or worse, MMR is a pretty good indicator of your ability to win pubs. This has been proven with various studies of high MMR players taking over low MMR accounts and low MMR players taking over high MMR counts; what follows is always a fairly sharp correction in the account’s MMR towards the level of the person now playing on it.

    Pip – of course competitive, organised tournament Dota is a very different beast to the chaos and cat-herding of solo ranked queue. The vagaries of being good at the former but worrying about being judged on the grounds of the latter, as you encountered with the bracketing thing, seems like an extreme edge case to lay at the foot of the MMR system when 99.9% of the player base are never going to have to deal with that situation. It sounds more like a problem with the organisers misappropriating that metric for a purpose for which it was never intended.

    I don’t think the move to seasons would be useful because it would only serve to lengthen the one journey that everyone has to take with MMR. There is indeed a road we all must travel with in ranked Dota: not to reach 5k, or 6k, or any other arbitrary number that we want or feel we deserve, but rather to come to terms with the number that we have.

    MMR works: if you want to test your self-assessed MMR versus the number on your profile, you can do so at any time. Just make a new Steam account, download the client and recalibrate. You’d be amazed (or maybe you wouldn’t) at how many people discover they were actually a 4k player all this time that Valve and the baddies and the servers had been holding them down in the 2k trench.

    Giving people the continual expectation of a reset and recalibration would have two effects. Firstly, it would lengthen that journey to the acceptance of the reality between one’s perceived and true MMR. Until these numbers are reconciled, the toxicity produced by the psychological disconnect between them perpetuates, bleeding into the attitude taken into the game and contributing to blaming, flaming, and all the other stuff which makes ranked games unpleasant.

    Secondly, it would be ruinous for the player population in ranked games. People who hadn’t come to terms with the MMR assigned to them would simply refuse to engage with the system, opting instead of playing and striving to improve, rather to wait it out for the next round of calibration and have another go at receiving their ‘true’ MMR.

    Or so I reckon anyway. Sorry if I came across as pompous or boreish in this post. If you disagree with me, console yourself with the knowledge that I’m just a 2.7k scrub and therefore have no idea what I’m talking about.

    • werbliben says:

      I see another possible problem with yearly resets. If everyone expects their MMR to be born anew in a matter of a few weeks/months, some players’ motivation to do their best in the ranked games in this threshold period might become lower than usual. They might try new heroes out without practicing them properly – or at all – and spread havoc in general, ‘cause, y’know, doomsday’s a-comin’, so we might at least have some fun! And the period in which such reckless attitude would be observed would vary depending on how frequently a certain player fires Dota up – the more occasional his visits to the three lanes, the longer the interval in which he/she feels that the reset is coming “soon”. It will be motivationally different from doing shit in unranked games that many perceive as less important ones, because it will provide a clearly defined time frame in which nothing matters in ranked – and even your hidden unranked MMR will be in danger.

      You might argue that such behaviour, if probable at all, would be mostly observed in lower MMR brackets (and as someone at 1.8-1.9k I – and around one third of the player base with equal or lower MMR – would be quite concerned with it). While this might be true, I have a friend whose MMR is floating between 4k and 5k and when he’s playing unranked, he’s in low-priority a large bulk of the time, because he feels being reported in unranked does not matter much. I believe that he would carry this attitude over to pre-reset weeks, should they indeed be introduced. He is rather exceptional among my dota-playing acquaintances in this regard, though, and I don’t come across too many people truly deserving a report even in my bracket, ranked game or not.

  20. soundofsatellites says:

    Well for me the problemas lies in that, so far in my experience, MMR leads to mostly unbalanced matches that are kinda stompy and winning or losing streaks. I sincerely don’t believe there is an ELO hell, but I do wonder if the ELO system is an accurate representation for the varied skill gap in video games, or the skill gap in dota.

    As for MMR experiments out there I think that while the do prove that it’s better to focus in playing better oneself instead of blaming others for losses, they have maybe methodological errors? I dunno, as far as I know there is only one of those done as kotol which is a kinda weak hero and a support. Most others trend towards carries and semy-carries that in the snowball-y gameplay of dota lessens the chance of throws.

    I dunno: tl;dr is just that I wish I had more often epic give-and-take games and less one sided stompy stomps of ez mid ez life ez rares

    • jrodman says:

      I think this is a game design criticism, not a matchmaking criticism. Dota is designed with snowballing possibilities, and so they often occur.

      There are some factors which exacerbate it of course, like the matchmaker preferring to start a game soon rather than wait for more perfect balance, and smurf accounts, but mostly I think this is game design.

      • soundofsatellites says:

        Point taken: The snowbally nature of dota, and of some of the heroes can be discussed at large as a game design criticism issue, but I should’ve put more emphasis as the how MMR calculates your impact in any given game.

        For a complex game like dota, it seems odd that everything is reduced to a single number. There are a number of objective things that might be quantifiable such as last hit statistics, stacking camps, gpm, xpm, wards bought, wards placed. But there are other than might be harder to quantify (and to reduce to a single number!): Are you a stronger support than carry? Do you understand hero matchups? How about game mechanics? Whats the MMR impact of diving tier 3 at lvl 6 and feeding yourself? Split pushing when no enemies are on map and dying to a gank.

  21. Chucky89 says:

    Exactly my thoughts and feelings about (solo) ranked matchmaking…

    I did my calibration when Ranked was implemented and I was relaiively new to the game, only went down ever after :-( I discussed this idea with a mate before and I think League of Legends does reset its ladder once in a while.

    I really hope they change the MMR in Dota 2 Reborn, though I don’t think it will ever be reset or drastically changed…