QUESTION TO DEVS - Determining Rank

(bgyoshi) #1

@stayfreshshoe @Moobabes @6runk etc

As far as I know your placement in Ranked is determined on

  • Win/loss ratio
  • Individual game won/lost
  • Previous season rank
  • Individual match performance

So what are the actual metric considerations around this? Is there a specific ratio of these things that go into determining a rank move? For example, something like this?

  • Win/loss ratio (20%)
  • Individual game won/lost (60%)
  • Previous season rank (15%)
  • Individual match performance (5%)

And what does “individual match performance” mean? Is that some combination of values on the end game screen like accuracy, headshot %, revives, etc? Or is it more like your exp per minute?

There have been lots of calls to change how a person’s rank is chosen, shifting from a sheer reliance on team performance and basing it more on individual performance. In a team game, it’s important that team performance should have an overall effect on your Rank (otherwise what will encourage team cooperation?) but it seems to me like that should be the minority stake in your rank. So for example, taking the numbers from above and making it more like this:

  • Win/loss ratio (5%)
  • Individual game won/lost (20%)
  • Previous season rank (15%)
  • Individual match performance (60%)

Justifications for this are within the spoiler

! In effect, it should be harder and slower to climb the ranks when you’re constantly on losing teams, but not impossible. If you’re being that one dude that refuses to cooperate and play your role, the team will lose, and you shouldn’t be rewarded for it. But at the same time, if you aren’t cooperating, crushing the other team with your skill, and the loss is more because your team is bad, you should get a net gain for it, even if it’s small.
!
! At the same time, games lost to factors we can’t control (like trolls, throwers, leavers) have an impact on our Rank, but if we perform well, less of an impact.
!
! If the game is matchmaking well, you claim, then players should have a W/L ratio of 50% (ish). So that shouldn’t factor in very heavily, it’s outside of player control. Individual games won and lost are somewhat outside of player control, and entirely reliant on team performance. It should have a notable effect, but not a huge one. Previous season rank (would be, with this suggestion) mostly the player’s control, so it should have a notable effect, but since it’s based on many factors and also in the past, not as great of an effect as individual game performance. Player performance, on the other hand, is 100% the player’s control, and should be a very strong factor in determining Rank.
!
! Even if this means Rank doesn’t reflect a person’s ability to cooperate on a team, it WILL filter stronger players towards the top and weaker players towards the bottom. Even if the top rankings are just a ton of non-cooperative players that are very skilled, they will eventually hit a cap when they start facing players that are both skilled AND cooperative, as these games will affect said player’s individual performance stat.

But my question is, what kind of development challenges are there in regards to overhauling the system that determines player Rank? I have a feeling it’s a daunting task, considering the general push back on making this change. But I do feel like the community as a whole would benefit from it, and players would be a lot more encouraged to play Ranked. It’s a lot easier to deal with leavers, throwers, and trolls when you know those people won’t cause you to plummet down the ladder.

Your input would be greatly appreciated. Thanks!

(K1X455) #2

over 80% of my matches (including some wins) involves a match thrower and a leaver.

The validity of ranked matches is totally questionable, which begs the question as to how it is moving the competitive scene forward.

(Nail) #3

while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

(K1X455) #4

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

telemetry data, player reports and their intuition…

(Nail) #5

you think that would make random stuff predictable

sorry, doesn`t compute

(bgyoshi) #6

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

What?

re-reads post… can’t find a single mention of this

What are you talking about dude? lol

(Ptiloui) #7

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

The goal is not to predict random oddities, but instead weigh each parameter involved in rank calculation in a way those unexpected cases do not affect your own rank.

On the other hand, your individual match performance should register the role played. For example, playing medic and ending the match with a KD ratio of 5 (random number) but 0 health given should result in a bad performance result. It’s a cooperative game after all.

(watsyurdeal) #8

@Ptiloui said:

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

The goal is not to predict random oddities, but instead weigh each parameter involved in rank calculation in a way those unexpected cases do not affect your own rank.

On the other hand, your individual match performance should register the role played. For example, playing medic and ending the match with a KD ratio of 5 (random number) but 0 health given should result in a bad performance result. It’s a cooperative game after all.

And how would it do that?

Just based on the merc played it calculates things differently??? Because I feel like Guardian and Phoneix would really mess with those numbers.

(Ptiloui) #9

@watsyurdeal said:

@Ptiloui said:

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

The goal is not to predict random oddities, but instead weigh each parameter involved in rank calculation in a way those unexpected cases do not affect your own rank.

On the other hand, your individual match performance should register the role played. For example, playing medic and ending the match with a KD ratio of 5 (random number) but 0 health given should result in a bad performance result. It’s a cooperative game after all.

And how would it do that?

Just based on the merc played it calculates things differently??? Because I feel like Guardian and Phoneix would really mess with those numbers.

AFAIK, pretty much everything is already calculated with all the different xp we earn. All the game has to do is, as @bgyoshi suggested, weigh things differently.

If your individual performance is basically the xp you gain during the match, a medic will see his support xp weigh for 80% of his perf. On the contrary, an assault will see his combat and ability xp weigh more than support or objective xp. And for cross-class merc such as Guardian or even fire support merc, their perf would weigh 50% support / 50% combat.

This kind of calculation will reward more players that tend to play with others, even if they’re average, than those who are highly skilled but play solo.

(bgyoshi) #10

@Ptiloui said:

AFAIK, pretty much everything is already calculated with all the different xp we earn. All the game has to do is, as @bgyoshi suggested, weigh things differently.

Though I don’t actually know if the game just assigns weights to parameters or if there’s something more hardcoded and difficult to unwind involved, hence my question. It seems like it SHOULD just be a system of weights that can be changed… but if it was that easy, I feel that the devs would be tweaking it all this time. So I want to know more specifics about it, and what challenges are associated with changing it.

(woodchip) #11

@Ptiloui said:

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

The goal is not to predict random oddities, but instead weigh each parameter involved in rank calculation in a way those unexpected cases do not affect your own rank.

On the other hand, your individual match performance should register the role played. For example, playing medic and ending the match with a KD ratio of 5 (random number) but 0 health given should result in a bad performance result. It’s a cooperative game after all.

The most important reason not to do this is that the Medic with a KD of 5 IS actually a much mechanically superior player, albeit one who for whatever reason didn’t score well at his role. Placing him radically lower than his KD and win loss record suggests only leads to him stomping even harder when he eventually switches to an assault or whatever.

Simpler metrics like ‘total score’ or if you really want ‘total score adjusted for role’, maybe just K/D would probably also do a better job placing people in roughly the right skill buckets.

(Ptiloui) #12

@woodchip said:

@Ptiloui said:

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

The goal is not to predict random oddities, but instead weigh each parameter involved in rank calculation in a way those unexpected cases do not affect your own rank.

On the other hand, your individual match performance should register the role played. For example, playing medic and ending the match with a KD ratio of 5 (random number) but 0 health given should result in a bad performance result. It’s a cooperative game after all.

The most important reason not to do this is that the Medic with a KD of 5 IS actually a much mechanically superior player, albeit one who for whatever reason didn’t score well at his role. Placing him radically lower than his KD and win loss record suggests only leads to him stomping even harder when he eventually switches to an assault or whatever.

I agree this is paradoxical but this is the goal of the addition I suggest : that player may be a great skilled player, but he will be more efficient as an assault merc. If he only focus on killing people, he’ll have to switch to a merc designed to complete this task. That way, this player may show his true talent, hence the purpose of playing ranked. Otherwise he’ll just be a bad medic.

(K1X455) #13

@Nail

You really don’t understand how it works do you?

  • telemetry data - shows your actual behaviour in match
  • player reports - will re-enforce what the telemetry data shows
  • their intuition - it’s the smoke and mirrors I don’t know about, but I trust they have sufficient knowledge unless Cam is right about them being chimps
(Nail) #14

@K1X455
“over 80% of my matches (including some wins) involves a match thrower and a leaver.”

I was asking how telemetry and player reports could predict either

(K1X455) #15

@Nail, it’s not supposed to predict. The whole system is reactive and not pro-active.

(Nail) #16

@K1X455 said:

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

telemetry data, player reports and their intuition…

so this is just bullcrap ?

(K1X455) #17

@K1X455 said:

@Nail said:
while I agree that the rank system is flawed, I’m uncertain how it could predict throwers or leavers, please, for the love of Dog, enlighten me

telemetry data, player reports and their intuition…

so this is just bullcrap ?

No. You’re just hard of learning.

(Xenithos) #18

@K1X455 said:
@Nail
You really don’t understand how it works do you?

  • telemetry data - shows your actual behaviour in match
  • player reports - will re-enforce what the telemetry data shows
  • their intuition - it’s the smoke and mirrors I don’t know about, but I trust they have sufficient knowledge unless Cam is right about them being chimps

So to put it a different way, are you saying the system should use telemetry data and player reports combined with developer intuition (aka developer added programming to help the system determine identities) on what is or isn’t a good match, and that when it finds “bad” matches with leavers and afkers etc it doesn’t weight the outcomes as heavily on either team unless the 4 man group beat the 5 man group etc?

(K1X455) #19

@Xenithos said:

it doesn’t weight the outcomes as heavily on either team unless the 4 man group beat the 5 man group etc?

You do understand that Exceptions exists do you?

I’ve been in that situation. We won short handed 4v5; it wasn’t a fluke either. It was just we had a player who made good calls and timed it correctly on when to go all out and keep them long spawned in more than a dozen waves which chewed up their clock significantly.

(Xenithos) #20

@K1X455 said:

@Xenithos said:

it doesn’t weight the outcomes as heavily on either team unless the 4 man group beat the 5 man group etc?

You do understand that Exceptions exists do you?

I’ve been in that situation. We won short handed 4v5; it wasn’t a fluke either. It was just we had a player who made good calls and timed it correctly on when to go all out and keep them long spawned in more than a dozen waves which chewed up their clock significantly.

Yeah, I’m aware of that. That’s what I meant by saying that. That would be an “exception” or “bad game” and there are exceptions to the exceptions as well. My only question was were you saying the system could either choose to balance the game as equal to a bad game, or as equal to nothing and remove it from having any affect whatsoever? Because I think, In the case of exceptions particularly, that it should be able to see those exceptions and weigh even MORE heavily because of it.