A rant on MMR, MMR changes, and why MMR changes fail (long)

Marvel Puzzle Quest MPQ General Discussion
1

MMR has a fundamental flaw. The assumption that players with weaker rosters should not be matched up against stronger rosters and still get same rewards is a flaw rather than the solution. Fundamentally, the goals of MMR are that flaw prevents changes to MMR from working. A better premise for a game might be that everyone should be able to improve their roster an equal amount with an equal level of effort.

Part of the problem is that MPQ has tiered rewards that give better rewards for top performances. A typical 400-day roster from a competitive player might have 2 maxed 4* characters. To improve those rosters, veterans have to be scoring 1000+ points in a PVP. A 300-day roster is about ready to make the 3* to 4* transition. They need probably have 6 maxed 3* characters and need to score 700-900 each PVP. A typical 200 days roster is in the middle of the 2*->3* transition and needs to score 500 points and so on.

A “no-MMR” system works because players naturally stratify themselves into these levels. Veteran players with 3*/4* rosters breeze through the first 700 points. They fight for placement in about a 400 point band (between 700 and 1100). 3* players breeze throught the first 500 points and fight for placement between 500 and 800. 2* players breeze through 200 points and place between 200 and 600. Intuitively a world with no MMR works and provides a strong incentive to get a better roster. This isn’t a perfect system - top players have to do 28 easy matches to get into the 700 point range to actually start competing. This leads to weaker rosters getting hit by people that they can’t beat in retaliation. However, on the whole, the “No MMR” world works.

With MMR-systems, when those players are matched exclusively against people of the same roster strength, scoring bands form at much lower levels. However, this would provide no way for most rosters to improve since the rewards at those levels are not sufficient to improve that type of roster. It also provides little incentive to improve a roster. It does prevent weaker rosters from getting trampled. However, on the whole, a “MMR based world” doesn’t work.

In other words, we want a system that has 3 goals:

  1. Give players a way to continually improve rosters
  2. Give players a strong incentive to improve rosters
  3. Remove “trampling” of weaker rosters and get players into meaningful matches sooner.
  4. Remove “tanking” and playing of non-meaningful matches

There are many ways this might be done. One way is to take inspiration from sports and stratify players into “divisions” where each division has progressively better rewards. For example, you might have 4 divisions where the top 20% of each division get promoted to the next higher division every 4 weeks and the bottom 20% of the higher division fall back to the lower divsion (kind of like the English Premier League). For doing 5 PVP fights at one level, you get all the rewards from the previous levels. This would be a lot like starting all veteran 4* rosters out at 600 points under the current system, 3* rosters at 400 points, and 2* rosters at 200 points, and 1* rosters at zero. This would reduce the “trampling effect” and provide a strong incentive to improve a roster - to get into the highest division.

But does it work??

  1. Yes - It gives players a way to improve rosters (yes - you get a really good “door” prize for just showing up in the top brackets)
  2. Yes - It gives a strong incentive to improve and get to the next level (and spend HP/boosts to get there!!)
  3. Yes - It largely removes the trampling effect - players at all levels can be successful in their division
  4. Yes - It also removes tedium of “tanking” and playing lots of meaningless matches against weak teams.

But wait!! There are still some kinks to work out - shield hopping and alliance ranks.

Super high scores for PVP exist in the current system because players coordinate hops, attacking shielded friends, to avoid retaliations and get 30+ point victories. This is the type of ridiculous behavior that should be eliminated from a game. This is a bit harder to address without system changes. So, lets get rid of variable points for fighting higher level people. Every fight gives wins 1 point, losses negative 1/2 point. Change shields so they prevent lost points but don’t make people invisible and create rule that you can only attack people once per PVP. The more matches you play, the better your score - no more looking for 50 point fights.

For alliance scores - rate every player in the game. Give a bonus to everyone that played in an event based on difficulty of the divsion. Set scores so that top 20% of people in lower division score better than bottom 20% of people in higher division.

2

I don’t mind it being long. I would add a tl; dr at the end for lazy ppl. We appreciate it.

3

Unless the reward system is seriously revamped (something like 3* vs. 3* battles gives 5x the ISO), I believe most veterans want trampling to occur.

Winning 5 “challenging” matches is far far worse than trampling 20 or so easy matches. Against the easy matches I have the option of using my weaker characters (my own 2*, under covered 3*, not very useful 3* etc.), saving kits + health for PvE or other events. Against equal strength rosters I have to use my best team and losing them can cause a massive setback. There’s a reason why there was a lot of negative feedback when the “accidental” hyper mmr patch was released (during the Mystique release PvP if I recall).

Easy non-meaningful matches is where we make our ISO (you can look up all those 3* iso starved / 2* cover starved threads). Challenging “meaningful” matches usually require us to use ISO boosts which makes it a net loss per battle. Unless you’re increasing rewards by a large enough amount to compensate, you’re just further compounding the issue of punishing players for leveling up characters.

4

Give points for losing in higher brackets and give credit towards 5 match count for winning or losing. Once you get rid of incentive to lose (because losing will kick you to lower tier with fewer free rewards), at that pont it is about how much you play.

5

This sounds really hard to balance and test. While the idea is really solid, the current PvP system has always been the backbone of MPQ, so the devs want to tamper with it as little as possible in fear of things not working out and everything going to hell. I think we’re at the stage of the app’s lifecycle where its basically impossible to make such a drastic change on PvP: new game modes will probably become the main avenue to progress (cough daily PvEs cough) as opposed to large changes made on the existing systems.

6

MMR is fine. It’s okay for someone to get baby matches and get paired against weaker opponents because they still won’t get top tier rewards. If you get the point values to get the rewards, you’ll have had to go through players who have similar rewards as your MMR will be comparable to theirs.

7

Question: would your concerns be addressed in any way if the current MMR system remained, BUT rewards – both points and ISO – were scaled based on the strength of the opposing roster?

So, for example, a team of a basic featured character plus two level 1 1* characters would give everyone only, say, 2 points and 10 ISO for beating them. An elite 4* team would give, say, 80 points and 400 ISO. And appropriate scaling inbetween. (These are just random examples; obviously a lot of balancing would be involved in getting the right numbers.) The bottom line is that if I have a regular 2* or 3* roster, everyone who beats me would get the exact same reward, regardless of whether they’re beneath me or above me.

Some obvious drawbacks: there’s less incentive to “punch above your weight”, unless the rewards are disproportionate. Also, the spread of points within the event would probably be enormous and highly, highly stratified. So maybe this isn’t workable. But maybe it is? Just throwing the thought out there…

8

Part of the problem with that is that you always have to use your absolute best characters. Most players use some type of climbing team and switch to their heavy hitters when speed matters.

Psychologically its beneficial to have easy to beat teams. Put yourself in this situation: You just joined a brand new PvP and are getting ready to climb. You’re facing an equal if not slightly stronger team right off the bat. The AI gets a lucky 5 cascade, gets an ability off and things go down hill quickly. 5 minutes later and you’re down 3 health kits and still at 0. You made no ISO and need to wait almost 2 hours for your kits to return. You could start pushing but its not the best idea because if you lose one more time you’re out for the full 3 hours and even if you do start winning you will have to stop sooner than if you had all 5 kits (which means you stop at a lower score and either shield at that low score or get beat back down to 0 while you’re waiting for kits).

A large part of the reason the mmr fiasco sucked was because you had a high probability of losing (or at the very least taking a ton of damage) and even if you did win, other players were in the same boat so you were getting attacked a lot more than usual at a very low score. There’s almost always a PvE running at the same time and currently I have to schedule when I can use my best PvP team and have kits ready around end times + rotations. In the Thor PvP right now I’m duking it out with a loaner and my 2* and sub-100 3* roster just having fun and earning ISO, saving my real Thor and heavy hitters for PvE nodes. If I was forced to face Xforce and 249 Thor’s I’d have to bring my own, only PvPing when I cleared a rotation taking minimal damage.

I just remembered, during the mmr fiasco the lightning rounds sucked because we were all murdering each other with 249 Magnetos Hoods and Xforces and no one could climb over 100. 2 players were hitting 200+ because they were invisible to maxed rosters.