They make teams like you’d pick kickball sides in elementary school.
Tommy the young jock, you have to play with Eugene the wheelchair kid. Otherwise it’d be a stomp. That’s fair right? Yeah, all the kids agree it’s about all you can do.
They figured they’d use that logic to create “even” teams in OW.
It doesn’t work.
Because every time you get hot, you’re sandbagged with worse and worse players until you lose enough SR for the game to say “okay, you no longer stand out so we’ll ease up on you”
If you don’t see that over and over and over, then you’re middle of the road and games just happen around you. Probably feels fair.
A lot of us have seen the blatantly obvious handicapping, resulting in the exact same pattern of streaks for two and a half years in a row.
Call it handicapping or whatever. It’s there and it’s been noticed by a TON of players.
I do not dispute that he means that SR is close to MMR. The distinction is that while close and accurate, MMR and SR have an internal separation as one can decrease/increase while another stagnates.
Whether or not this separation is within statistical err is completely hidden from us.
It can be likened. to a degree, to statistical +/- of data points. Data can sway up or down on that data point calculation, but because what they measure similarly reflects enough of the same variables, the disparity is such that they are closely linked in their outcomes.
Eg I have a couple of bad games. I am set to drop. I am given potato team mates until such point as I can win a game. But it’s a lot harder to win with potato team mates.
Eg, as Orisa in a solid team I go 15-2 kd and smash Kings Row in 40 seconds then full hold.
Vs.
Instalock Moira on my team that does 1000 heals in 8 minutes and a trash Widow that does nothing to stop enemy shield comp. I go 8-12 kd as I desperately try to stabilise our 4v6.
It’s easy to do well with good teams, and hard to do well with potato teams.
Your idea corresponds to mhz theory, and the game does feel like that sometimes, for sure.
As has been mentioned many times, this is completely false. There’s no concept of a player “getting hot”. There’s just your MMR. You are matched with people around your MMR. This is how MMR systems work. There’s no difference between someone at 2500 MMR who just climed from 2300 or fell from 2700. Both are viewed as equivalent players, because they have the same MMR.
We literally have people arguing exact opposites in this thread. Some people think if you play well the game “sandbags” you as you said, others believe if you perform poorly the game thinks you belong lower, and thus gives you games you are meant to lose to drop your rank. Again, these are total opposites. People “noticing” that both playing well or playing poorly makes the matchmaker give them supposedly unfair matches.
The fact is everyone sometimes gets bad games, it doesn’t mean the match was intentionally rigged. It’s obviously just people finding patterns in random noise, and then confirmation bias.
People have observed that your SR isn’t used in matchmaking at all ranks. (The old “drop SR after placements” used for early season showed it clearly). Once you accept that fact, all these conspiracy theories fall apart, because they all rely on something besides MMR to be used for matchmaking. Some idea where the system thinks you “deserve” to be higher or lower, or that it must do some compensation, besides just matching players based on their MMR.
In reality, your MMR is the systems best guess of your skill, by definition. It doesn’t mean your MMR is perfect, but it’s the best estimate the matchmaker has. This is how MMR systems work: every other thing that might influence your rank in any MMR system (streaks, past ranks, performance, whatever) all gets wrapped up into the update calculation. So the final MMR is the result of all of that information.
Again, there’s no “he’s at this MMR but should be lower”. There’s just “he should be at this rank, set his MMR to be that rank”.
Every Mercy didn’t suddenly shoot up in rank. Mercy’s who played for mass rez got more SR, and those who didn’t get less.
But let’s say they decided to change how personal performance worked, so it wasn’t based on averages anymore. It would still be a ridiculous way to redistribute ranks. So you set a certain accuracy as “diamond” accuracy. Then as the game progresses over the years, and the playerbase generally gets better and more mature, more and more people climb to diamond?
The whole point of the tiers is they are roughly equivalent to percentiles. Diamond is roughly the top 10%, Kaplan himself has posted the numbers multiple times showing it steady over years.
PBSR as it’s implemented can’t change the SR distribution, and using it to do so anyway (with a different version) wouldn’t even make sense. What’s the point of even arguing for this? Why are you insisting that there is this possible, terrible, way of scaling the rank distributions. I’m sure there’s a lot of terrible ways of doing it that aren’t in OW.
I don’t believe MHz’s theory said, that SR is used in matchmaking. The way I understand it, he speculated, that your MMR can drop lower than your SR, if you are not fulfilling the criteria for said rank set by the personal performance factor.
What about my hypothesis? Or you are not putting me into the “conspiracy theorists” bunch? I mean sure, I didn’t say they are deliberately making the matchmaking crap, but I did say this might be a result of something, say a bug. I also said, that there is indeed an issue, they don’t have much incentive to look into it, sicne people are willing to buy multiple accounts and try their luck.
“they all rely on something besides MMR to be used for matchmaking”
In my hypothesis, this extension does not affect your matches directly. It affects your MMR, and then MMR is used for matchmaking. So theoretically, you can go on a win streak, while your previous opponents might be throwing. The TTT extension then is like “we overestimated his ability, the opponents this guy defeated are actually losing a lot, so he is not as good as we thought” and it lowers your MMR. Yu get weaker allies, losing streak kicks in.
And while I’m not at all certain it works this way, it is at least a plausible explanation as to the streaky nature of the game.
Also, there is this other thing, that I’m not certain of, because I don’t play in Diamond. The personal performance factor. It doesn’t affect your points exchange at the end of a match, but is it out of the picture completely, or does it still affect your MMR.
There’s still not enough transparency.
The entire idea relies on SR being used. It doesn’t matter if your MMR is different than your SR if SR isn’t used. His whole theory relies on that BECAUSE your MMR is lower than SR, than you get weak teams designed to make you lose games.
In reality, someone at 2500 MMR gets 2500 MMR teammates and 2500 MMR enemies (on average, at least).
Okay let me answer that:
Both these statements are false. The vast majority of competitive games use hidden MMR. League of Legends, Starcraft, CS:GO, Rocket League, etc. etc. Dota 2 is the only one I know of with visible MMR.
Similarly, they all run into the same problem that matching players who are equal is best, but not always possible. In particular, at the rank extremes. How do you deal with top 500 level players when there aren’t 12 queued up? You make matches as close to 50% as you can. This is a universal principle because any other decision makes a fundamentally broken matchmaker (or simply does not allow these players to actually play the game).
It’s really annoying that people keep twisting the best and only real solution to the problem of “there aren’t 12 equal players queued up” into the idea that it’s done at their average ranks specifically just to hold them back. Even if you really think your teams MMRs are too spread out, you’ll be the lowest MMR player as often as the highest.
Finally, as for the Activision patent, you yourself said it wouldn’t even make sense in OW. There’s nothing players can buy to help them win. The best thing the OW team can do to make money is make the game fun so people keep playing.
This really makes me question this entire discussion. It’s clear you are somehow fundamentally misunderstanding what MMR is.
If your MMR goes down, your games will be easier, not harder. This is a very basic fact about how any MMR system works. With a lower MMR, are your teammates worse? Yes, but so are your enemies. Lower MMR means playing at a lower rank. It’s obviously easier to win a game in silver than in masters, even though in masters you will get better teammates.
For top 500 players there can be other solutions. For the rest of the leagues, the allowed difference in MMR should be much smaller.
This really makes me question, whether you, despite having higher rank than I do, understand Overwatch.
It’s a heavily team focused game and your performance is dependant on that of others. And especially if you main support, you’re not going to hard carry by jumping the entire enemy team and killing them (with the exception of Brig when she was released).
Again you forget, that the roles are interconnected, and you can die, because people on your team refuse to help you.
We’re just circling on the same points. This is the same excuse reworked into a different format. Your enemies are just as bad as your teammates (on average). You can die because of bad teammates, yes, but so will enemy supports. You’re not magically given extra bad teammates all the time. Sometimes you’ll get bad teams, just like everyone else.
Your games don’t be come harder to win when you are given 6 enemies who have lower MMRs than before. They become easier.
And you can carry pretty hard on support. I’ve definitely seen Ana’s (for instance) who make themselves almost impossible to kill, while still pumping out great healing, which puts you in a catch 22. Ignore Ana and your targets don’t die, or spend too many resources to kill her, and then die while the other 5 enemies had free reign to do what they want. Or you can just pick Zen and heal allies, shutdown enemy ults with trance, while still putting out DPS level damage yourself.
Honestly, how are you seriously proposing that games get harder at lower ranks, and easier at higher ranks?
Face it, you’re wrong on so many levels here. Like how you can’t see that the Mercy patch changed an outcome of a stat, but was still compared to old averages.
I honestly believe you have no credibility. Hopefully others see this too.
That’s realistically only a short-term problem. The averages fix themselves in time.
How would you propose they establish a metric for comparison’s sake when they have no reliable data to base it off of when making a large change like the Mercy scenario?
As it introduced new SR that wasn’t there before, proven.
You could manually change the average of a stat to cause increase and decrease of ranks. The metrics are already there, from lowest to highest rank, so you can adjust accordingly to create a bigger spread. This may surprise you, but mmr currently does just that.
My comment wasn’t off-topic at all, but it’s clear you lack either the ability or the desire to engage in the discussion. So, I’ll not waste anymore of my time with you on it.
Can you explain how when averages re-stabilize (ie in regards to the average rez/healing of 2500SR) will remove the additional SR (due to rez/healing of all players increasing for a period of time compared to the pre-redesign average)?
If it is not going to remove that SR, then that is called permanent. That is, until further action is taken by devs to readjust where players sit.
The timing of the performance based factors and their impact would depend on the length of history used for performance based SR, which is something we don’t know. If they use a lookback period of a month, for example, it would take less time to resolve than if they used a year.
If we assume they use all historical data, the performance based SR gets more and more accurate every day that goes by. Just looking at the Rez rates for kicks and giggles.
If performance based SR used Rez numbers as a factor and prenerf mercy averaged 6 rezzes a game and postnerf mercy averages 3 at a specific SR range (completely made up numbers) - then we can use the following to examine the trend:
History of 10 games at 6 rezzes establishes a baseline prior to nerf. Nerf happens, but the metric is still used. If you go above, you’re rewarded, if you go below (which we will use for the example), you’re punished on SR gains. At 11 games the metric shifts to 5.727 (63/11), at 12 it goes to 5.5(66/12)…at 200 games it’s down to 3.15 (630/200) so on and so forth until it normalizes where the PBSR system is working as intended all over again.
Is it a problem with the system and how PBSR works? Yeah. Is it something that requires an overhaul to the entire system to account for? Not imo. People will still rise and fall when their performance doesn’t match their MMR and the minor factor of PBSR in the equation for those below Diamond fixes itself as players continue to play the game.
Sheevah, I still think you’ve misunderstood where this conversation started. I pointed out that a claim that MMR is needed to redistribute players is wrong, as you could do the same with something i.e. PBSR. I am not suggesting it is the only method available, just that you could make it into a viable method.
That being said, FF then claimed it impossible as PBSR is a zero sum equation. I pointed it out that it actually already generates and removes SR, by design.
So in your example, while (on average) everyone under performs on rez, the overall shift of SR will be negative. Their SR gains are affected by PBSR, so as a group the total SR of the server will go down.
To claim it’s a zero sum, you now need to explain how those SR will come back naturally without any new changes. Instead in your example they reach a new plateau. SR would then change at a rate that for all intents would be a zero sum from that point onwards, but the removed SR still happened.
If you are having trouble understanding this part, then we can’t get to how you could use this to balance ranks.
This is possible. Wouldn’t be the first time I’ve jumped in on something and missed the mark by a mile.
I think I see where you’re coming from. I don’t necessarily think that PBSR is a zero sum thing on its own in short term scenarios… I could even see an argument that could be made to expressly state the PBSR aspect is there to prevent the lower ranks from being zero sum. Which, if you’re only specifically discussing PBSR and whether it should be in the game… well, I’ll leave you to it.
That said, I fear you’re focusing too much on the minor aspect of PBSR modifiers instead of the major aspect of the MMR system as a whole. Even if we assume that the entire system wasn’t zero sum (Trueskill isn’t strictly zero sum at a technical level, IIRC), how do you go from “not technically zero sum” to “handicapping system”?
Pardon me for being daft on this one, but I don’t see the dots you’re connecting. Just to try and make this as clear as possible - If we assume for a moment that SR is roughly equal to MMR, and that we only care about ranks below Diamond where PBSR gains affect the SR we see. Additionally, we’ll say no players enter or leave this small system between 2 days. Day 1 - the average SR is 2200, Day 2 - the average SR is 2150. This is clearly not zero sum but my question would be, how does this prove the matchmaker is rigged or that handicapping is a thing?
Edit:
But you would be the lowest person in a match as often as you’d be the highest person in the match over time (1/12th frequency). How is that an issue?
Yes exactly! In a lot of matches you’d likely see the highest MMR player matched against all 6 of the next highest MMR players, and teamed with the 5 lowest. That might sound extreme, but Blizzard’s handicapping system is designed to create those 50% odds for either team to win, and it’s free to do whatever it wants behind the scenes.
