Why Handicapping (MMR) is Wrong for Competitive Play

Competitive Discussion
2075

lol so you don’t know. If you are going to claim the system is rigged you need to provide proof that it is.

what?!?! how does the distribution among ranks and new account placements prove a forced bell curve?

First: the % of players in each division are absolutely not the same… not even close

Previous released player distribution:

https://i.imgur.com/LoldcuU.png

Second: it is more a log-normal distribution than normal distribution(see links below) and it’s just a display % in X rank. There will always be a peak in any of the distribution types

Log-Normal Distribution:

https://en.wikipedia.org/wiki/Log-normal_distribution

Normal Distribution:

https://en.wikipedia.org/wiki/Normal_distribution

Third: new players do not place anywhere, they place where their MMR(from QP) and placement matches put them. This is just too ridiculous…

2076

I mean…I assume you realize that “average” by definition means that 50% is lower?

How can it possibly be any kind of problem that the middle is in the middle? Plus, the average is actually mid-gold, so you have 2.5 of 7 ranks below average and 4.5 above average. Which, let the GMs have their rank. It’s ok to have a top rank be a small percent.

Now, IF we grouped by rank rather than MMR I could see this as a problem, but we don’t do that. If you’re 1999, the chances are high that your team and opponents will be between very low gold and very high silver. They’re not 1500-1999. Except for groups, of course, which screws up the matchmaking.

2078

“Think of the intelligence of the average person. Now realize that half of all people are dumber than that.” -George Carlin

3 Likes
2080

lol Indeed. I tend to forget who I’m arguing with in these type of threads. thank you for reminding me :rofl:

2081

Bell Curve is just a slang for normal distribution. it is only a representation of given population separated by ranks.

Forcing a specific distribution on an entire population is actually absurd in that all distributions relating to human behavior(yes gamers fall into this) can be easily modeled using well know population distribution models.

As far as your question. yes a curve can be manipulated by falsifying data, No it cannot be forced in the sense you cannot rig a system used by humans to give a specific result.

Distributions:

https://www.itl.nist.gov/div898/handbook/eda/section3/eda366.htm
2 Likes
2082

I’m not sure I understand you. How do you think MMR is used to form matches? Where is the following incorrect?

(4) Overwatch Forums “In Overwatch, whether your MMR goes up or down is contingent on winning or losing. But there are a number of factors that determine how much that rating goes up or down.”
(7) Overwatch Forums
(23) Overwatch Forums
(32) Overwatch Forums
(47) Twitch

2 Likes
2083

Not only that, but studies have shown that 90% of people think that they are above average…

Which means, if my math is right, that 40% of the population is delusional or, at BEST, woefully misinformed.

I suspect that the 10% who recognize that they are below average don’t stay below average for very long.

2 Likes
2084

Ok I just saw the comment and felt compelled to make sure people reading this understand that a bell curve can be artificially created, induced, manipulated etc.

2 Likes
2085

I agree with you that you can force one or skew one depending on what you’re attempting to do or how you want to present data.

That said, it is important to note that when it comes to the ranks in Overwatch, no manipulaton is necessary for the distributions that we’ve seen. Not to say that there isn’t anything more sinister going on under the hood, but we can’t look at the distributions we’ve been shown and reasonably reach the conclusion that something sinister actually is going on.

1 Like
2086

You are talking not about average, but about median. What I meant was: with natural normal distribution, middle gold is supposed to be rank of average skill. But with new players generously placed in there, the skill level of gold is dispersed, and it makes some games (around 35%) outright painfull.
I know, we all here are to please this 1% of gm players, pro oriented environment, etc. But what about other players? All I want, for example, is to avoid one sided games, which dominate the comp experience. Truly equal games would be a lot more frequent, if that mmr to sr relation was a little more definitive.

2 Likes
2087

All hail George Carlin!

1 Like
2088

There is nothing sinister. Just intentional design according to business needs.
I just keep telling the same thought: there is nothing natural in players distribution across ranks. Games happen in controlled environment, the outcome of next game is dependent on the previous game result, and “random” players in the same solo que match are not so random.

1 Like
2089

No one is saying the system is perfect, because it isn’t. But most of the “suggestions” will never work. If you can’t see what is wrong with what you said ill point it out for you… I’ll tell you what. Start noting down every session you play. I bet you will find that over the course of a season or two that the enemy will have more leavers than your team. It is just how statistics work. Everyone has had issues, hell i almost tanked into plat at the start of last season due to back luck… Made no difference to my rank at the end of the season.

2 full MMR resets a year? Multiple issues with that:

  1. It’s a kick to the teeth that have actually spent seasons working on climbing, because some people have a fallacy that it is impossible.

  2. Resetting MMR every 6 months would not give the ladder enough time to sort itself out… The system would have a worse mix than it currently has.

  3. The first season after the MMR reset would be a complete write off, with GMs/Masters/Diamond players getting put in the mix with Bronze and silver… It would be horrible

Honestly if you think an MMR reset will fix all your problems you are delusional.

I have helped many many people climb through the ranks, from silver to Diamond… And guess what? The strat is the same for literally every rank. How you play and who you play might be different… But the same overarching strat is the same.

Play better than the average person, playing your hero at your rank. Simple as that. Usually the only thing holding people back is their inability to see what they are doing wrong.

I’ll give you some free advice, one healer main to another.

Your stats are abysmal. The system isn’t holding you back, you are.
You need to almost double healing and damage done per game.
You need to almost triple beam healing done per game.
You need to almost double elims.

Your stats literally put you in the bottom 1% of Moiras…

The system isn’t the problem. If they did a full MMR reset, you would tank to bronze again, i guarantee this.

1 Like
2090

I certainly don’t think Blizzard intends to elevate specific people while shoving others down. I think that’s just a problem inherent with how they are calculating MMR on top of SR.

In every other game mode, I am playing with and against golds. Both low and high. I had a secondary plat account last year and maintained it. I clawed my way to gold on this account once and then went on a losing streak literally the day before they removed streaks.

The problem is something with MMR. I don’t know exactly what it is, since none of it is transparent, but here lies the problem at low ranks:

I have decent stats when I have decent teammates. If I am playing a tank and have DPS focusing on who they need to counter, as well as having healers who are paying attention, I can poke, soak, and wreck with that synergy as a tank.

If I am a dps with tanks actually soaking and protecting the team, and healers who don’t believe every DPS can go find a health pack, I can position correctly and do my role.

If I am a healer and I have tanks soaking and DPS actually killing things instead of feeding healers Ult charge, I am golden to heal them and provide extra dps in teamfights when everyone is topped off.

The problem is, when I’m a tank and get 0 heals, DPS are running in 4 different, I can’t perform my role. Swap to healer you say? Ok, let me swap to heals and attempt to salvage this mess of a team that all go in from 4 different directions 1 by 1.

Same for DPS and same for heals.

If you have that team of complete YOLO enthusiasts who will not work cohesively at all, there is not much you can do except try to stay positive. The issue here is those groups affect your MMR. My healing is going to be garbage if everyone is spread out. Soldier is taken by someone who keeps running in solo, so I can’t pick him to self heal. About the only character I can reliably play in groups like that is Moira since she is very self sufficient and her damage is nothing to sneeze at in 1v1.

The issue isn’t that the bell curve is forced or incorrect exactly. It is that the way MMR is forced into the SR calculations, you have a lot of people from bronze to gold who really don’t belong exactly where they are. You have tons of gold tier matches in bronze with people who have fallen and are trying hard to climb back up. You have tons of high bronze/low silver tier players in gold who stagnate in gold because the system tries to force a 50% win rate by mixing good players and lower tier players together. A single streak of bad luck can spiral into hundreds of SR lost, but since your MMR is also calculated, you’re still playing the same quality matches once you get decent groups who work well together and your stats rise again.

I don’t know why people who have never been to bronze in the last few seasons always say that people can’t aim, nobody ever works as a team, or positioning is nonexistent. That isn’t the case at all. Is it the same as plat players? No. The difference is in bronze, the “meta” is essentially nonexistent, so you have to play differently to succeed.

I know great positioning on route 66 defense. I know not to push that corner and pile on the cart. The enemy spawns 50 yards behind it, they have a tremendous advantage. But I can’t very well sit behind the corner to watch the back tunnel or sit on the gas station when we have morons as Rein, soldier, D.va, Zarya, Lucio, etc refusing to stay back at the choke. When important characters insist on sitting on the payload, I have to adapt and figure out other positions that are somewhat safe to heal from while keeping an eye on flankers so I’m still able to hopefully get help from my team.

3 Likes
2091

Wow, I didn’t want to type out statistical nonsense that wouldn’t be understood because it seems a lot of people don’t even understand how bell curves are easily manipulated.

One post saying they can’t be forced and then continues onto squishing and stretching (apparently not manipulation, but ok) and then a discussion on how bell curves are natural?

Power distribution guy gets it. Think of a percentage based bellcurve for people who can tie their shoe laces. Now think of a bell curve of people who can hit a bullseye when shooting an arrow from moving horseback.

The metrics being measured in overwatch are not defined, and the weight of each metric can and has been changed.

2 Likes
2092

I agree with most of what you are saying. The problem I run into is that Overwatch is a game with so many variables that I’m not sure the problems we routinely encounter are avoidable. :frowning:

Edit: At least from a purely matchmaker driven solution.

1 Like
2093

Again, had a plat account that I held for most of last year (missed the season over the holiday). It’s absolutely amazing what teamwork can do to your stats. As Moira, the best I can do to heal DPS going everywhere except the objective is put an orb in their direction. If people refuse to get anywhere near the objective, and scatter, I cannot beam heal them. This is the entire problem with mixing MMR into SR. I’m also not going to have high healing when we are absolutely steamrolling the enemy team, either. That is how many matches go in Bronze. A steamroll either way. I also don’t tend to play Moira in higher brackets as much, because I can actually be effective on Mercy and Zen.

I’m not saying an MMR reset would fix everything. The issue is, yeah, it may be terrible for the first season. Guess what? We had 3-4 terrible seasons to start with where the system went through numerous changes.

I appreciate the tips, and I always strive to improve. I have no delusion that I’m diamond and above. I largely got lucky with Plat because I had a lot of groups where we just all were on the same page and group queued for a while, and it is pretty easy to float in plat as a healer when groups are cohesive. I’d settle for gold at this point. I am grouped with them in every other mode anyway.

Edit: I also did tally a full season with the leaver issue. I’m constantly tabbing to check ults and character selects to see if I need to swap. 113 groups with at least 1 leaver on my team. 68 groups with at least 1 leaver on the enemy team. Sure, it definitely evens out if many of those 68 matches have 2 leavers STATISTICALLY, but that doesn’t really mean a thing in practice when the loss of a single member can turn a game. No, not all those 5v6 matches resulted in losses. Doesn’t change the fact that the average player is not playing literally thousands of games for the outcome to stabilize at 50%.

1 Like
2094

Yes. I had a low of 600 in season 2. Climbed out to gold. To get out of low bronze, you should only rarely lose a 1v1 (Have good mechanics). Regardless of hero you are playing as or against. You should almost never have a 1v2 (It’s dangerous to go alone). You should often have a 2v1 (be better grouped than the enemy). I did it as Zenyatta, but Moira should work as well.

2095

Your examples (shooting bullseyes on horses vs. tying shoes) are based on mastery and ineptitude, basically the maximum standard deviations from what an average person could do.

How do you not get that?

You make a game called Shooting Bullseyes on Horses. It’s all the rage. 1 million random people play.

If you took a million humans at random and saw how well they could shoot a bullseye on horseback, probably very few would be able to, but there would be some humans who couldn’t even get on the horse at all, let alone shoot an arrow from it, and somewhere in the middle there would be a LOT more people who would do something “average,” whatever that means.

You make a new game called Shoe Tying Simulator! It’s all the rage. 1 million random people play.

Some don’t know how to tie their shoes yet. They are bronze. Most people can tie their shoes pretty efficiently. Some people can tie their shoes faster and more secure than most anyone else playing!

You are taking extreme examples and even those can be shown to have standard deviations.

1 Like
2096

I agree with this. There is no easy fix. It would have to be multilayered, and Blizz would have to have some transparency, at least with pros in NDA agreements for feedback. I’m under no delusion that it can be fixed over night. I’m just tired of there being absolutely no work to fix it, or at least that is the impression given since Blizzard wants to keep everything under the hood hidden.

2 Likes