Let's Talk about Matchmaking


(Jythri) #1

This week, we’re going to roll out another round of changes in an effort to improve matchmaking. Our key goal, as always, is to try and help us find a way to play Battleborn that you enjoy. What makes this challenging is that we all enjoy playing in different ways. Today we are focusing on Versus Matchmaking.

The Challenges of Matchmaking

All of our players want a “good match,” but “good match” means different things to different players.

We can really sum up most of the requests in three statements:

  • Match as quickly as possible
  • Match as closely as possible to your skill
  • Give you some choice in what you play

At times, those priorities are at odds within the matchmaking system.

For example, we can make you wait until you have a very precise skill match, but that might take a very long time as we wait for just the right players to be online and available to match. We could also give you the options of queuing up for exactly the map you want to play; however, for each available option, we split our playerbase into smaller and smaller sub-groups which lowers the pool of potential match candidates for all players and increases the match time.

On top of all this, our system attempts to find the best possible connection speed for you.

TLDR – Better skill matches require longer waits. More choices divide the community, creating separate matchmaking pools, which results in (you guessed it) longer waits.

Where We’ve Been

Up to this point, all competitive queues have used ELO-computed skill-based matchmaking. That means, when you play a match of Battleborn, we use a skill rating number to track how you compare against other players. If you win, your rating goes up. If you lose, your rating goes down.

We use that number to find teammates and opponents of similar skill within a certain tolerance range. If you want to play a match against players of very similar skill, we need to use as narrow of a tolerance rating as possible. If you are very high (or very low) in ratings, however, this can create a problem: too few players within your skill range to create a full match. This can sometimes result in a condition where it takes a very long time to find a match, if indeed you ever find a match.

Broadening the ranges creates an opposite problem: you’ll have more viable candidates to match with (aka, you’ll match faster) but those candidates might be of much higher or much lower skill than you (aka, you could have a match against very differently skilled players).

So far, most of our weekly changes to matchmaking have been to change our allowable tolerance ranges, and to adjust how fast and how far ratings can change within our scale. We’ve tried to respect your choice of play mode by giving you the option to match into a specific mode type, and each mode has used a rating number specific to that mode queue.

We’ve tried to find the sweet spot between how long it takes you to match and how good of a match we make for you. We’ve seen some success, but not as much as we would like. So, we’re going to try some different approaches.

The Great Command Rank Fallacy

A common concern in matchmaking is Command Rank (the number next to your name on the front-end menu). Command Rank increases as you play but it does not factor into your behind-the-scenes skill rating. While Command Rank CAN help indicate how much time someone has spent with Battleborn, it’s not always an accurate indicator of how skilled they are.

If a new player is a good gamer, and goes on a multi-game winning streak, they could be in the top ranks of skilled players by Command Rank 5 or 6. Likewise, a Rank 100 player has a lot of hours in Battleborn, but if that player has lost as often as they have won, they are matching right in the middle of our curve with other average players.

Hiding Command Rank may reduce some confusion. However, we believe Command Ranks are valuable to mark and display your progress, so we’ve avoided this change. We’ve also considered displaying your skill rating, and may eventually.

Where We’re Going Soon

This week, we’re going to adjust the flow of competitive matches to try a different approach at handling some of our challenges.

Here’s how we’re changing Versus Public:

  • The three existing mode-based queues are going away
  • Two new queues will be available: CASUAL PLAY and COMPETITIVE PLAY
  • Intermittently, a third option, SPOTLIGHT BATTLE, will appear featuring a single specific map or mode.

The CASUAL PLAY queue will contain all maps (and modes) available in Battleborn. Each time you match, you will vote on one map from the three that were randomly selected. CASUAL PLAY will not use any skill-based matchmaking, so you will match with all available players. CASUAL PLAY should cater to those players who care less about what you play specifically and more about wanting to get in and play quickly.

The COMPETITIVE PLAY queue will contain exactly three maps, one from each of our three modes. Generally, these will be the most popular map for each mode. When you match in, you will vote on the map you want to play from the three selections. COMPETITIVE PLAY will use skill-based matchmaking. It might take a little longer to find a match, but the match should be closer to your skill level. To help make up for the additional time you have to wait, we’re going to give you bonus experience and credits for playing matches in the COMPETITIVE PLAY queue.

SPOTLIGHT BATTLES will feature a single map in its own queue, without any voting. It will use the same skill-based matchmaking ramp as used by the COMPETITIVE PLAY queue, and like the COMPETITIVE PLAY queue, SPOTLIGHT BATTLES will give you bonus experience and credits. We will change out this map on a regular basis for those of you who really love to focus on one particular map.

TLDR – CASUAL PLAY matches fast but randomly; COMPETITIVE PLAY will take longer to match but be closer in skill; SPOTLIGHT BATTLE matches by skill on one specific map.

The Future

We’re trying some things this week. We need your feedback! Try out the new changes to Versus Public match flow. Let us know if you like it. Your feedback will help determine how we change in the future.

Send us your feedback here in the Gearbox forums!

Even with these changes, though, we aren’t done. We are working on ways to make our skill-based tolerance ranges smarter so that they will do a better job of balancing speed of matching with the quality of your matching. Those changes will appear in our next big patch. We’re also working on changes to how our queues work, trying to give you the choices you want, while keeping the time it takes you to match as short as we can.

We’re making progress, but are by no means done. Try out our changes, keep providing feedback, and work with us to continue to improve Battleborn!


Population issues?
Explain how ELO would work
Matchmaking Suggestions (Fixes etc)
Why did you lax on the ELO so much
Matchmaking at all time worst! Losing hope for the game
Matchmaking Why
Matchmaking freedom
Problem with competitive and causal
Please let me know matchmaking will be fixed
Spotlight Battle and the Other Queues
Let's talk dev feedback and future content UPDATED: 7/1/16
Let's talk dev feedback and future content UPDATED: 7/1/16
I don't want to play f****ng Incursion!
Matchmaking thoughts
Matchmaking Change is hurting the game
Versus change... my last 6 games
Great game, bad choices
Quitting at character select
Battleborn has been ruined!
I am forced to play modes and maps that I hate, what a bad decision!
Feedback on matchmaking changes
Let's Talk About Lack of Choice!
Team Matchmaking Question
Battleborn PvP matchmaking changes!
There NEEDS to be consequences for leavers
(Kujako) #2

How are you going to manage skill based match making with only ~700 players?


(MLGx420YoloSwag) #3

ranked play yesssssssss. Now if only could ban legendaries and do pick/bans in competitive playlist.


(Natsume Ryu) #4

excited for this, hope i can get on before changes happen. cant wait to get my internet back :x


(pizzonage) #5

Its not quite ranked play yet but it’s definitely a step in the right direction.


(Iiits Kyle) #6

That is on steam, consoles are fine.


(Mshldm1234) #7

Personally I think the skill level should definitely be shown in tiers with rank play. It’s silly to have rank play and only know your rank based off of calculating your W/L.


(Ambra's Arbiter) #8

THANK YOU. THANK YOU THANK YOU THANK YOU. YOU DID EXACTLY WHAT WE HAVE REQUESTED. ALL WE NEED NOW IS AN OFFICIAL RANKED.a


#9

Just out of curiosity, how does the current matchmaking handle groups? Will the new one be the same?


(battleaxe0) #10

Possible, but very unlikely. Also, at CR 5 or 6, their gear will be subpar to non-existent if they’ve been playing only PVP for those ranks. I’m not sure how the skill rank is calculated, but low level players I’ve seen always play the wrong game - they invariably go after 1v1 kills instead of objectives, get creamed, and feed the other team.

One option to try is shuffling. Take 2 players of very close skill and shuffle them onto 2 teams, keep doing that until you have a team of varying skills, but matched 1-to-1 against the other team’s skills.

For example (made up skill numbers 1-100):
Team 1 Roster skills: 99, 74, 33, 22, 11
Team 2 Roster skills: 95, 72, 38, 20, 12


(badguy5 on YouTube) #11

@battleaxe0 mentioned something very similar to what I was about to say.

While I understand that CR and MMR are different, there is certainly a correlation. So, when it’s me, a CR 12,25,11,30 against CR 100,100,100,90,87, it seems like there’s two things happening:

  1. the system is making teams of comparable skill to each other (trying to pair similar MMRs together)
  2. the system is comparing the skill of fully constructed teams, eventually loosening the reigns from there

This results in having two teams, who are in theory internally balanced facing each other, which may not be a fair team-to-team pairing.

What would make more sense is to find pairs of people to play against each other first, rather than with each other. Players JON and DAV have an average MMR of 50, pit them against SAM and KYL with average MMR of 49, and continue to do that until you have 10 people total.

This is based solely off of my observation, but I could be wrong.

Also, obligatory: HARSHER PUNISHMENTS FOR LEAVERS!!!
Really, though, stop putting them on my team.


(Corrison) #12

Any numbers or % for what you consider [quote=“Jythri, post:1, topic:1539170”]
bonus experience and credits.
[/quote]


(jmsitzman) #13

Mega, mega thanks Jythri. Keep up the good work.


(battleaxe0) #14

When losing Meltdown 300-0, I offer surrender, the lowbies say “But I’ve got 2 kills! We can’t quit now!” and vote down the surrender, and it’s the 3rd game like it in a row - I’m sorry - I leave to go find a scotch and something else to do.


(badguy5 on YouTube) #15

Sure, but that problem should be fixed with better matchmaking.


(SuperBadJuJu) #16

This is fantastic! Looking forward to tackling the competitive play with some friends! Will the system still evaluate “Skill” the same way? Or will we expect to see tweaks there too? How is “Skill” defined? Curious how things like Kills, Assists, Buildable etc are weighted.

Thanks for putting the work in and tweaking the matchmaking, can’t wait to dive into it!


(12326kjr) #17

@Jythri

This is exactly what i posted a little while back.

I expect to paid for this, handsomely i might add, as you are using my idea in your game. I will accept payment in the form of Doritos or Mtn Dew, which ever you prefer. I also expect my name to be added to all credits (must be the very first name as my idea will have had the largest impact on the game).

But seriously, this is a great idea and I may start playing again even though i just bought overwatch and they just released ranked play as well.


(Rickbaker88) #18

Am I missing something here??? No one in comments cares about the fact that we are going to be forced to play certain game modes?? My family and I dont give a flying F*** about meltdown or capture.

I look forward to having us 3 team members auto quit every time its meltdown or capture that gets voted on.

Seriously, what the hell?


(Sam50gregg) #19

a little off topic, but can you add a FOV slider for consoles?


(12326kjr) #20

I fyou have 3 team members you control 3/10 of the votes. For most games you are going need 1-2 more votes to win. Not that hard. If you end up playing something else every once in awhile its not a big deal. The games modes arent all that different.