Make it 0.
Just because you say that loses won’t be a problem doesn’t mean wins wouldn’t be either. Nor that you debunked anything.
Currently a win still boosts a carried person (on the short term), but due to statpadding, a carried person will earn more points than the one who carried them.
Plus as said many times -which you basically left without answer- that there are things (providing vision, anchoring, trading and stuff) that helps the team win but cannot be measured or can look “bad” with a stat based metric.
Your idea doesn’t work, but my guess is that you weren’t around when PBMMR was a thing for a month(? More?). You didn’t see its chaos and glaring issues.
It’s not a new idea, we had PBMM a long time ago and GM/Master players demonstrated how easy it was to game and exploit that particular system by padding the appropriate stat on any given hero.
I know the reason for removing PBMM was not due to the reason I stated above, however we never did see it return in any form.
It’s not only stat padding though, that’s just one of the problems. My overall point is that stats don’t necessarily correlate to good play. Good play can still result in (seemingly) bad stats. Bad plays can result in (seemingly) good stats. This can happen whether it’s intentional or not. That’s why stats are not a useful way to determine rating.
Your solutions don’t solve this. They just make your useless suggestion slightly less detrimental, but still useless overall. It’s still an issue when you’re rewarding bad play (but good stats) and punishing good play (but bad stats).
And if wins/losses are still the only thing that matters for determining mmr gain/loss, what’s even the point of this system? We already have a system that gives you mmr gain/loss depending on whether you win or not. It’s nice to have failsafe to prevent your idea from doing anything, but it seems a lot easier to just not have the pointless idea in the first place.
I’m not sure what this has to do with anything I’ve posted, but I can give it a try. You actually have 2 mmr’s, one public (the points you’re at) and one hidden mmr (the rank the game thinks your skill is at). The public mmr goes up and down much faster than your hidden mmr, which takes longer to catch up.
From what I understand placement games are essentially the same as ranked games, only with potentially much looser (internal) mmr matchups and bigger changes (and I think also varying per game).
1-2 into rank up might’ve happened by first being matched against much higher ranks and winning, resulting in (for example) an internal mmr +3 ranks higher than you used to be, then dropping a bit from losing the next two games (against similar opponents) and ending 1 rank higher. Wheres the 3-0 might’ve been against similar opponents and thus not result in as big a climb (or possibly your internal mmr was still catching up from a climb, and the 3-0 merely ‘settled’ it at your rank. )
I don’t know the specifics though (nor do I know your exact situation), so this might be utterly wrong.
True in that there are hundreds of possible metrics. False in that there still is no properly definable ‘good play’ to be measured. Just pointless numbers.
Or just someone wasting his time on pointless clicking, especially if their constant movements don’t match up with the stutter step timing.
(Also again context dependent. A heated teamfight is going to involve a lot more clicking than casual map rotations).
Again, an utterly meaningless number without context (and arguably even with). Teamfight A is a long poke war with dozens of skill shots, Teamfight B ends in quick one wombo combo. Teamfight A has far more skillshots, was that better play?
Also really depends on how hittable the enemies are. You’re going to hit a lot more skillshots firing at a Johanna standing in the frontline than you are on a Zeratul that’s constantly blinking away. And if they have a healer filling her back up those Johanna hits might still be useless, and possibly even a detrimental waste (both in mana and cd’s) depending on what you’re firing at her.
How about bush checking and other predictive shots? Is it bad to throw a missing skillshot into a bush? Your hit% is going to drop from that after all.
And let’s not forget team roles either. I’m a Nova. I’m usually not going to have that many minion/merc shots, my team is supposed to mostly take care of that. Am I doing ‘badly’ for having a low number of minion skillshots hit? (And what if my team doesn’t soak, so I should be Exploding Sniping minion waves to clear?)
Again, depends on how you die. Dying to set up a team-winning wombo combo can be great, even if your death timer goes up. Dying because your team doesn’t save you when they could/should is bad for your timer. Not to mention the person that doesn’t die most of the game, but does stupidly throw feed near the end of the game and causing the 4v5 teamwipe for the loss.
Once again, depends on team roles. And what the team needs. The solo laner that soaks 2 lanes while their team agrees to stall obj 4v5 will have more xp than the one coming to help the team. So will the solo laner that ignores the team that did expect him to come help, even when their entire team is unhappy about them not coming.
Or again, think about long stand-offs near an objective. Low Xp per minute, but is it bad play?
So now you’re comparing one pile of useless numbers without context with another pile of useless numbers without context. That’s sure to give more reliable numbers /s.
My point is that there IS no ‘magic formula’. Just an infinite mountain of numbers that all need another infinite mountain range worth of context to be actually useful. You’d need to analyze every single game state to have any use for this, and that’s just not realistically possible.
I feel you are severely underestimating how much people care about this kind of thing. If there ever was such a ‘performance calculator’, i’d give it a week or two before someone posts the calculations online.
Ok, let me slightly update that then; There is no realistic way it can be done. There are ways to judge and recognize good plays, our brains do it all the time when we analyze plays. With infinite time and resources, I’m sure it’d be possible to eventually build a HotS-Playing supercomputer that can do similar things, compare all the data and game states, and then use that information to recognize ‘good play’ and provide a semi-accurate judgment of people’s performance. But just because it can technically be done doesn’t mean it’s remotely plausible.
That’s actually not true any more. It was changed when Storm League was introduced to have just one MMR now, and that is the public one seen by your rank and the points within that rank.
To be honest, I am highly skeptical of the OP’s claim to have gained rank going 1-2 in placements. Although the points gained/lost can very quite a bit due to things such as playing in a party, favored team advantage/disadvantage, or if you have leaver forgiveness going on, I can’t see any scenario where someone actually gains ranks with a losing record in Storm League.
The HUGE point you are missing is that these niche scenarios you keep complaining aren’t accounted for in this proposal… aren’t accounted for now! The bad plays you worry about are absolutely rewarded right now… so long as you win (and dont try to tell me that the team that wins is always free from bad plays). You made the right plays, but your teammates didn’t? Who cares? You lost so enjoy your -200 rank points.
The point of the system (which is listed in my original post) is that the best player in the game (who happens to be on the losing team) and the worst player in the game (who may have caused their team to lose) get the same number of rank points at the end of the game.
The problem is your rank and your skill are only very loosely correlated. Please read the whole thread.
Because you keep saying that that measuring performance outside of wins/losses either can’t be done, or is “not realistic” when that is exactly what is happening in the placement games. I went 3-0 and was give the same exact rank I had from the previous season. So if it is not measuring my wins, what was it measuring?
Man… If I gave you a single puzzle piece you’d say, “this is pointless and doesn’t tell me anything. I need more context.” I’m advocating that you take all of the puzzle pieces and put the puzzle together.
Its called statistical analysis, business intelligence, propensity to buy, Big Data, etc. Please research these concepts, you seem completely oblivious to them.
Why do people on this forum cling to rock-bottom? The current system has all the flaws you seem to be so worried about.
You can make all the right moves and if your teammates are weak, you will lose and you can collect your -200 points. Congrats… your good play was just punished.
Conversely I had a game just yesterday where my teammate (xul) just relentlessly pushed bottom lane and died 15 times and never helped in teamfight or with the Obj. We ended up winning by a hair. Congrats… his bad play was just rewarded.
Almost anything more than a 1 single metric would be better. Yet rock bottom is where we will stay.
Sorry to steal Hoku’s old picture again, but who do you think was the MVP in this match?
(It’s not Ragnaros who got the MVP award.)
Do you think the enemy Ragnaros deserves any extra points or a loss deduction for his performance? They played the match completely wrong and did not understand they were being backdoored with portals.
This same thing can be done with a Medivac (replace Medivh with Lt. Morales and charge structures on one of their lane, then the core), it’s just much harder with portals.
You can have absolutely crappy stats, zero kills and as long as you kill their core first, your team played smarter and better.
Every single action in the game either helps your team win… or helps your team lose. The stat pad argument is saying that that a player will ignore good plays that a help their team win in order to gain more performance points.
That argument makes no sense. Its like saying a person will throw away $50 in pursuit of $20. No one is going to throw away half of their points… in pursuit of getting more points.
Right now, the carried person gets +200 rank points for the win. The players who lost but played much better than the carried person, get -200. Seems like about as bad as you can get in terms of trying to make points = skill/good play.
I addressed this many times. Intangible things that increase the likelihood of a win are rewarded because a win is still half the points awarded. Right now, can you provide vision, anchoring, trading and stuff and still lose? Absolutely. Well that is punishing good play.
This is the same response given any time someone wants balanced QM matches. The old ‘they tried to fix it but only made it worse’ argument. So now QM will continue to be a joke for the remainder of the game’s lifespan. Oh well, guess its hopeless.
The problem is metrics cannot track this, and it will often result in what looks like bad stats. Anchoring means that I am not dealing damage, taking damage, generating XP, or anything at all, really. But it means that my team is doing all of the above in a safe environment. Your proposal actually punishes good tanks because as far as any algorithm can tell, they were just afk in a bush for periods of time.
This is what I mean by you simply cannot put a numeric value on good play. What you personally may think is “good” may be very bad from the perspective of a Master ranked player.
Personally, I think who is playing good or bad is irrelevant to points awarded. Like I said very early on, is the current system perfectly fair? Definitely not, but it is more fair than what you are proposing, and most other suggestions, just by the nature of the game design itself. The main reason I consider your system actually less fair than the current system is that it will be impossible to determine how many points you will be deducted, especially since such a complex system can’t possibly be 100% transparent. There will be matches where you will do everything right, lose, and lose more points than you think you deserve because the algorithm doesn’t agree with your personal assessment.
This will lead to even more frustration as you won’t have tools to tell you what you needed to do better, and the points will feel arbitrary.
Life isn’t fair, but as long as a points system is consistent and predictable, I can live with it.
My personal problem with the current system that is driving this is:
You can have two ‘Silver-3’ players with a large skill disparity. That makes no sense to me and is a gap worth fixing.
Why bother getting better at the game when you just have the same likelihood of getting a bad teammate and losing 200 points just as before? It greatly reduces my desire to play ranked.
Of course, but as mentioned, that sort of thing evens itself out in the long run. In that sense, the system is perfectly fair, because you can control your own play, so if you are one of the better Silver 3 players, your team will always have a slight advantage every match, and you will climb. It may not be as fast as you think you deserve, but it will happen.
The ranked system is not designed for someone who only plays a handful of matches a season, and I don’t think it should be. If you personally find the system unlivable, all I can say is that ranked play may not be for you.
It was raised several times already : a terrific idea that is badly implemented can worse than the current system. This is one of the concerns that are raised and the example of the PBMMR already was given. One of the fear people have is that the system cannot be actually implemented.
For instance, I would love to have a teleportation devise. Now if each time I use the devise, I miss one of my members, I not taking it. For a less science-finction like example, just take a auto-pilot car. Sure it seems to be a good idea, but if each time I use it, I have an accident, I am not taking it either.
Now, if you want to go over the discussion “Yeah but it is a good idea, why don’t you accept it?”, I think you need to talk about the implementation. You mentioned having a hundread of metrics or so but you listed only some of them. I suggest you begin to list the metrics you have in mind and the formula you would use for your performance based system.
PS @ Hoku : we already discussed it somewhere else, but I still think there are two separate numbers for your ranking, even if they are mapped .
If you honestly believe you are a much better player than your rank, coach your team during the match, ping them back if they are overextending, call boss at an opportune time. Ask people politely to get a camp before an Obj if you are on the other side of the map. If you are really making good decisions they will listen to you because they want to win also. From your OP here is a big hole I would shoot in what you want from MM. I know this has happened because I have done it. I have been carrying the entire team on my back all game. Then at the end of the match 20-25 minutes I get too cocky because I have been carrying all game and I die late in the game and that loses us the match. My numbers are amazing but I am the reason we actually lost at the end of the match.
I just want to point out an important difference. Maradonna joined a premade team and practiced with the team. I would assume that a player of his caliber has quite some influence on his team, like giving advice on how to enable him. His teammates probably knew that they just have to pass the ball to him and he will probably be able to score.
It’s not like a solo player in hots who gets into a match with random players that all individually think that they will have to be the carry of this match.
The point is that unordinary players can absolutely carry a team and that’s why they became legends. One super good player can lift up a team if he spends more time with the team.
But if you are better you already get there faster. It’s why you can gain 500-600 points per game while rest of your team still gains the regular 200 points per game. You get this by consistency, by going on winning streaks. Just like you can lose extra points if you go for long enough loss streaks. We tried PBMMR and it was utter mess mostly rewarding bad players and punishing good ones. To make it functional you would need 3rd gen AI which is never gonna happen with the few devs left working on the game with minimal resources available to them.
Heh, I almost made a similar point. A good idea, poorly implemented… is still a good idea.
The first car ever built might have been an unmitigated disaster. In fact, riding a horse might have been better than using that first car. But if the assumption is that the implementation will always be botched… whats the point of trying to improve anything?
.