Discussion on PVP MMR

entrailbucket

@pepitedechocolat said:
MMR system could be replaced with one that ties MMR to results rather than roster . (It could be much fun as high pointers would fight other high pointers, and double 550 roster could drop their MMR to abuse tiny roster that would be so cool)

I don't recognize your IGN so I'm not sure if you were around back then and this is a joke you're making, but for everyone else's benefit...it really did work this way!

In the early days, you had an actual internal matchmaking rating that was your overall PvP win-loss record. It carried over between events, as well. Since losing on offense is MPQ is pretty rare, and always was, grinders were very quickly only matched against other grinders.

Well, it turns out that when you have 0 points at the start of an event and you lose a match, you lose 0 points. So we'd start an event and retreat 100 or 200 times (for a net -0 points) until our W/L record went way negative, then get matched up with day 1 1* teams.

The devs picked up on that relatively quickly, and now matchmaking is roster-based.

dianetics

The devs figured out pretty quickly they did not want grinders to dominate the game. I did it for a few events...my wife hated it, the devs hated it, and I generally skated by because my wife didn't divorce me.

entrailbucket

@DAZ0273 said:
They could make it a one time deal. I am not so keen on the idea myself> @entrailbucket said:

@dianetics said:

@entrailbucket said:
So either: the gaps in power level are intentional, and the algorithm needs to know about that, or they're unintentional and they need to get fixed.

I'm 100% on the same page and I said the same thing regarding Okoye vs Wasp.

It would be an interesting thing to see them try to ilevel each tier, but I think it would be fruitless. So if Okoye is gold ilvl Wasp is white ilvl. I'm not a coder, but I'm not sure how you could function multiple tiers with a variety of power levels this way. It is a complex problem.

ie... a tier has a base power level but there is a modifier for the each hero's assumed power.
It becomes very convoluted, unless you shrink the power band of each tier. It sounds like a lot more work to try to normalize each tier. If you bring everyone closer together you could more easily modify, but you also reduce diversity.
It is almost an impossible task,, like how do you balance Beta Ray Bill v Rescue or Kitty v Knull?

Right...they should try to create some sort of obvious, easily-understandable indicator for each character to show how powerful they're meant to be...like some sort of star level? Maybe the best characters should get 5 stars and the worst should get 1 star!

(That bit of snark wasn't directed at you, as much as at the many, many people who've earnestly proposed the same sort of idea before -- you seem to understand what a silly concept it would be).

Controlling for power level within a tier is really hard/probably impossible. It also brings up other questions. Like, if Chasm is DESIGNED to be better than Rescue, why does PvP award the same number of Rescue shards as Chasm shards? Shouldn't we get twice as many Rescue shards, since she's purposely designed to be half as good?

All of this stuff is part of the (frankly monumental) pile of evidence indicating that character tiers are meant to be roughly balanced. When they produce a character who's too good or too bad, that is an accident, and not part of some grand scheme.

The answer given in the Q&A (point no. 27) regarding Chasm is at odds with your last sentence though?

Let me clarify a bit. Basically every underlying system or mechanic in the game assumes that characters at a tier are meant to be roughly balanced.

If BCS is doing things differently and are purposely creating good/bad characters, that means that some of these mechanics will need to be updated, because they don't work correctly if that assumption no longer holds.

Bowgentle

@entrailbucket said:

@pepitedechocolat said:
MMR system could be replaced with one that ties MMR to results rather than roster . (It could be much fun as high pointers would fight other high pointers, and double 550 roster could drop their MMR to abuse tiny roster that would be so cool)

I don't recognize your IGN so I'm not sure if you were around back then and this is a joke you're making, but for everyone else's benefit...it really did work this way!

In the early days, you had an actual internal matchmaking rating that was your overall PvP win-loss record. It carried over between events, as well. Since losing on offense is MPQ is pretty rare, and always was, grinders were very quickly only matched against other grinders.

Well, it turns out that when you have 0 points at the start of an event and you lose a match, you lose 0 points. So we'd start an event and retreat 100 or 200 times (for a net -0 points) until our W/L record went way negative, then get matched up with day 1 1* teams.

The devs picked up on that relatively quickly, and now matchmaking is roster-based.

It was glorious.
Tank in LRs until you saw seeds, then start regular PVP.
Climb a bit, then tank down again.
So... Peaceful.

dianetics

@entrailbucket said:
If BCS is doing things differently and are purposely creating good/bad characters, that means that some of these mechanics will need to be updated, because they don't work correctly if that assumption no longer holds.

BCS has to deal with legacy development and also move forward with their own development. They are not responsible for Havok, but they have to account for him now.

We can couch call this all we want, but we have to accept that the game we love has to move forward.

entrailbucket

@dianetics said:

@entrailbucket said:
If BCS is doing things differently and are purposely creating good/bad characters, that means that some of these mechanics will need to be updated, because they don't work correctly if that assumption no longer holds.

BCS has to deal with legacy development and also move forward with their own development. They are not responsible for Havok, but they have to account for him now.

We can couch call this all we want, but we have to accept that the game we love has to move forward.

Right, so like, if they've decided to purposely make some 5* better than others, then they need to adjust these systems to account for that. Matchmaking needs to know about it. They also need to adjust things like rewards, token drop rates, and essentials. And they should create some kind of indicator in game so that players aren't forced to go out of game for necessary information.

KGB

@entrailbucket said:

@KGB said:

@entrailbucket said:
The problem now is these giant gaps in power level between the best and worst characters. If you champion a bad character you're going to be matched with all the best characters, because the algorithm still assumes they're all the same.

This they can fix (accidentally leveling a bad character).

Just give players the ability to unlevel a character. So if you unknowingly champed 5* Wasp and found PvP impossible, you could unlevel her back to whatever level you wanted (getting the ISO refunded). Any champ levels are lost (since you got champ rewards).

This helps new players too who level up a 5* to say 300 because they lucked into a couple of covers. they could unlevel back to 255.

KGB

They can't do this because it breaks their business model. If you could unlevel and relevel characters, it makes things like essentials meaningless.

It'd also make PvP matchmaking pretty hilarious if I could level and unlevel my roster however I wanted...I could unlevel everyone and climb off 1* until I ran out of points, then selectively level up folks until the next tier became visible.

How could you do this? The min level of your 5 stars is 255 so how could you climb off 1 stars.

As I said, you would not be able to undo champion levels so once you applied any champion level you would not be able to unlevel that character. I doubt anyone would forgo champion levels just to attempt PvP climbing with 1 stars.

Also a simple tax on the ISO returned (say 10 or 20%) would make sure players weren't doing this all the time and only doing it for a mistake.

It's by far the easiest fix for new players accidentally leveling characters and finding PvP impossible because it means they don't have to do anything to MMR to try to figure out if character X = character Y in terms of power.

KGB

entrailbucket

It sure sounds like you're designing a system specifically so it can be manipulated by softcappers, not by innocent new players who made a mistake...

Anyway I'd much rather, y'know, fix the massive discrepancies in power level that cause these problems, rather than encouraging everyone to purposely underlevel their characters.

Massive power level discrepancies are great for optimizers and powergamers and awful for everyone else. In a casual phone game, I don't think you want to cater to the most fanatically entitled optimizers (who, incidentally, are also the least likely to contribute financially).

pepitedechocolat

@entrailbucket you are overplaying much the "massive discrepancies" in both "relative power level" and "number of outliers". I fail to see any of the problems you are describing, while playing maybe 2 hours every day. Only issue I see is chasm is kind of an outlier in 5* land, and he is being fixed, like the other ones before him since the very beginning of the game

U are actively suggesting clueless player should be rewarded as much as player playing "seriously". This is not fair, nor desirable, nor sensible economically. I mean if you want to play a game with no competition this is not MPQ.

entrailbucket

Great! I have a roster of 15 fully maxed out 550 5*, and another 40 or so over lvl 500. I've been playing for about 10 years. I am as far from a new player as it gets. On your scale from "clueless" to "serious," I'm about as serious as it gets.

By your logic I should be rewarded with easier matches, correct? Why am I being "punished" for playing well?

As far as power levels, if you seriously think the gap between Polaris and Spider-Woman isn't a massive discrepancy, I don't know what to tell you. There are tons of examples at the 4* and 5* tier -- these gaps in power level are not small.

KGB

@entrailbucket said:
It sure sounds like you're designing a system specifically so it can be manipulated by softcappers, not by innocent new players who made a mistake...

Anyway I'd much rather, y'know, fix the massive discrepancies in power level that cause these problems, rather than encouraging everyone to purposely underlevel their characters.

Massive power level discrepancies are great for optimizers and powergamers and awful for everyone else. In a casual phone game, I don't think you want to cater to the most fanatically entitled optimizers (who, incidentally, are also the least likely to contribute financially).

How does deleveling characters create a system that can be manipulated by softcappers (or anyone for that matter)? Can you explain your concerns or ideas on how exactly that would work. Softcappers can already softcap now so they gain nothing by deleveling characters.

You've already admitted (in many threads) there is virtually no way to fix the massive discrepancies in power levels in a game with 130+ characters in the 4 star tier and close to 100 in the 5 star tier.

First, the community has repeatedly said no nerfs, buffs only. So even the tiniest nerf sets off massive wailing and demands for compensation. So they are essentially stuck with only boosting (which of course is a chase your tail game of requiring ever more boosting of older characters).

Secondly, the resources needed would be off the charts because even if they could (or would) change 10 characters a week it would take 20 weeks to go through the 4 and 5 star tiers (200+ characters). Then each change would need some time to percolate and go through a boost cycle and of course need to be evaluated with future characters who get changed in a few more weeks etc. It would drive players nuts to be constantly having to deal with characters getting new power sets and dealing with bugs and newly over/under powered characters so on (esp for newer players trying to figure out where to devote their limited resources).

Thirdly, as you have noted many times, they must constantly add new characters. These new characters have power creep built in (esp 5 star match damage) as a feature to get players to chase. So that of course upsets power levels.

It's a never ending hamster wheel and we all know it. Pretty much all they can do is handle the extreme outliers like Chasm and boost some under powered characters now and then and wait some time to see how that plays out. But it's going to take a VERY long time (years) to play out.

KGB

ShionSinX

@DAZ0273 said:
If you think about it then it is not too surprising that you will find MMR changes at certain levels simply because what those levels represent.

>
I started seeing champed Polaris when my 4★s were around lv210~225, but her partners would usually not be champed themselves (undercovered Grocket 90% of the time if it was a 4★). I also saw more 4★s than 3★s at that point, even tho maxed 3★s level were closer than champed 4★s. Stepping into 240, both Pol and whoever was with her are champed, and lots of Juggs, who I didnt see at all until then. I can also add that Sabretooth, the 'Polaris counter', is rare and Gorr is inexistent on my slices (usually 5, rarely 4).

What you say about the algorithm pushing whatever is closer makes some sense, but on top of that it seems that it gives you things that are almost always at a higher MMR, rarely at the same level (possibly for the lack of 'perfect targets') and even rarer are people below you (at least until they break the roster MMR barrier with high scores).

Phumade

@ShionSinX said:

What you say about the algorithm pushing whatever is closer makes some sense, but on top of that it seems that it gives you things that are almost always at a higher MMR, rarely at the same level (possibly for the lack of 'perfect targets') and even rarer are people below you (at least until they break the roster MMR barrier with high scores).

Yes, The algorithm default is to give you a slightly tougher match. It will never give you an easier cupcake until your point total breaks MMR threshold (at that current second). Don't forget people shielding and unshielding can repopulate or deplete your que. This creates either cupcake chances or clogs your que and you'll only see that same 4-5 opponents. This is all very nuanced and highly dependent on the initial state of whos actively playing/ floating etc. There are pretty well established point bands, that people can guide you through, but its an inexact science at best.

DAZ0273

@ShionSinX said:

@DAZ0273 said:
If you think about it then it is not too surprising that you will find MMR changes at certain levels simply because what those levels represent.

>
I started seeing champed Polaris when my 4★s were around lv210~225, but her partners would usually not be champed themselves (undercovered Grocket 90% of the time if it was a 4★). I also saw more 4★s than 3★s at that point, even tho maxed 3★s level were closer than champed 4★s. Stepping into 240, both Pol and whoever was with her are champed, and lots of Juggs, who I didnt see at all until then. I can also add that Sabretooth, the 'Polaris counter', is rare and Gorr is inexistent on my slices (usually 5, rarely 4).

What you say about the algorithm pushing whatever is closer makes some sense, but on top of that it seems that it gives you things that are almost always at a higher MMR, rarely at the same level (possibly for the lack of 'perfect targets') and even rarer are people below you (at least until they break the roster MMR barrier with high scores).

The game AI does not understand softcapping and is not programmed to take into account players attempting to avoid fighting certain rosters. It also assumes that the closer you get to 270 that you are a champed 4* player simply because there are more champed 4* players with 270 than there are max champed 266 3* players WITHOUT a champed 4*. Likewise with levelled 5*. It doesn't have the capability of evaluating what those levels mean and whether they create a false equivalent - it is just trying to match you up with what is available and most of what is available accords to these certain levelling milestones. It isn't set in stone of course but once you reach a certain MMR level then you might as well go to the next Milestone as fast as possible because you are simply kneecapping yourself as the better teams will not stop coming.

There is a way to see what your MMR current ranking looks like through team ups received from Alliance members although I can't remember exactly how it works - maybe if a player who has a higher level character than your version sends you that team up you get it at your MMR rank? That could be wrong. Someone on here will know exactly how it works. In my experience that rank is usually slightly higher than your top character but I assume it varies.

Surprised people are not using Morbius - is he not a better counter than Sabretooth to Polaris?

entrailbucket

@KGB I am going to try very very hard to not write a million words in response, but that reply is worthy of it...you said a lot of thoughtful things that I absolutely agree with.

As far as the delevelling thing -- I assumed that softcappers chose not to champion their highest level characters. If so, being able to delevel any non championed character would allow those players to rapidly increase and decrease their matchmaking range. I can think of a ton of ways I'd exploit this currently if I could. In PvP I do something similar by rapidly increasing and decreasing my score to change who I can and can't see, who can and can't see me, and how badly my attacks hurt, at any given time.

For the rest: the balance problem is not insurmountable, although it's big. Yes, the "community" has told them no more nerfs ever, but the community also wants things that are most easily accomplished with targeted nerfs (fairer matchmaking, more variety, interesting fights, better rewards). Ignore the short term reaction. The long term reaction to 90% of nerfs has shown they were correct to act.

I've written this before, but it's a three part plan.

1) Fix the outliers at the 4* and 5* tiers -- everyone who's either way too good or way too bad. There aren't a million of these, I'd say it's less than 20 or 30. Yes, this would take awhile at their current pace. I don't know if it can be sped up, but I'd also do less extensive changes to these guys as a first pass. Either increase or decrease some numbers to start, then go to step 3.

2) Try really, really hard not to release any more outliers, either too good or too bad. BCS are honestly not bad at this.

3) Monitor, monitor, monitor. If there's a power level problem, act quickly to address it, not with some enormous power level swing, but with small tweaks up or down.

The expectation is not 100% parity, but giant gaps like Chasm vs Rescue or Juggernaut vs Spider-Woman can't happen, because they break the fundamental mechanics of the game.

KGB

@EB,

I suspect it's a lot more than 20 or 30 characters across both tiers that need adjustments. You yourself have identified at least 20 5 stars that you've said can't be played even when boosted.

I think the useless characters are easy to identify. But I suspect the too good characters are going to be a LOT harder and more contentious to identify. For example consider Polaris, who the Dev's have mentioned. Is it her engine (creating more SAPs/passive damage on SAPs) that's the issue or is it Grocket's and BRB's fuel (their passive tile placement at start of battle) that's the issue. Because she's rarely used without one of those two and when she is, she's not remotely scary or over powered if she has to rely on someone (including her single SAP placement) to place the SAP's via a fired power. It's almost the exact same argument from several years ago when it was Kitty/Grocket and the argument was whether it was Kittys boost that was the problem or his initial strike placement that was the issue.

I also think there would be a huge push back on changes to how powers work. Look at the recent change to Dark Beast (a character I've essentially never seen used in PvP) and how many got angry that his powers got changed. Or remember when they buffed Robbie Reyes and how many are still mad that his Green power got changed even if he's better overall than he was. There is just so many downsides in terms of player anger for them to change how powers work and yet so many of the useless characters or even semi useless ones need power reworks more than number changes and getting consensus will be near impossible. It's easy to see why they don't want to do it.

KGB

entrailbucket

I think if they hit those 20 or 30 that needed adjusting the most, a lot of other things would get fixed.

And yeah, a small percentage of entitled vets get really mad and start throwing tantrums when you nerf their overpowered guys (or change anything at all -- were you around for the insane overreaction to the game board rework? Nobody even remembers the old board now.). The reaction to the Bishop nerf, Gambit, etc, were crazy. But those changes all made the game better, and now nobody remembers that they lost their minds over them, or cares about those changes.

I do think the reactions of the extremely loud minority stop them from doing some things, but 1) it's a minority and 2) those players generally aren't spending and likely never will. They also don't care about the overall health of the game, just what's best for them and the few other players like them (which is winning everything forever).

The main thing is that you get all this other good stuff. The example that started this thread -- matchmaking is screwed up now, and one of the biggest reasons is overpowered/underpowered characters. If you level up the wrong guy, because maybe he's your favorite from the comics, or RNG decided to give you enough covers, you can screw up your game. That shouldn't happen.

KGB

I've been around for over 3000 days but I didn't join this Forum until a couple years after I started playing because the first year or longer was just my step son playing the game on my phone till I got hooked and took over the account. So for a year or longer there was no real progress on the game beyond daily login and I never looked at Forums for help etc.

So yeah, I remember well the new game board (and all those old nerfs). To be fair, a lot of the anger about the game board came from the annoying flashing. Once that was fixed most players were fine with the current look/feel. It was amusing to look at those links you posted from 2014 about the interview with the CEO of the game back then. Those screen shots really took me back.

Bishop and Gambit were asked for nerfs for a long time. So is Chasm and that's why he's going to get one too. Not sure any other character has been asked to be nerfed though (by that I mean regularly asked for and for a long time). So I'd be extremely cautious about nerfing anyone else besides Chasm esp until we see how big of a nerf he gets (no one wants a Demi nerf). I think players would much prefer the bottom 30 characters (say 15 in each tier) get brought toward the middle before anyone else gets brought down.

KGB

Bowgentle

In Chasm's case, he NEEDS a Demi nerf.
Tweaking his numbers a little isn't going to stop people using him, he'll still be too good.
He just does everything, he needs to lose either his endless resurrection, the AP drain, or the healing.
Possibly even two of the above.

Alex502

@DAZ0273 You're very insightful even when you say you don't know something.

@KGB You always bring a great perspective and got excellent points, Demi took their time with nerfs more than character production, but they always had a certain eye around roster syngery. Weird balance between the two and I think BCS is missing the mark a bit.

@entrailbucket I've been saying for several weeks now that BCS aught to focus more on quality assurance on their additions to the game and maintaince on the game's overall functionality before adding new content, and they've not once commented to that end. Not sure if its a staffing thing, priority issue, or if they're just more interested in new development over the QA process, however all that said I was going to say to

@Bowgentle Chasm is in the process to be nerfed, BCS has stated that a change for him is in their QA process, however as I just mentioned, that QA team might need some more support from the Dev side before the Dev's can get their newest additions across the line.

Overall, I'd say that BCS has a pipeline issue of under prioritizing the user side of their developments and not carefully testing through their changes before execution into production. Its pretty clear they've slipped a time or three or five on sending out a production model before it was ready. Easiest examples are the Annihilis Torch and the Mother's Day Vault. 0/5 on execution there, guys. Demi never had THAT kind of problem.