Ranked mode too punishing?

Award

Squid Savior From the Future
Joined
Dec 18, 2015
Messages
1,661
@BlackZero : Very possible about TW. And if so, if it fundamentally uses the same system as the ranked system it leads me to believe the ranked system's problem may be that it's severely broken versus rigged. I mean broken in a fundamental evaluational way. TW has similar issues as jsilva notes below. Matches have started to hit for me the same problems as ranked: It's either so easy that I start to AVOID attaining more splats, and pull back, even if it gets me splatted, or so brutishly difficult I get very tempted to just sit on the spawn and squidbag for the last minute. Seldom is it in the middle, but occasionally it is. The easy matches are fun, but once I get to 10/0 or so I start feeling like I'm messing up my own player value in addition to killing everyone's game. Though it could be rigged to load balance the "invisible" ranks as well.


You had some good points although this one was my favourite. I can say my attitude in the past during losing streaks affected my gameplay. A self-fulfilling prophesy as you said, but in retrospect I'd say my attitude might have made it worse but the result was more or less inevitable.

I've been observing these things intently for about four months when I started feeling like there was more to the matching system. Throughout that time I've humbled myself and tried to analyse my own playing and tried to reevaluate situations objectively. I can absolutely respect your cool-minded approach to this and I've really appreciated your input. Even after all of my observations I've been trying to reevaluate.

However, all of my observations with my three accounts going up to S, watching the accounts of my two sons, hearing reports from other people, etc., just don't fit into a matching system that doesn't either have an agenda or a severe error that effectively results in bias, whether it be in losing streaks or easy winning streaks. I've seen way too much to dismiss it as random or an anomaly, even if things balance out in the end, which I question if they always do.

I'm glad Splatoon has done well because fundamentally it's an awesome game. But ranked matchmaking and points award/loss is not leaving a good taste in my mouth at all.
I have to agree with all of it. I really don't know for sure if it's rigged, or just broken so badly that it can't be repaired without an overhaul. But what I do know is the results are not right, and are not simple algorithmic errors that pop up from time to time due to complicated data. The results are RELIABLY bad, not anomalously bad.
 

jsilva

Inkling Cadet
Joined
Oct 30, 2015
Messages
262
mandatory modes
I hadn't really thought about mandatory modes when you brought it up before but I think you have sound reasoning. I used to be terrible at Splat Zones and I'd rarely play it. Suppose I went into a game as an S with 3 A+'s against a team of all A+'s, that means my teammates would probably lose more points because I'm on their team, and they'd be more likely to lose because I was bad.

I think it'd be overly complicated for most players to have separate ranks for each mode and we'd have more segregation, but the issue you brought up is another reason for the Splatoon developers to revise their ridiculous point system! Or perhaps they could match players based on performance in a particular mode rather than overall performance.

In any game of chance the systematic losing/winning streaks should not happen in series. It's not statistically possible to happen with such consistency.
Absolutely! That's precisely what I observe. Besides the many other things I've observed, I'm currently on my 3rd or 4th cycle of wins and loses that is overall happening the same way each time. Some variability of course, but overall it looks and feels the same and has the same fundamental outcome. I tried to break the losing side of the cycle with some heroics I didn't think I was capable of but the weaknesses of my team were revealed each time. One game recently, after getting the lead in tower control late in the game after I took out the entire opposing team on bluefin depot (after my team was dominated for the entire game), two of my teammates decide to fiddle all around the map instead of defending in the last seconds of the game and we lost. Of course! :)
 

Award

Squid Savior From the Future
Joined
Dec 18, 2015
Messages
1,661
I hadn't really thought about mandatory modes when you brought it up before but I think you have sound reasoning. I used to be terrible at Splat Zones and I'd rarely play it. Suppose I went into a game as an S with 3 A+'s against a team of all A+'s, that means my teammates would probably lose more points because I'm on their team, and they'd be more likely to lose because I was bad.

I think it'd be overly complicated for most players to have separate ranks for each mode and we'd have more segregation, but the issue you brought up is another reason for the Splatoon developers to revise their ridiculous point system! Or perhaps they could match players based on performance in a particular mode rather than overall performance.
Yeah, I don't particularly love the idea of separate ranks, but I'm not opposed to it either. I still think if ranked consists of 3 game modes, the rank should reflect your skill in all 3. Otherwise why have 3 modes? And it's insane to have people able to play high rank in any mode that they couldn't even fight their way out of C in. Exactly as you say, if you're terrible at a given mode or haven't played that mode, how can they seriously through you into matches in that mode at that rank with people that do know what they're doing? You become the one' that's several skill levels behind their rank. It makes no sense with the current system.

BlackZero's point is valid: He hates RM and wouldn't want to be forced to play RM if he's good at the other two games. But, then it lets him play RM which he hates should he decide to play it, at whatever rank level he's in where he can harm other's scores. That's bad. The ideal system would involve mandatory skills on all 3, but if they wanted to reconcile with people in BlackZero's position, the only way to do it is rank them separately.

Absolutely! That's precisely what I observe. Besides the many other things I've observed, I'm currently on my 3rd or 4th cycle of wins and loses that is overall happening the same way each time. Some variability of course, but overall it looks and feels the same and has the same fundamental outcome. I tried to break the losing side of the cycle with some heroics I didn't think I was capable of but the weaknesses of my team were revealed each time. One game recently, after getting the lead in tower control late in the game after I took out the entire opposing team on bluefin depot (after my team was dominated for the entire game), two of my teammates decide to fiddle all around the map instead of defending in the last seconds of the game and we lost. Of course! :)
It was pretty much the same for me. I always cite my 18/12 TC with two triple splats game, I think because that's the one that made it clear to me something is really wrong. I accepted my winning/losing streaks as "maybe it's just me, maybe I suck, maybe I got carried before" - but then I noticed I was usually the best or second best. So then I decided "I'll just win this thing myself!" And play 3x harder than ever, and own the whole map. And I did. And we still lost. Given the brutality of TC, I could have even accepted that things just went wrong and I couldn't land the tower. They'd always be waiting near the goal to take me out even after I finished everyone else (but I got it near the goal every time!) But when my team can't get them off the tower after I get shot off to the point that I have to be the one to shoot them off EVERY TIME they get control, to the point that I had to be the one to get them off when it was ONE POINT away from our goal in addition to being the only one driving it forward, even when I'd virtually cleared the entire map twice. The losing streaks are weird enough. But when you can perform that well and your team is SO bad that it can go THAT badly even when you yourself are tearing the whole map single handedly. And the losing streak continues from there down to B-. It's weird. Then after that I could soar back up, win after win, like I hit the "easy button"

The other part that's weird is TW. I go similarly back and forth with "too easy" and "impossible to leave spawn" rounds. But consistently when it's "impossible to leave spawn" I check and the enemy team is mostly S/S+ (or has a player so good they're clearly an S/S+ alt.) Why would games in the B ranks against players that are mostly B- and B consist of matches as difficult as when I'm pasted in an all S rank room? (And how did I have an S rank team in TW that I performed equally with? (I actually had the best k/d of the team, but I was eliter so that's not unexpected if we're on the same skill level.)

Maybe the root of the problem is just that it passes SO many C- players to ALL other ranks that it's colossally broken. But the continuous pattern of win series/loss series that defies statistical probability sticks out like a sore thumb. Wins and losses simply can not happen primarily in series. And when the win series happen when your team all has positive k/d, and the loss series happen when you have positive k/d and your team has negative k/d (in multiples.) It's hard not to point at the team assembly.
 

Award

Squid Savior From the Future
Joined
Dec 18, 2015
Messages
1,661
I'll just leave these here...


I think once we're to the point of comparisons to games that exist largely for financial gains based on a particular side winning and/or illegal gambling, we're pretty much answering our own questions ;)
 

BlackZero

Inkling Commander
Joined
Nov 3, 2015
Messages
350
I think once we're to the point of comparisons to games that exist largely for financial gains based on a particular side winning and/or illegal gambling, we're pretty much answering our own questions ;)
I don't follow. The video is using an example from Poker, but this phenomenon is rooted in human psychology. The concept of tilt is the same regardless of what game it is or whether money is involved.
 

Award

Squid Savior From the Future
Joined
Dec 18, 2015
Messages
1,661
I don't follow. The video is using an example from Poker, but this phenomenon is rooted in human psychology. The concept of tilt is the same regardless of what game it is or whether money is involved.
Yeah. I'm just not buying into the whole "it's all in your head, the patterns are normal" motif. The win streak/lose streak/win streak/lose streak pattern is NOT normal in any kind of random variation competitive system. I realize you're trying to give the game system the benefit of the doubt, but even in a best case scenario, it still comes down to the game has SUCH a wide variation in player skills within each rank bracket that a useful match does not exist. But even if so, the "streaks" would not be in bursts. And the "streaks" would not change with map rotation or rank change. That is something else.
 

BlackZero

Inkling Commander
Joined
Nov 3, 2015
Messages
350
Yeah. I'm just not buying into the whole "it's all in your head, the patterns are normal" motif. The win streak/lose streak/win streak/lose streak pattern is NOT normal in any kind of random variation competitive system.
It actually is. LoL forums are full of people asking "why am I getting win streaks followed by losing streaks?" The answer in all of them is the same: as you win, you play better opponents until you reach a point where you are outmatched. At that point, you go on a downhill slide until you reach a level where your winning upswing starts again. Whether people care to acknowledge it or not, this cycle seems fairly normal for this type of matchmaking. People aren't getting screwed by anything other than their own winning streaks and good fortune. For what it's worth, my own play experience is consistent with this. When I have win streaks, it's almost always followed by a losing streak. When I have a mix of wins and losses, I don't notice the same dramatic shifts. Make of that what you will.

I suspect Turf Wars isn't purely random, rather stochastic. Let's say someone is shooting at targets. In a truly random system, your shot would have as much chance of hitting a bulls-eye as it would exiting the barrel, turning around, and flying right back into the gun. Rather, a paper target represents a stochastic system: there's no real logic to shot placement, but their grouping is consistent enough that one can perform statistical analysis on it to determine trends.

So, I suspect turf wars are more random than Ranked, but you are still grouped within a specific range of player values. The more you win, the higher that group of player values goes until you eventually find yourself in a value range that is beyond your capabilities. From that point, you suffer a series of losses until you find yourself back in a comfort zone. That's not the game manipulating or screwing anyone. That's a natural by-product of how this ranking system works. I am not saying it's a good system nor am I trying to defend it. Rather, you guys asked how one can go from win streaks to losing streaks within this matchmaking framework. This is the simplest and most common explanation (from what I've seen) that is consistent with how Elo works.

It's very easy for tilt to take over in a losing streak, which will pull you down even further because your quality of play drops. This is why many people in competitive games with Elo matchmaking set a loss threshold. If they lose more X (mine is 5) number of consecutive matches in one session, they take a break. It keeps them from getting frustrated to the point that their play suffers and they prolong the losing streak longer than necessary. People who hang on for at least one win before they quit can easily drop a whole tier because, no matter how much harder they try, they will only get worse.
 

Award

Squid Savior From the Future
Joined
Dec 18, 2015
Messages
1,661
It actually is. LoL forums are full of people asking "why am I getting win streaks followed by losing streaks?" The answer in all of them is the same: as you win, you play better opponents until you reach a point where you are outmatched. At that point, you go on a downhill slide until you reach a level where your winning upswing starts again. Whether people care to acknowledge it or not, this cycle seems fairly normal for this type of matchmaking. People aren't getting screwed by anything other than their own winning streaks and good fortune. For what it's worth, my own play experience is consistent with this. When I have win streaks, it's almost always followed by a losing streak. When I have a mix of wins and losses, I don't notice the same dramatic shifts. Make of that what you will.
Well, LoL makes poker look clean... o_O When looking for dirty money in video gaming, there's no better place to start...

I see your point, and if we were talking about individual skill matchmaking and a point system not so dreadful what you describe is exactly how I'd expect a floating rank system to work. you'd reach your cap then lose then win, lose etc. The idea would be you'd settle into your rank by being matched within your skill bracket by losing to the stronger players and winning to the weaker ones. But because of all the things we've already discussed, that's not what happens here. You end up losing to weaker players and winning to weaker players and being prevented from entering the actual skill bracket that provides the correct level of skill and challenge.

In addition to the bad use of Elo for teams, and the bad scoring system overall, I'd say a fluid/floating rank system is ALSO inherently bad for team games. the progressive ladder Holidaze discribed would be a better fit for a team game to ensure teams of a given rank are indeed of the same skill more or less. The floating system by design muddles the ranks, which makes the system that much worse.

But the oddness of the win/lose streak pattern here is that you're not being defeated by superior opponents but by weaker ones not because you were directly defeated as in LoL, but because of scoring technicalities. And the fact that you can "tilt toward victory!" When you go into a losing streak, the LoL logic suggests you're tilting to your own doom. But in Splatoon you tilt to a magic point, and then you get fixed on a winning streak. Tilting toward doom (or victory) suggests that the pattern wouldn't change during a map rotation, or a rank change on the same map, but it does. Also in TW (with or without a wider swing.)

And when you have a mix of wins and losses, don't look at your own results, and look at both teams. Versus the massive imbalances in the winning streaks you'll likely find the results of all players to be closely grouped: I.E. you were given good team mixes.


I suspect Turf Wars isn't purely random, rather stochastic. Let's say someone is shooting at targets. In a truly random system, your shot would have as much chance of hitting a bulls-eye as it would exiting the barrel, turning around, and flying right back into the gun. Rather, a paper target represents a stochastic system: there's no real logic to shot placement, but their grouping is consistent enough that one can perform statistical analysis on it to determine trends.

So, I suspect turf wars are more random than Ranked, but you are still grouped within a specific range of player values. The more you win, the higher that group of player values goes until you eventually find yourself in a value range that is beyond your capabilities. From that point, you suffer a series of losses until you find yourself back in a comfort zone. That's not the game manipulating or screwing anyone. That's a natural by-product of how this ranking system works. I am not saying it's a good system nor am I trying to defend it. Rather, you guys asked how one can go from win streaks to losing streaks within this matchmaking framework. This is the simplest and most common explanation (from what I've seen) that is consistent with how Elo works.
Again it comes back to being very broken. It makes sense that the more you win you should face tougher enemies. But what it appears is if you win the easy rooms, it just throws the best at you. If you lose to them, it sends you back to the noobs. (TW) You might pass through the midsection momentarily before going back to the best. It's weird.

If win/loss streaks are THAT common in Elo, it makes me question Elo as a whole, tried and true though it might be, it produces highly unnatural results from a competition system that is designed to match skills. I don't know if Elo itself is flawed, or if all implementations of it outside chess just don't work.
 

Manwell

Inkling
Joined
May 11, 2015
Messages
9
NNID
Manwell
I think the ranked system is incredibly fair and balanced when you compare it to other games such as League of Legends where if you start to even out winning and losing, the system traps you in a hole where wins give you barely anything until you manage to go on a big winning streak. In Splatoon, its much more forgiving and with a couple of wins you can recover losses easily. Note: i am talking about the points system and ranking and not about matchmaking. That's a whole other thing.
 

jsilva

Inkling Cadet
Joined
Oct 30, 2015
Messages
262
For what it's worth, my own play experience is consistent with this. When I have win streaks, it's almost always followed by a losing streak. When I have a mix of wins and losses, I don't notice the same dramatic shifts. Make of that what you will.
I appreciated the videos, thanks BlackZero. When I first started experiencing ridiculous losing streaks I came across people saying not to waste the energy looking at anything but yourself, how you can improve, and I did begin taking that advice. I am definitely a much better player now vs. then. I'm always on the lookout for ways that I let my team down, in-game and afterwards.

In the example I gave above, my first reaction to that annoying loss was a wish that I had played better in the end and hadn't been splatted by the Luna Blaster from the tower because that left my poor sniper vulnerable. But the reality is that when I see my team getting completely dominated, completely unable to push the tower any signifiant amount (except when I pulled off better-than-I-normally-am heroics), three games in a row, in a game where I'm on the significantly higher-ranked team (mostly S's vs. A+ and A), and I lose 7 points each game, and all of this in the midst of a number of repeated losses, that's just one of many examples where something doesn't seem right. Why was I even matched three games in a row on pretty much the exact same rank imbalance but with mostly different players?

Also, my experience doesn't match yours. My winning streaks often happen after substantial losing streaks. Consider the time I mentioned above when my youngest son was losing. I knew he wasn't playing very well, but what's the sense in me losing something like 10 games in a row at that rank? And then after getting down to B+ 47 I won most of the games back up to A- 54?
 

BlackZero

Inkling Commander
Joined
Nov 3, 2015
Messages
350
If win/loss streaks are THAT common in Elo, it makes me question Elo as a whole, tried and true though it might be, it produces highly unnatural results from a competition system that is designed to match skills. I don't know if Elo itself is flawed, or if all implementations of it outside chess just don't work.
Again, Elo doesn't measure skill; only probabilities of winning. You'll get no argument from me that this is not a great system for team battle games because it doesn't factor individual ability or effort. This is why things like TrueSkill were developed. TS measures several different stats that correlate with higher skill, and uses that to run several calculations to decide not just how teams will stack up, but also how the individuals on both teams compare to each other. It still aims for a 50/50 match, but pays more attention to each player on a team rather than considering them as one single entity.

Why was I even matched three games in a row on pretty much the exact same rank imbalance but with mostly different players?
Because not all S rank players are created equal. Even within a single rank, you're going to get a wide range of actual skill and experience levels, favorable and infavorable weapons match-ups, and possibly other players who are experiencing tilt. These are things that the matchmaker won't consider simply because it can't. It can only match people based on their player values. If those values fall within a specific range, regardless of how they got it or whether they deserve it, the matchmaker will include them in the pool of people it draws from.

As for rank imbalance, I've already explained this. It shoots for close to a 50/50 match. If it has to play fast and loose with ranks to get there, it will. Also bear in mind that 50/50 doesn't mean you'll win one and lose one. You could easily chain together a streak of 10 wins, then hit a wall and lose for the next 5, then win another 3, etc. Seriously, I'm not pulling this out of thin air. Browse some LoL forums and you will see threads where people make posts that are almost word-for-word verbatim of yours right down to people accusing the matchmaker system of working against them. It's not some conspiracy, though. It's just how Elo matchmaking works in team battle games. You'll get good match-ups and bad ones, you'll get winning streaks and losing streaks. There's nothing strange about it; It's just the nature of the beast as I have explained here and countless others on LoL forums and subreddits have explained there.
 

97Stephen

Pro Squid
Joined
Sep 27, 2015
Messages
146
Location
Wisconsin
Again, Elo doesn't measure skill; only probabilities of winning. You'll get no argument from me that this is not a great system for team battle games because it doesn't factor individual ability or effort. This is why things like TrueSkill were developed. TS measures several different stats that correlate with higher skill, and uses that to run several calculations to decide not just how teams will stack up, but also how the individuals on both teams compare to each other. It still aims for a 50/50 match, but pays more attention to each player on a team rather than considering them as one single entity.



Because not all S rank players are created equal. Even within a single rank, you're going to get a wide range of actual skill and experience levels, favorable and infavorable weapons match-ups, and possibly other players who are experiencing tilt. These are things that the matchmaker won't consider simply because it can't. It can only match people based on their player values. If those values fall within a specific range, regardless of how they got it or whether they deserve it, the matchmaker will include them in the pool of people it draws from.

As for rank imbalance, I've already explained this. It shoots for close to a 50/50 match. If it has to play fast and loose with ranks to get there, it will. Also bear in mind that 50/50 doesn't mean you'll win one and lose one. You could easily chain together a streak of 10 wins, then hit a wall and lose for the next 5, then win another 3, etc. Seriously, I'm not pulling this out of thin air. Browse some LoL forums and you will see threads where people make posts that are almost word-for-word verbatim of yours right down to people accusing the matchmaker system of working against them. It's not some conspiracy, though. It's just how Elo matchmaking works in team battle games. You'll get good match-ups and bad ones, you'll get winning streaks and losing streaks. There's nothing strange about it; It's just the nature of the beast as I have explained here and countless others on LoL forums and subreddits have explained there.
I've got to ask, is there any way to "beat" an Elo system? Not cheating, but a way to help alleviate the problems? I'm wondering if I could try something to to see if it improves my experience. I know there's know replacement for skill, but a better overall experience would be good.
 

BlackZero

Inkling Commander
Joined
Nov 3, 2015
Messages
350
I've got to ask, is there any way to "beat" an Elo system? Not cheating, but a way to help alleviate the problems? I'm wondering if I could try something to to see if it improves my experience. I know there's know replacement for skill, but a better overall experience would be good.
That depends on what you mean by "beat." If you want to prevent losing streaks, the best advice is to set up the loss threshold I mentioned earlier. Basically, make yourself quit playing after so many losses in a row (3 is a pretty good number, though some LoL players go as far as quitting after 2 consecutive losses) and come back to it after an hour or so. This will keep your sigma under control as you won't go from a 10 win streak to a 10 loss streak and back again. It will also help control your descent and keep your tilt in check, as you won't be taking a nosedive with a huge losing streak. If you're losing, ditch the "I'll win the next one" attitude and minimize the damage by accepting that you're having a bad day.

As for how you can get yourself better matches, there's no surefire way to do this. Elo doesn't measure skill, so there's no way to play the system to guarantee you get skilled teammates and opponents other than to play with squad mates that you know are on your level.
 

97Stephen

Pro Squid
Joined
Sep 27, 2015
Messages
146
Location
Wisconsin
That depends on what you mean by "beat." If you want to prevent losing streaks, the best advice is to set up the loss threshold I mentioned earlier. Basically, make yourself quit playing after so many losses in a row (3 is a pretty good number, though some LoL players go as far as quitting after 2 consecutive losses) and come back to it after an hour or so. This will keep your sigma under control as you won't go from a 10 win streak to a 10 loss streak and back again. It will also help control your descent and keep your tilt in check, as you won't be taking a nosedive with a huge losing streak. If you're losing, ditch the "I'll win the next one" attitude and minimize the damage by accepting that you're having a bad day.

As for how you can get yourself better matches, there's no surefire way to do this. Elo doesn't measure skill, so there's no way to play the system to guarantee you get skilled teammates and opponents other than to play with squad mates that you know are on your level.
I guess I'll have try that. It's good advice, I have kept playing to get that one win which probably didn't help my streak. If I lose that much I'll play tw I guess. It's a pity though that there isn't any way (if it's an Elo system) to help alleviate the root of the problem.
 

jsilva

Inkling Cadet
Joined
Oct 30, 2015
Messages
262
Because not all S rank players are created equal. Even within a single rank, you're going to get a wide range of actual skill and experience levels, favorable and infavorable weapons match-ups, and possibly other players who are experiencing tilt. These are things that the matchmaker won't consider simply because it can't. It can only match people based on their player values. If those values fall within a specific range, regardless of how they got it or whether they deserve it, the matchmaker will include them in the pool of people it draws from.

As for rank imbalance, I've already explained this. It shoots for close to a 50/50 match. If it has to play fast and loose with ranks to get there, it will. Also bear in mind that 50/50 doesn't mean you'll win one and lose one. You could easily chain together a streak of 10 wins, then hit a wall and lose for the next 5, then win another 3, etc. Seriously, I'm not pulling this out of thin air. Browse some LoL forums and you will see threads where people make posts that are almost word-for-word verbatim of yours right down to people accusing the matchmaker system of working against them. It's not some conspiracy, though. It's just how Elo matchmaking works in team battle games. You'll get good match-ups and bad ones, you'll get winning streaks and losing streaks. There's nothing strange about it; It's just the nature of the beast as I have explained here and countless others on LoL forums and subreddits have explained there.
Yes, I understand this. The examples I bring up all have an explanation when considered individually, and if they occurred individually and occasionally I wouldn't think anything of it. However, in the actual real-life context the explanations don't make as much sense.

Even when there is an explanation (which usually includes an assumption or two), it isn't always sufficient. Such as my example of playing on my son's account. In light of the fact that I got on a winning streak with no change of mood, saying I was experiencing tilt doesn't work very well, so then the only real but very generic explanation is to say the (assumed) elo matching system matched me repeatedly on the substantially weaker team for 10 or so games in a row and then stopped doing that, for whatever reason. That sounds very problematic at best. Not to mention that the dude in the video seemed to be saying the elo system is good and well adapted for multiplayer games.

And in reference to the three games I brought up, I understand your perspective but it leaves some things unanswered. I lost 7 points for those games so the system perceived an imbalance. So if it perceived an imbalance, why did it deliberately match me pretty much with an identical imbalance three games in a row from a mostly different player pool? Was there really not a better option? And what's the likeliness that different player pools of higher ranked players in three consecutive games were notably weaker than different player pools of lower ranked players? While there's certainly a possible explanation for this, when you consider it in context the explanation is less convincing. The context is basically a number of other 'highly coincidental' circumstances that lead to repeated losses that are nearly unavoidable.

I realise there is a history of these kinds of complaints, to the degree that there are those videos you linked to. But not everyone drawing conclusions about the matching systems is an emotionally charged non-thinking person :)
 
Last edited:

Award

Squid Savior From the Future
Joined
Dec 18, 2015
Messages
1,661
That depends on what you mean by "beat." If you want to prevent losing streaks, the best advice is to set up the loss threshold I mentioned earlier. Basically, make yourself quit playing after so many losses in a row (3 is a pretty good number, though some LoL players go as far as quitting after 2 consecutive losses) and come back to it after an hour or so. This will keep your sigma under control as you won't go from a 10 win streak to a 10 loss streak and back again. It will also help control your descent and keep your tilt in check, as you won't be taking a nosedive with a huge losing streak. If you're losing, ditch the "I'll win the next one" attitude and minimize the damage by accepting that you're having a bad day.

As for how you can get yourself better matches, there's no surefire way to do this. Elo doesn't measure skill, so there's no way to play the system to guarantee you get skilled teammates and opponents other than to play with squad mates that you know are on your level.
What you're saying about sigma makes sense, but if we do ignore tilt, that I think with the methodologies most of us are using is not really a primary factor here, but not just for Splatoon, but for any game, the way to prevent the problem is to stop playing until you get a magical winning team is just so bad I don't even know what to say about it. (Though it also does confirm a lot when it's so easy, after you drop a rank to just pick right up and go on a winning streak, that the issue is not tilt and is matchmaker related.)

A fun experiment none of us have the time or patience to do but I'm sure someone on Youtube would love to do it: Play NON STOP. Don't change lobbies unless dc'd, just keep playing and watch the trends through the rank changes.

Because not all S rank players are created equal. Even within a single rank, you're going to get a wide range of actual skill and experience levels, favorable and infavorable weapons match-ups, and possibly other players who are experiencing tilt. These are things that the matchmaker won't consider simply because it can't. It can only match people based on their player values. If those values fall within a specific range, regardless of how they got it or whether they deserve it, the matchmaker will include them in the pool of people it draws from.

As for rank imbalance, I've already explained this. It shoots for close to a 50/50 match. If it has to play fast and loose with ranks to get there, it will. Also bear in mind that 50/50 doesn't mean you'll win one and lose one. You could easily chain together a streak of 10 wins, then hit a wall and lose for the next 5, then win another 3, etc. Seriously, I'm not pulling this out of thin air. Browse some LoL forums and you will see threads where people make posts that are almost word-for-word verbatim of yours right down to people accusing the matchmaker system of working against them. It's not some conspiracy, though. It's just how Elo matchmaking works in team battle games. You'll get good match-ups and bad ones, you'll get winning streaks and losing streaks. There's nothing strange about it; It's just the nature of the beast as I have explained here and countless others on LoL forums and subreddits have explained there.
I think the point jsilva is trying to make though is, if this were random error by Elo not able to great a great match with the lobby selection available, how is it possible to end up with lobby after lobby of the same exactly type of Elo error / pool for elo to select from to create the same type of mismatched team? We're focusing on the problems with Elo, but does that not imply the most specific problem here might be the lobby creator since it shows a habbit of having streaks of "4 mid teir, 1 top tier, 3 scrubs" lobby after lobby with some consistency? The problem with Elo would assemble the same faulty team reliably given a lobby of the same bad mix-up for all the reasons you've explained. But maybe the top question should be: Why is the lobby creator creating the same kind of bad lobby again and again, especially in light that it should be designed with elo's limitations in mind? Alternately how is it in "streaks" that you get lobby after lobby of the same deeply mismatched group, but then get streaks of lobbies that have a more balanced pool to pick from? Should it not be more random if it were picking randomly from the pool? Or is it picking a defined set? Maybe we've been focusing on the wrong broken element.
 

Award

Squid Savior From the Future
Joined
Dec 18, 2015
Messages
1,661
Even when there is an explanation (which usually includes an assumption or two), it isn't always sufficient. Such as my example of playing on my son's account. In light of the fact that I got on a winning streak with no change of mood, saying I was experiencing tilt doesn't work very well, so then the only real but very generic explanation is to say the (assumed) elo matching system matched me repeatedly on the substantially weaker team for 10 or so games in a row and then stopped doing that, for whatever reason. That sounds very problematic at best. Not to mention that the dude in the video seemed to be saying the elo system is good and well adapted for multiplayer games.
And I've experienced the same, what I called "tilting to victory" above. A time when I started at B47 or so, and from the start, went on a massive losing streak down to B-40 due to these imbalanced teams (that's the session where my oft mentioned 18/12 "I can't possibly play this thing any harder" game and the "we got spawncamped in TC" rounds came from) If there was ever a "tilt" it was that session. And then magically at B-40 the switch flipped, and I went on a massive winning streak with a few losses here and there, all the way to B80 (I went down to B60 after map/mode change to RM but that wasn't the matchmaker, that was me using a weapon in a mode/map combo I'd never played before and got it wrong.) If I was tilting to frustration and burnout ever, it would have happened in that dreadful sequence. And yet I went on an epic winning streak by simply NOT stopping no matter what. That shouldn't happen based on all the theories. I shouldn't "tilt" to a higher place than I started (easily I might add.) And that's what's weird. I had to play super-mega-tryhard through the whole losing streak. And then the winning streak played EASY. But I was still the team leader!? And the other team was not filled with a bunch of 0/7 players when they lost like mine was in the beginning.

Tilting to victory.

And in the video, they may be saying that Elo is good for multiplayer games, but we're talking about an Elo that doesn't actually score YOU, but averages you and 3 other people to be "the player." It's never actually matchmaking YOU at all. It's matchmaking a fictitious player made up from an average player value of 4 random players. The problem would thus be that Elo is fine for multiplayer, Elo is not fine for TEAM multiplayer where the feedpool is not pooled based on individual merit.

If I'm evaluating DarkZero's explanations properly the idea is:
  1. Lobby builder takes 8 people from an undefined pool of a given playstyle/rank and suitable internet connection and assigns them a lobby. (*This here might be the key, I've been thinking.)
  2. Elo creates a list of "team values" for all permutations of 4 players from the list of 8.
  3. Elo creates a list of "win probabilities" from all permutations of crossing the team values of teams that don't overlap players from the list above.
  4. Elo assembles the two closest "win probabilities" teams from the above calculation.
  5. Rage.

And in reference to the three games I brought up, I understand your perspective but it leaves some things unanswered. I lost 7 points for those games so the system perceived an imbalance. So if it perceived an imbalance, why did it deliberately match me pretty much with an identical imbalance three games in a row from a mostly different player pool? Was there really not a better option? And what's the likeliness that different player pools of higher ranked players in three consecutive games were weaker than different player pools of lower ranked players? While there's certainly a possible explanation for this, when you consider it in context the explanation is less convincing. The context is basically a number of other of 'highly coincidental' circumstances that lead to repeated losses that are nearly unavoidable.
+1

The more I think about DarkZero's explanations, the LoL scenarios, your experiences, and my experiences, the more I feel like we're questioning the wrong broken part of the system. They're all broken ,but our suspicions are focusing around Elo, while ignoring that Elo is being given. Given an appropriately disproportionate lobby of 8, Elo would consistently and reliably pair the 8 in the same general way. If you're the best of the 8, and there are 2-3 that suck, you'd be reliably paired with the sucky ones, time and time again. I.E. If your calculated player value places you as one of the best in yoru rank, any time you're in a lobby containing people among the worst in their rank, you will always be paird with them, predictably by Elo.
Elo is still a disastrously bad choice, to be sure, but it would be predictably bad.

What we're not questioning sufficiently, probably because we have the fewest answers for, is the LOBBY BUILDER. If we are getting predictable Elo losing pairings (until we don't), then that implies it is the LOBBY BUILDER providing the 8 players to Elo that are guaranteed to lead to a bad match, whether it's by defect or conspiracy. Our real question, now that we've likely identified Elo in the team pairings and why it works as it does, should probably change focus to determining how and why the lobby builder works as it does, how it divides "playstyle" and whatever other pools it may have. And/or if there is an agenda in the lobby builder, building lobbies it knows Elo will by nature match a certain way and generate a particular outcome.

I realise there is a history of these kinds of complaints, to the degree that there are those videos you linked to. But not everyone drawing conclusions about the matching systems is an emotionally charged non-thinking person :)
Again, I think once we're comparing things to LoL we're already pretty deep in a hole. LoL is effectively a video gambling machine that lets you play zero stakes games for free. It exists for gambling. Riot games is more or less a gambling equipment company without having to define itself and license itself as such. Would you trust a Bally video poker machine to play clean? Would you trust a Konami Pachinko machine to play clean? Would you trust your state-run lotto to play clean? Similarly should you view a Riot game, specifically LoL. It's not that LoL can not be fun, just as a Pachinko machine or video poker machine can be fun. Each of these is worth playing if it appeals to you and is a perfectly valid form of entertainment to play. But there should be no illusions about exactly WHAT you're playing.

Thus, if we're down to comparing Splatoon's scoring to what is effectively a Pachinko machine, it's an indication of the problem without even needing an answer ;)
 

BlackZero

Inkling Commander
Joined
Nov 3, 2015
Messages
350
Even when there is an explanation (which usually includes an assumption or two), it isn't always sufficient. Such as my example of playing on my son's account. In light of the fact that I got on a winning streak with no change of mood, saying I was experiencing tilt doesn't work very well, so then the only real but very generic explanation is to say the (assumed) elo matching system matched me repeatedly on the substantially weaker team for 10 or so games in a row and then stopped doing that, for whatever reason. That sounds very problematic at best.
You aren't going to experience tilt on a winning streak. Tilt is exclusive to losing streaks due to how our minds perceive losing. More info here and here, if you are still uncertain as to what tilt and amygdala hijack are. As I've explained before, a winning streak propels you higher in the ranking than your actual skill should place you. A subsequent losing streak is the system "rebalancing," i.e. pushing you to a position that's closer to where you actually need to be.

Not to mention that the dude in the video seemed to be saying the elo system is good and well adapted for multiplayer games.
I disagree with the video on that note. Elo isn't very good at team based games. With that said, I think he was trying to tell people to stop using it as an excuse for losing rather than saying saying it was an ideal system.

And in reference to the three games I brought up, I understand your perspective but it leaves some things unanswered. I lost 7 points for those games so the system perceived an imbalance. So if it perceived an imbalance, why did it deliberately match me pretty much with an identical imbalance three games in a row from a mostly different player pool? Was there really not a better option? And what's the likeliness that different player pools of higher ranked players in three consecutive games were notably weaker than different player pools of lower ranked players? While there's certainly a possible explanation for this, when you consider it in context the explanation is less convincing. The context is basically a number of other 'highly coincidental' circumstances that lead to repeated losses that are nearly unavoidable.
First, unless a player has ESP or can look at the objective game data of the players they faced, there's no way a player can determine whether they were at an identical disadvantage in any two matches. It may feel that way, and one may believe that to be the case, but what a person thinks and what is objectively going on are two very different things. This is why I take any anecdotal experience with a grain of salt: people are biased and operate with limited information. Unless you can prove that you had the same disadvantage in all three matches without simply saying it happened that way, I have to base this on a few things. First, rank =/= skill. They may be higher or lower rank than you, but they may also be experiencing a losing streak or winning streak that placed them above or below their actual ability. So, higher or lower ranked teammates is not a guarantee of anything with regards to how players actually perform.

Second, the game calculates a 50/50 win ratio. There's nothing impossible about losing three games in a row with 50/50 odds, especially if a win streak put you higher in the ranks than you actually need to be. There's nothing impossible about losing three times in a row with 50/50 odds period. If your son's account was C rank or B rank, I'd give this more weight. It sounds like his account is also on the upper level of ranks (A-S). It's entirely possible for someone S ranked to get put in an A rank lobby after a losing streak.

I realise there is a history of these kinds of complaints, to the degree that there are those videos you linked to. But not everyone drawing conclusions about the matching systems is an emotionally charged non-thinking person :)
I'm not saying you guys are. I'm saying that losing streaks have been objectively shown to have a negative impact on people's performance. That's a statistical fact. Unless you guys are going to tell me that 10 losses in a row have no effect whatsoever on you, tilt is a factor regardless of how disciplined you are. There's a reason why top tier competitive LoL players stop playing after 2-3 consecutive losses. I've also noticed less dramatic shifts in the matches I get when I do the same. It doesn't matter how good or level-headed a player thinks they are, tilt is a fact of human psychology that has been well documented in competitive games.

If I'm evaluating DarkZero's explanations properly the idea is:
  1. Lobby builder takes 8 people from an undefined pool of a given playstyle/rank and suitable internet connection and assigns them a lobby. (*This here might be the key, I've been thinking.)
  2. Elo creates a list of "team values" for all permutations of 4 players from the list of 8.
  3. Elo creates a list of "win probabilities" from all permutations of crossing the team values of teams that don't overlap players from the list above.
  4. Elo assembles the two closest "win probabilities" teams from the above calculation.
  5. Rage.
Yep. That's pretty much it. I think people get screwed when the game tries to force a 50/50 match. Win streaks that launch people into higher ranks than their performance can sustain makes this even worse.

What we're not questioning sufficiently, probably because we have the fewest answers for, is the LOBBY BUILDER. If we are getting predictable Elo losing pairings (until we don't), then that implies it is the LOBBY BUILDER providing the 8 players to Elo that are guaranteed to lead to a bad match, whether it's by defect or conspiracy. Our real question, now that we've likely identified Elo in the team pairings and why it works as it does, should probably change focus to determining how and why the lobby builder works as it does, how it divides "playstyle" and whatever other pools it may have. And/or if there is an agenda in the lobby builder, building lobbies it knows Elo will by nature match a certain way and generate a particular outcome.
The simplest explanation is that both lobby builders have a rank system (it's just hidden in TW). Their player values go up when then win and down when they lose. When their value reaches a certain point, they get grouped with a certain tier of player. This is why win and losing streaks are so dangerous: a win streak can launch you through a couple of tiers until you wind up at a level that your actual abilities can't sustain. At that point, you go on a losing streak until you drop back to a level you can either sustain or climb out of. If you drop too far, the players will be well beneath your ability and you'll go back on a win streak and the whole process repeats. This is why it may appear deliberate: you bounce between two different extremes in a predictable fashion.

LoL is effectively a video gambling machine that lets you play zero stakes games for free. It exists for gambling. Riot games is more or less a gambling equipment company without having to define itself and license itself as such. Would you trust a Bally video poker machine to play clean? Would you trust a Konami Pachinko machine to play clean? Would you trust your state-run lotto to play clean? Similarly should you view a Riot game, specifically LoL. It's not that LoL can not be fun, just as a Pachinko machine or video poker machine can be fun. Each of these is worth playing if it appeals to you and is a perfectly valid form of entertainment to play. But there should be no illusions about exactly WHAT you're playing.
The difference is, the owner of those machines has an invested interest in winning. What does Riot Games care if you lose? It's not like you're betting real money that they can take if you lose a match. Same thing with Splatoon: there's no gain for engineering a biased system. It's actually against their interests to create a game that screws people because people won't spend money on it. Unless you are suggesting that you guys are on Nintendo's hit list for some reason and they are willing to throw money away on a game just so they can screw you over with biased matcmaking, I truly don't see the incentive for a "the house always wins" system. Why would it target you? Do the random opponents you play with that you have never met before and will probably never see again bribe Nintendo to screw you over? I really don't understand this line of thinking, but many people here seem to think that the only way they can lose multiple times is if someone engineers the game to sabotage them. In that case, I really don't what to say other than Nintendo has better things to do than to make sure a handful of people who bought their game get taken down a peg on a regular basis.
 

Award

Squid Savior From the Future
Joined
Dec 18, 2015
Messages
1,661
You aren't going to experience tilt on a winning streak. Tilt is exclusive to losing streaks due to how our minds perceive losing. More info here and here, if you are still uncertain as to what tilt and amygdala hijack are. As I've explained before, a winning streak propels you higher in the ranking than your actual skill should place you. A subsequent losing streak is the system "rebalancing," i.e. pushing you to a position that's closer to where you actually need to be.
I'm not saying you guys are. I'm saying that losing streaks have been objectively shown to have a negative impact on people's performance. That's a statistical fact. Unless you guys are going to tell me that 10 losses in a row have no effect whatsoever on you, tilt is a factor regardless of how disciplined you are. There's a reason why top tier competitive LoL players stop playing after 2-3 consecutive losses. I've also noticed less dramatic shifts in the matches I get when I do the same. It doesn't matter how good or level-headed a player thinks they are, tilt is a fact of human psychology that has been well documented in competitive games.
The only issue with that is the example that both I and jsilva cited the losing streak or "tilt" led us directly into winning streaks instead of prolonging the losing streak. Meaning it wasn't tilt at all.

I'm also unsure about the "rebalancing." While it makes sense that it's "rebalancing" that doesn't really explain the people that can need "rebalancing" out of the same rank again and again. Or that "are propelled too far" over and over again.

If you're right, and that happens, to that degree, that still goes back to the point that the system in this game and/or others is so horribly, utterly broken as to serve almost no purpose to begin with. If an elo system is so capable of bouncing between solid blocks of extremes without ever settling on placing a player at their appropriate bracket, then elo itself demonstrates repeated failure in its intended purpose and must be discarded as an invalid solution. If it does that purely because it's being wrongly used for team matching, then we're back to the original topic that the ranking system needs to be tweaked or overhauled.

Yep. That's pretty much it. I think people get screwed when the game tries to force a 50/50 match. Win streaks that launch people into higher ranks than their performance can sustain makes this even worse.
I can agree with the latter in terms of myself being apparently propelled too high in TW, and due to my teammates being propelled too high in ranked. Losing streaks aside, I can safely say I have yet to be propelled too high in ranked as I've yet to encounter superior opponents there (despite the losing streaks.) I know superior opponents, I fight them in TW on a regular basis. I can say for sure I haven't encountered them in Ranked, or rather only encountered a few who were obviously destined (or alts) for top ranks. ;)

The simplest explanation is that both lobby builders have a rank system (it's just hidden in TW). Their player values go up when then win and down when they lose. When their value reaches a certain point, they get grouped with a certain tier of player. This is why win and losing streaks are so dangerous: a win streak can launch you through a couple of tiers until you wind up at a level that your actual abilities can't sustain. At that point, you go on a losing streak until you drop back to a level you can either sustain or climb out of. If you drop too far, the players will be well beneath your ability and you'll go back on a win streak and the whole process repeats. This is why it may appear deliberate: you bounce between two different extremes in a predictable fashion.
Makes sense, but the part I'm stuck with is how within the same session, even entering multiple lobbies, I seem to be grouped with the top tier players (I mean, really, lobbies with multiple S+??) across several lobbies. And then I'll get grouped with the lv5-25's. I have to hold back and NOT splat them. And then a few rounds later I'm back in the invisible S ranks. If it's propelling me too far, it does so pretty fast, and if it "rebalance me down" it seems to do it pretty slow and then just drops the floor out.

Either way, even if it's happening the way you say, a system that bounces between two extremes rather than ever settling on the intended goal of matching equals is a devastating failure from any angle you view it. It is unable to achieve its primary goal.

The difference is, the owner of those machines has an invested interest in winning. What does Riot Games care if you lose? It's not like you're betting real money that they can take if you lose a match. Same thing with Splatoon: there's no gain for engineering a biased system. It's actually against their interests to create a game that screws people because people won't spend money on it. Unless you are suggesting that you guys are on Nintendo's hit list for some reason and they are willing to throw money away on a game just so they can screw you over with biased matcmaking, I truly don't see the incentive for a "the house always wins" system. Why would it target you? Do the random opponents you play with that you have never met before and will probably never see again bribe Nintendo to screw you over? I really don't understand this line of thinking, but many people here seem to think that the only way they can lose multiple times is if someone engineers the game to sabotage them. In that case, I really don't what to say other than Nintendo has better things to do than to make sure a handful of people who bought their game get taken down a peg on a regular basis.
The machine maufacturers don't have a vested interest in the machines (except when they're installed in their own casinos.) Their customers who buy the machines do. Further, Konami's pachinko machines should have no vested interest at all since gambling is illegal in Japan....riiight? ;) Riot is the machine manufacturer in this case. LoL isn't a normal video game. It's made for torunaments and *nudge nudge, wink, wink* but of course not gambling *nudge, wink*. Yes you can play it if you're not participating in that. I can import a pachislot and play it as much as I want in my home, too. But it's made with an eye toward customers who do whish to participate (and, lo, they manage their own tournaments, like Bally's manages their own video poker machines.) There's real money at the end of LoL and the heart of the games Riot makes in a way that is not true for most games.

I'm not implying Riot's doing anything really wrong or abnormal in that industry, they're as clean as the big names in that arena. I'm just saying the games Riot makes are not of the same spirit as what we expect from an E3 class game. It's a really different thing.

LOL, I don't think Nintendo is "targeting" anyone specifically. I think, if there is a house always wins system, it can apply to anyone and everyone within certain thresholds when they need to grow or shrink a given lobby population to keep the different lobby groups sufficiently populated in a low population environment. Not targeted sabotage, just "hmm, S now has 20% more population than A+ and A+ is experiencing an increase in wait times, lets balance the S matches to move about 8% of the population down so the A+ lobby doesn't get too barren.
 

Users who are viewing this thread

Top Bottom