The definition of an unbalancedz match


(Meetrock) #21

@Meerkats There are a lot of factors to take into consideration when define who’s more skilled. Go to your profile in DB and you’ll find a bunch of stats the devs can use to calculate how skillful you are (there are a looot of information the game gathers we can’t even see in those stats). While ranked has matchmaking, casual matches have a different type of balance, which is the lazy one. Of course, you won’t spend as much time and resources on a casual match with kids as on a real ranked match where pro players can compete. SD simply doesn’t care about fixing the balancing problem, I am 101% sure they have all the resources to fix it, whether they are devs or data they can use, they simply just don’t care. And just so you know, “matchmaking” is just a word describing BALANCE.


(mrdisco) #22

There’s no miracle that can fix balance when a game have a small player base.

Just get stomped and try to get better like everyone else. If the teammates suck, start instructing them “vigorously”.


(Meerkats) #23

Yeah, and I literally just told you…

NONE of that information is collated into anything useful.

If I join a match that is 5 v 6, but the average MMR for the team with five players is 800 and the average MMR for the team with six players is 400 and my MMR is 2,700, which team should I be put on? Cause lemme tell you want DB will do: DB will put me on the team with five players.

NO MATCHMAKING.

Kinda like how games used to be.


(Your worst knifemare.) #24

Those are nothing compared to 90% of my faceit matches.

You also have to take into account leavers, medics, spotting xp, etc when it comes to comparing scores. Also generally the better one person does on their team the better the teams score becomes overall.


(HadronZodiac) #25

Wish i still had the screenshot of a match, our team has an average of 6k exp, and enemy team had ~ 18k


(STARRYSOCK) #26

To be fair unbalanced matches like this are common in every game, not just DB. And considering DB literally has no matchmaking, it’s actually not that worse off in terms of balance from some other PVP games I’ve played. Hell, even CMM had a pretty large amount of unbalanced matches.

Ultimately there’s no point in complaining about it. DB’s playerbase is too small for CMM to be worth the additional queue time. Best thing that could be realistically done is encourage vote shuffles for unbalanced matches and try to break the stigma around losing so people aren’t afraid to switch to balance a team.


(Teflon Love) #27

DB’s matchmaking essentially came down to assigning a number to each player that intends to represent their skill, then compute the sum of this number for the entire team and then find two team with a number that is about equal with adjustments for premades. With that you end up with matches where you win about half and enjoy hardly any because the result is random and most of the matches are unbalanced.

What would have been needed is a game mode that enforces a sensible meta instead of allowing 3 snipers no medic, enforces a single merc rule, allows you to announce your preferred role beforehand (to include this information in the match making). It could then maintain a rating for you for each role (e.g. decent medic, mediocre engi, awful assault). Taking this information into account, the matchmaking could build teams where all roles are covered.

Of course there still should be some servers without match making to mess around for fun. But then you would not mix up with the people who play semi seriously and want to enjoy the flow of proper teamwork.


(STARRYSOCK) #28

Ideally that’s what ranked should be, a more serious environment. Nothing wrong with 3 vassilis in a casual match, since those are supposed to be, well, casual.

Besides, poor team comp is really only a small part of it. In DB mercs are very versatile, a medic can push like an assault and an assault can do the objective. This means that bad team comps impact balance a lot less than they would other games. From my experience, whether a match is balanced or not mostly boils down to pure skill. A level 500 player will usually dominate no matter what merc they play, and a level 5 player will usually suck no matter who they play as well. That’s why I figure it’s better to encourage shuffles or switches to balance a match, rather than to try to force people to play one way or another in a casual match


(Kirays) #29

No, you’re wrong. If your MMR were to reset each time you join a new server the way players are distributed between the teams would truely be random since the balancing occurs right before a match is about to happen. The devs might have adjusted this already since then, but from what I know the last 30 matches have an impact on your rating. Now we don’t know about all parameters they use, but levels have been confirmed to be a part of it, and the amount of points you gather throughout a match/your location on the scoreboard most likely has an influence on your rating, too.


(Meetrock) #30

@Kirays That’s what I’ve been telling them all along.


(Teflon Love) #31

And it still doens’t take into account that I might feel like playing as a somewhat decent medic or a utterly awful sniper (for the lols’n’trolls). A still enter the server with the same MMR.

And it still doesn’t take into account that the team already has two decent medics and now needs a competent assault instead of a third medic.


(kittz0r) #32

starting on Defenders = attackers leave, I switch to Attacker and just roll through the Enemy and yet People complain about Balance because they aren’t able to Defend against 1 Person and Attackers without an Engineer?

No there is no Balance Problem, there is a trash Players problem. Thats the usual Pub experience and nobody in the right Mind would want to play on Pubs.

Like this Guy on Defenders: i knew it was unbalance!
uhm what? its unbalance because Players aren’t able to Kill anything or create a good Team with their possibilities.


(Meetrock) #33

When the entire team is just bad, including yourself, there is no one to blame but yourself for not playing good. But when your entire team is just stupid and you’re the only one actually doing something, then surely you’re not the one to blame.


(Smooth) #34

So there are a few different points where the game will do it’s best to balance the teams.

  1. When a player is Quick Joining in to a server

At this point we typically only factor in the average level of the server, and try to get players into servers that are close to their level of experience.

  1. At the end of the Lobby, just before loading the map

Here is where the game will use ‘Public Skill Rating’ (PSR) of players. Which takes into account thier level and their recent (typically something like last 25 matches) performance in recent matches, with around half coming from level and half coming from performance.

Recent Performance factors in pretty much everything, kills, accuracy, xp, wins etc. Player level stops adding anything extra to the PSR after Level 50, which is where player skill plateaus. From Level 50 to Level 1,000+ the impact Level has on PSR is the same.

  1. Just before match start if teams are imbalanced

This will only take place if players have left/joined since the Lobby balance and uses the same rules as the Lobby.

  1. When a shuffle vote passes

This uses PSR but also factors in the current match performance. The longer the match has been going, the more heavily ‘current match performance’ is weighted.

  1. When a player connects

If teams are uneven, the connecting player will always join the smaller team. If teams are even then the connecting player will join the team with the lowest overall PSR.


The main struggle we have with balancing teams is usually the vast range of skill from the best to the worst player. Given the high skill ceiling of Dirty Bomb, it’s possible for one expert player to completely dominate another team.

The second challenge we face is team composition. The data shows that lopsided teams (too many of the same Merc, lack of diversity across the roles) have a dramatically lower winning rate than those which have a more even spread of Mercs. Since team composition can change on the fly, it’s not something we can really factor into any balancing algorithm.

The final struggle we have is that player skill varies dramatically from Merc to Merc. Someone who kicks ass with Fragger might decide to try Sparks for the first time and not do so well. Again, as switching Mercs is something that can occur at any point during the match, it’s not something we can really factor into any balancing algorithm.

Overall match balance (the difference between two teams at the end of the match) is something we measure on a daily basis. We’ve made changes to the algorithm that has improved things over the years, but nothing has had as big of an impact as player base size. The more people we have playing, the far easier it is to get players of closer skill in the same match, which makes balancing the teams so much easier.

Now that the server issues have been resolved, we’re seeing a slow recovery in player numbers, which will gradually improve things here.


(Jigstraw) #35

I’m inclined to disagree about player skill plateauing at level 50. I’m in the 400s, and have gotten significantly better than I was at whatever level was equivalent to 50 back before the level scaling was changed. Account levels don’t correlate directly to skill levels, but they do tend to indicate the amount of time that has been played. Which often means the player has learned more about the game.

Since i was level 20-25ish which i’m assuming would be around what 50 is now, i’ve learned a LOT more about the map layouts, how to predict other players movement, tons of shortcuts and trickjumps both intended and unintended, how to choose the correct loadout to match not only the merc’s playstyle but my playstyle as the merc, when to switch mercs to fill a missing role, when to stick with a merc i’m strong with rather than filling a missing role poorly, just tons of things that have made me a better player than I was back then. I might not have improved as quickly past the supposed “plateau” point, but I definitely still improved, and I’m sure other players do too. I think bumping the level “cap” in the balance algorithm up a bit higher would be a good idea to account for certain things that won’t necessarily show up in a page of stats. I think people who use voice chat in-game should also have a slight modifier applied, as communication can have a MASSIVE impact on the outcome of a game, especially if only one of the teams are communicating. this would be flawed of course as some people communicate in third party voice chats, but it would be nice if the in-game voice chat was considered as well.


(Smooth) #36

Well that’s exactly why player level only accounts for around 50% of the total rating.

Generally though, beyond Level 20 (under the old levelling system - so around Level 50 now) things did plateau. Yes skill still very gradually increased over time, but that gradual increase is covered by the other 50% of the total PSRating.

Directly factoring in player communication is a good idea but not something that’s guaranteed, especially since people do use out of game chat and also because using voice chat in previous games doesn’t mean you’ll use it in the current match.

Additionally, successful use of voice-comms already somewhat factored in, as if players are doing better because of it, their recent performance will improve, which will impact their PSRating.


(Jigstraw) #37

Well. I guess all I have left to ask is if CMM will be brought back if the playerbase continues its upswing.

It addressed one of the major problems with the quickjoin/server browser in terms of balance, in spite of CMM having its own flaws. Often times, several players will leave after a particularly bad match. Then the next match begins with a half-empty server, which usually becomes more and more imbalanced as those empty slots are filled, due to the player that joins in when there are an odd number of players in the server not being subjected to the PSR check to decide which team they should be placed in. This wasn’t as much of a problem with CMM, as the matches always started with full teams. But once this cycle begins in the quickjoin/server browser system, it often becomes a loop of people leaving because the match wasn’t balanced, which starts the cycle all over again. I think I preferred CMM’s flaws over the flaws of the quickjoin/server browser system.

Another thing to consider is that there are a fair few negative reviews on steam with the reviewer saying they’d enjoy the game more if there were a party system that allowed them to play with their friends. Bringing CMM back could draw those players back to the game and help increase the playercount, which can only help solve CMM’s balancing issues. I personally hated the party system, as getting matched against full premade teams as a solo player was a nightmare, but I think it’s an evil i can live with if it means a larger playerbase. The only change I would suggest for CMM is adding a “rematch” button, for when you find a set of teams that are particularly well balanced against one another and they enjoyed the match. This would also solve the long queue time issue, as people could just keep hitting the rematch button once they found another team to play against.


(Teflon Love) #38

Thanks for theses insights, much appreciated.

One thing you should be able to measure comparably easily is how willing a player is to fill gaps in team composition. I’ve joined plenty of games where a team had no medic or engineer (on EV escort) but 2 or more snipers.

For instance, if you play on a team without medic and everyone keeps playing a different role, all the numbers counting for your performance are reduced during this time for the entire team. If however you switch to medic in a team without one, your individual performance numbers are boosted for the next spawn cycle, and the rest of the team reverts to the normal performance count once your your boosted period is over.


(Teflon Love) #39

I agreed that you cannot do this without information about the player’s intention. But why not let the player state his intent before the game?

Add a checkbox to the merc selection screen “I intend to play only mercs I’m experienced with”. Once selected, the player can only put mercs in his squad that he has shown to having mastered to some basic extent. This checkbox will only be available once you mastered at least 3 mercs, so new players won’t be confused by it.

If you play mercs you mastered without checking the box, your ingame credits and XP is reduced during this time. So if you realize that you inexperience drags down the team too much, you can switch to your old dogs. But it’s not encouraged.

In order to decide whether someone has “enough” experience with a certain merc you could go the easy route (having actually played them for e.g. 3 hours) or add a system where performing certain actions indicates that someone knows what he’s doing (e.g. doing charged revives as medic, cause a sum damage of N to the EV, blow up the door generator on trainyard with artillery or laser, …) .

Make these actions transparent, bind them to achievements and maybe even some minor credit rewards and players might actually strive to go there.

This of course will not be able to separate the mediocre players from the really skilled ones, but in my experience they can coexist on the same server quite well.

It’s the bottom skilled players that ruin matches, and currently they have hardly any feedback what they are supposed to do to get better.


(Begin2018) #40

@Smooth : Then why there isn’t 2 kinds of servers? Max level 75 and Min level 75?