|
Tuxedo Catfish posted:Again, it isn't straight win/loss ratios, that would actually, in fact, be dumb. I... know? That's why I said I liked it. I was referring mostly to MOBA-like systems in other games that use MMR based off ratio. I like OW's system more but think the weights need tweaked. I don't know what you mean by relative cross-reference, though. Kaplan already made a huge post regarding how Skill Rank is calculated and it takes performance stats into account with W/L being the biggest factor. So, yes it is relative to opponent MMR like all systems, but it does "baseball stat" your game to a point as well. Axel Serenity fucked around with this message at 17:57 on Jul 5, 2016 |
# ? Jul 5, 2016 17:50 |
|
|
# ? May 9, 2024 23:28 |
|
It's the "baseball stat" part that definitely needs some more tweaking; support players are being undervalued in the current system, and supposedly there's a major issue with KOTH maps counting for way less than assault maps in terms of rank points. I am curious about the data behind the perception that it takes so many wins to counteract a single loss in competitive. I've experienced it too, but it can't be universally true or else everyone's ratings would be tanking simultaneously.
|
# ? Jul 5, 2016 18:01 |
|
Axel Serenity posted:Overall, I like the idea of their system and how it uses performance as a measure. Straight win/loss ratios being used as MMR is really dumb, and it's weird to see so many gamers try to stand up for it when no sport in the world uses that for individual player analysis. But, the numbers really need adjusted. Punishing the winning team for a leaver needs to be fixed. If a person is going to hover, don't make it feel like a depressing inevitability that they will not keep their rank. No sport in the world analyzes individual performance? What? Are you just saying they don't use MMR to put together random teams? I mean there are a billion "player rankings" for every sport. Even if I'm misunderstanding, you can't equate this to normal sports (or even esports) since you don't have a set team, so you can't have a "team rating" in the ranked system.
|
# ? Jul 5, 2016 18:05 |
|
Brosnan posted:I am curious about the data behind the perception that it takes so many wins to counteract a single loss in competitive. I've experienced it too, but it can't be universally true or else everyone's ratings would be tanking simultaneously. I was actually wondering why quick play MMR is invisible but after seeing how people react to competitive rating I now understand completely.
|
# ? Jul 5, 2016 18:05 |
|
Sindai posted:Confirmation bias is very powerful. Yeah I have no problem with that, having seen the result of how Heroes of Newerth and DOTA's respective approaches affected player behavior. (HoN surfaces every stat possible and had a community that was about as lovely as you'd expect as a result. Yes, worse than Dota's.)
|
# ? Jul 5, 2016 18:10 |
|
A Saucy Bratwurst posted:100% chance your messages are 100% more aggressive More than probably in tone if not in content, agreed. I don't call people names or curse excessively at them, but for me it's the same as team sports. When I'm playing football and someones passes are sloppy, I'll tell them to get some focus on those passes instead of just spraying them out. It's an expected part of playing a team game for me. If I'm not standing where I should be on the field, the guy next to me is expected to tell me that my man is free and I need to start moving. Will we sound annoyed at each other? Sure. But once the game is over it's to be understood that it stays on the field and everyone tried their best. Obviously stuff like "you're poo poo" or "loving christ at least try" isn't ok. Better to just play with likeminded people. It's not like there's anything wrong with playing double dwarf symmetra on sudden death defense, but I'm perfectly happy to leave that to quick play personally and just eat playing the boring hero to try and win in comp.
|
# ? Jul 5, 2016 18:22 |
|
Brosnan posted:I am curious about the data behind the perception that it takes so many wins to counteract a single loss in competitive. I've experienced it too, but it can't be universally true or else everyone's ratings would be tanking simultaneously. I'm definitely interested in this too as my entire regular crew (of 8-ish people) has experienced the same thing.
|
# ? Jul 5, 2016 18:24 |
|
Sindai posted:Confirmation bias is very powerful. I think it's kinda fun watching the dreams shatter as people find out they aren't as good as they think they are. it's the system that must be rigged instead of the fact that with millions of people playing it's not inconceivable that they might not be even average atm. (I'm bad and happy playing mei forever with/against scrubs)
|
# ? Jul 5, 2016 18:29 |
|
It's funny that the high mmr guys in competitive mode are doing 3 mans instead of 6 stacks just to get paired down a bit.
|
# ? Jul 5, 2016 18:33 |
|
Axel Serenity posted:I... know? That's why I said I liked it. I was referring mostly to MOBA-like systems in other games that use MMR based off ratio. I like OW's system more but think the weights need tweaked. None of the major MOBAs use MMR based solely off of win/loss ratio either, which is why I was confused. If that's what you're complaining about you're complaining about something that just... doesn't happen. There are basically two major systems at work: pure MMR/Elo systems, where you rank up or down based on win and loss and how much you rank up or down is based on your rating and your opponent's rating, and hybrid ones where the amount is affected by things like in-game stats (KDR, healing done, gold earned...) or win streaks (I know HoN adopted these a few years ago, dunno about DotA 2 or LoL), etc. I'm for the former and against the latter. Overwatch falls into the latter category. Tuxedo Catfish fucked around with this message at 18:36 on Jul 5, 2016 |
# ? Jul 5, 2016 18:33 |
|
subx posted:No sport in the world analyzes individual performance? What? Are you just saying they don't use MMR to put together random teams? I mean there are a billion "player rankings" for every sport. Huh? That's not what I wrote at all, dude. I said no sport uses straight win/loss to judge individual performance. And I don't see how you can say that this isn't equatable to other sports. MMR is essentially a Fantasy League that is being constantly updated in real-time. It's also important to note that performance means a lot more than just medals. When talking about the POTG system, the devs were saying they have all sorts of things they can adjust and track as a game progresses, including how fast someone is moving when they are sniped by a Widowmaker. No one knows the exact formula they are using to adjust ranking. Brosnan posted:I am curious about the data behind the perception that it takes so many wins to counteract a single loss in competitive. I've experienced it too, but it can't be universally true or else everyone's ratings would be tanking simultaneously. No one knows but Blizzard, but it's a very common complaint. Even Seagull was getting upset over it the other day after seeing how the losses are weighted versus wins in a semi-pug group. There's also a fairly large amount of people complaining about losing rank despite having a positive win rate. How much of that is statistically true and how much is anecdotal from loud posters is up to Blizzard to decide. However, there are a lot of legitimate complaints about how Ranking is calculated that are more than the usual "I should be higher!" The Leaver Penalty is probably the most egregious right now.
|
# ? Jul 5, 2016 18:35 |
|
Well....I mean...... Chess kinda does use only W/L for player ranking.
|
# ? Jul 5, 2016 18:39 |
|
piratepilates posted:Well....I mean...... Chess kinda does use only W/L for player ranking. no? elo adjustments are based on the differences between your and your opponent's elo
|
# ? Jul 5, 2016 18:40 |
|
piratepilates posted:Well....I mean...... Chess kinda does use only W/L for player ranking. I guess I should say "team sports." i.e. ones that actually matter
|
# ? Jul 5, 2016 18:40 |
|
I think displaying your ranking like you're leveling up/leveling down is an issue because it makes it feel like "why didn't I get more experience points" which isn't what it is.
|
# ? Jul 5, 2016 18:41 |
|
Using only wins & losses to inform ranking is NOT the same thing as using your winrate as your ranking. The former is fine, the latter is bad.
|
# ? Jul 5, 2016 18:41 |
|
ImpAtom posted:I think displaying your ranking like you're leveling up/leveling down is an issue because it makes it feel like "why didn't I get more experience points" which isn't what it is. Actually yeah, there are probably some UI changes they could make at the results screen that would visually communicate "You were here, your opponents were here, this was the match result, so now you're here." Only seeing your bar tank down or nudge upward leaves some blanks that you're left to fill in mentally with either educated guesses about how the system works, or the alternative "this is bullshit gently caress you Blizzard!!" explanation.
|
# ? Jul 5, 2016 18:44 |
|
MacheteZombie posted:It's funny that the high mmr guys in competitive mode are doing 3 mans instead of 6 stacks just to get paired down a bit. It's not just that they get paired down, I don't think most of those guys hate good competition. The problem is it seems to also give you more xp based on your team's average level vs the enemy team. So like maybe you are 70, but if your average team is 55, and you are playing against an average 55, you get a lot more xp. It seems like how much you LOSE depends less on relative levels too, so a 75 vs 75 game, the loss is still huge, so you are basically hoping to dodge other good teams, which sucks. So what incentive is there to play in a 6 stack and only queue against lower-ranked averages where a win gives 5% of a level, and a loss loses an entire level? And even at rank 60-65ish where I hover, teams of 2-4 seem to gain a lot more xp than full stacking. No idea why though, as at this level there has to be other 6 stacks to play against.
|
# ? Jul 5, 2016 18:45 |
|
Matchmaking prioritizes group stack patterns above almost everything else that it considers (i.e. it really really doesn't want you to be a solo player against a 6 stack or whatever), but the likelihood of two 70+ 6-stacks that are otherwise a good match for each other queuing simultaneously is fairly low, so a full stack of pro players will likely end up matched with a full stack of 60ish players, and that's kind of a losing proposition for them. (The little-to-gain, lots-to-lose problem we've been talking about.) Blizzard is considering how and when to bend the matchmaking rules to help cover edge cases like this. Maybe having a 6 stack against a 4-man and 2-man team is better when everyone is 70+, or maybe that wouldn't do much to improve matchmaking times and wouldn't be worth bending the system for. You'd need to see their data to know.
|
# ? Jul 5, 2016 18:49 |
|
ImpAtom posted:I think displaying your ranking like you're leveling up/leveling down is an issue because it makes it feel like "why didn't I get more experience points" which isn't what it is. Bingo bongo, once they get some lovely feeling behaviour about the MMR change ironed out they should just get rid of that bar and not play up the rank so much. I like having that number, I just hate seeing these number change every time.
|
# ? Jul 5, 2016 18:51 |
|
ImpAtom posted:I think displaying your ranking like you're leveling up/leveling down is an issue because it makes it feel like "why didn't I get more experience points" which isn't what it is. Right, in an ideal system where blizzard got the algorithm perfect and no one got better or worse, the bar basically never go up or down because it has you pegged perfectly.
|
# ? Jul 5, 2016 18:51 |
|
ImPureAwesome posted:Right, in an ideal system where blizzard got the algorithm perfect and no one got better or worse, the bar basically never go up or down because it has you pegged perfectly. That's not really what he's saying and it's also not how the system is supposed to work.
|
# ? Jul 5, 2016 18:53 |
|
ImPureAwesome posted:I think it's kinda fun watching the dreams shatter as people find out they aren't as good as they think they are. it's the system that must be rigged instead of the fact that with millions of people playing it's not inconceivable that they might not be even average atm. This is easily the best part of competitive mode "Those puh-puh-pubbies cost me another rank," he said, while slam-picking Hanzo for the fourteenth time that night
|
# ? Jul 5, 2016 18:55 |
|
Brosnan posted:That's not really what he's saying and it's also not how the system is supposed to work. It's supposed to figure out who you're competitive against and keep you there unless you get better/worse
|
# ? Jul 5, 2016 18:57 |
|
ImPureAwesome posted:I think it's kinda fun watching the dreams shatter as people find out they aren't as good as they think they are. it's the system that must be rigged instead of the fact that with millions of people playing it's not inconceivable that they might not be even average atm. (I'm bad and happy playing mei forever with/against scrubs) so how do you account for having a 53% win rate in one mode and a 20% win rate in another? that's beyond 'git gud'
|
# ? Jul 5, 2016 18:58 |
|
Soothing Vapors posted:This is easily the best part of competitive mode i play Lucio tyvm
|
# ? Jul 5, 2016 18:59 |
|
Efexeye posted:so how do you account for having a 53% win rate in one mode and a 20% win rate in another? that's beyond 'git gud' I assume the algorithm for what gets prioritized when matching is different from qm to competitive. maybe that?
|
# ? Jul 5, 2016 19:02 |
|
ImPureAwesome posted:I assume the algorithm for what gets prioritized when matching is different from qm to competitive. maybe that? seems like that's a legitimate complaint, no?
|
# ? Jul 5, 2016 19:03 |
|
Efexeye posted:so how do you account for having a 53% win rate in one mode and a 20% win rate in another? that's beyond 'git gud' Sample size, different player pool, and your ranking still being adjusted relative to that player pool in one versus being fairly well established in the other
|
# ? Jul 5, 2016 19:04 |
|
Efexeye posted:so how do you account for having a 53% win rate in one mode and a 20% win rate in another? that's beyond 'git gud' This is really the problem. There's "complaining about your rating" and then there's things like "why does an opposing player leaving give the winners minimal XP, too" or "How can I win as an Underdog and only gain marginal XP but lose a rank by losing against an underdog team" or "Why am I droppin ranks with a positive WR?" I would say those are some pretty legitimate concerns. That doesn't even get to the disparity in play modes, which I would be curious to know. Mine was about 60% in QP with over 300 games played and it's definitely not even close to that in Competitive, though there are a lot of factors that probably change due to the extended modes. It's not even necessarily a complaint about Competitive; it's purely the ranking xp system as it is right now. In terms of the games themselves, I've been having a lot of fun the past couple of days and actually enjoy the mode changes. Sudden Death can suck a dick, but that's really the only issue I've had with the actual playing of CM so far. That and the one drunk Hanzo I got stuck with two games in a row. ImPureAwesome posted:I assume the algorithm for what gets prioritized when matching is different from qm to competitive. maybe that? That could be, but then I wonder why they would use QM MMR as the template for your placement matches and CM mode ranking, though. Plus, I think they've already outlined what they use as priority. After MMR, I believe latency was the next biggest thing.
|
# ? Jul 5, 2016 19:05 |
|
ImPureAwesome posted:It's supposed to figure out who you're competitive against and keep you there unless you get better/worse "Keep you there" means your level will go up and down but trend toward a particular number. Your performance from match to match (and moment to moment within a match) will vary irrespective of whether your "skill" is generally improving or not, and no system of this kind works without allowing fluctuation. The fluctuation isn't the problem; the way it currently communicates the reasons for that fluctuation is.
|
# ? Jul 5, 2016 19:07 |
|
Efexeye posted:seems like that's a legitimate complaint, no? Only if you never settle into a ~50% win rate eventually
|
# ? Jul 5, 2016 19:09 |
|
People really do think of skill rating as a progression system, and the way it's presented probably has a lot to do with that. They expect their rating to go up if they put the time in. They don't expect it to settle in a certain range, and they definitely don't expect it to drop. Maybe having non-numerical classifications is the answer, like Starcraft's leagues or Splatoon's letter grades. People feel like numbers in video games are supposed to increase, especially when they're represented by bars that fill up. If your rating were summed up as a qualitative class, it might curb that feeling.
|
# ? Jul 5, 2016 19:10 |
|
winningasamedic.jpg
|
# ? Jul 5, 2016 19:12 |
|
Supreme Leader Kaplan says that changes to help with the competitive leaver issue are coming this week http://us.battle.net/forums/en/overwatch/topic/20745746206#post-11 quote:We have some improvements coming that should address some of the "Leaver" issues. We should be able to get some in this week and more are coming later on.
|
# ? Jul 5, 2016 19:14 |
|
How about a system where your skill rating wasn't shown, but instead you had an XP bar that filled more or less at the end of each match based on your performance? Also, you could have shorter matches so that losses feel like less of a waste of time.
|
# ? Jul 5, 2016 19:16 |
|
fortocero posted:Hate to burst your bubble but there's a bug where you see that as your icon. I get that every other game as well. You're not top 500. I was aware it was a bug as I wasn't done with my placement games, just thought it was funny. Also, is that what the top500 folks see?
|
# ? Jul 5, 2016 19:16 |
|
Zikan posted:Supreme Leader Kaplan says that changes to help with the competitive leaver issue are coming this week thanks for quoting the actual text so i don't have to try and find this crap on my phone to see what he said SaNChEzZ posted:I was aware it was a bug as I wasn't done with my placement games, just thought it was funny. Also, is that what the top500 folks see? yeah that's the icon for top 500. kinda weird that it would be covering your rank number though.
|
# ? Jul 5, 2016 19:18 |
|
Efexeye posted:winningasamedic.jpg God this screen is so useless
|
# ? Jul 5, 2016 19:19 |
|
|
# ? May 9, 2024 23:28 |
|
Rocketlex posted:How about a system where your skill rating wasn't shown, but instead you had an XP bar that filled more or less at the end of each match based on your performance? Also, you could have shorter matches so that losses feel like less of a waste of time. Then it ceases to be a skill evaluation and becomes a measure of grinding. You grind less/get farther if you're better, but that doesn't really matter (see: people in Quick Play getting bent out of shape about level differentials even though those levels don't indicate skill in any way). Edit: Brosnan fucked around with this message at 19:44 on Jul 5, 2016 |
# ? Jul 5, 2016 19:24 |