Ranked point system MUST change!

[jsilva]

I have a lot of salt about the point system in ranked, but today I observed the worst ever.

I was watching someone playing splat zones and they lost two teammates, one right away and the other less than a minute after the game started. The opposing team exclusively held the zone to the 40s, and then this person's 2-player team managed to get control and hold it (off and on) down to 16. They ended up losing but they played really well! The person in question had a 20/4 K/D ratio.

Still this person lost 12 points! (A rank)

Ridiculous.

 

[Ryuji]

The amount of points lost when you or your teammates dc needs to change, not points in general. Problem is it can't really be fixed because Nintendo has no way of determining whether it was a genuine dc or a ragequit. Still though, I can understand the frustration. I tend to dc when it's pretty clear I'm going to win a match but never dc when it's safe to say I'm going to lose it, basically getting robbed of points in the former situation where I could have gained some. Really irksome.

 

[Redx115]

Man, I know how that feels. I get so mad when I lose a teammate 30 seconds into a match just to lose a ton of points. I agree with the post above me, though, Nintendo has no way of knowing if it was a genuine dc or not. Sucks.

 

[sammich]

it doesnt matter if it was genuine or not, though. it's still screwing over the people you're playing with.

why not make it time based? maybe the first ten-thirty seconds = just a few points lost, after a minute = more, etc. depending on the mode, they can link it to enemy team progression? like if it happens AFTER the enemy gets to 50p it's pretty much normal, but if it's earlier before that you get a slight cut.

 

[jsilva]

it doesnt matter if it was genuine or not, though. it's still screwing over the people you're playing with.

I agree. It doesn't matter why the teammate disconnected (in ranked, at least). Is it reasonable to penalise players for losing a 2vs4, especially when those two players managed to push the objective 83 points? It's completely unthinking on Nintendo's part. It wouldn't be difficult for the system to monitor enough metrics to make a better decision in a game such as that.

The amount of points lost when you or your teammates dc needs to change, not points in general.

I'd disagree with this ... there are many times when the point system is ridiculous! :) For one, you can never be completely sure how many points you'll gain or lose, and it's not weird to lose/win a certain amount of points in one game and win/lose a different amount in a game with no discernible difference or circumstance.

 

[Varonth]

I agree. It doesn't matter why the teammate disconnected (in ranked, at least). Is it reasonable to penalise players for losing a 2vs4, especially when those two players managed to push the objective 83 points? It's completely unthinking on Nintendo's part. It wouldn't be difficult for the system to monitor enough metrics to make a better decision in a game such as that.

There is always a simple way to fix that sort of issue. If someone discconects/ragequits, that person will get all the point penalties of his team, except when playing in a party. If you are playing in a party, the point penalty is split among the party members. Non party members will be save if part of a party leaves the match for whatever reason.

A person with regular disconnects should be playing ranked anyway (as it will cause trouble for his randomly assigned teammates). An odd disconnect every once in a blue moon will have no relevant impact on the rank. Rage quitters will basically stay in C-. Parties aren't save in a way that they get a dummy player who disconnects so everyone else will keep their rank points.

 

[binx]

If someone discconects/ragequits, that person will get all the point penalties of his team

There is no easy solution. Someone having one disconnection every 10 games, which is pretty bad, would lose way too many points and would basically play most of his game with and against weaker people. I feel they should have the right to play like everyone. Sure, it's bad when they fall in your team, cause you'll probably lose. But it's ok, as they disconnected, you won't see him again before long, so it doesn't affect you that much. Plus, if you change the way you count points for the losers, why not for the easy winners? They were 4v3, why not removing their points then?

Anyway, with time, the number of disconnections of your allies and the number of disconnections of your opponents tend to be equal. If you're thinking "I don't have luck, disconnections always happen to my teammates and never to my opponents", then you either didn't play much and had bad luck, or just didn't see it. Among all these easy wins you get, sometimes, it's because they were actually three.

I would say you need to lose points when you disconnect, to prevent ragequits. But you can't punish them too severly either, as there are some nice people with just bad connection. I would say the suggested time-based thing is not that bad, applied on teammates.

About the good push from the 2v4, that's nice. I won some 3v4 games and was quite proud, but never won 2v4 games outside of private games. I'm pretty sure the players will be quite proud and won't care much about the -5 or -10 points they got.

Oh by the way, I got a 3v3 from the very start the other day. Everyone spammed the nice button, and then we won the game. Would be sad to see one of the two disconnected guy taking a big malus for that, as it was a fair game.

 

[BlackZero]

The point deduction is supposed to change depending on the probability of one team beating the other. If a team is missing half its players and the other isn't, that probability shifts dramatically and the point deduction should reflect that. I would leave the -12 for people who DC/Rage Quit and set it to -8 for 3v4, -6 for 2v3 and -4 for 2v4. I wouldn't deduct points for 1v4.

 

[jsilva]

The point deduction is supposed to change depending on the probability of one team beating the other.

Presumably... ;)

If a team is missing half its players and the other isn't, that probability shifts dramatically and the point deduction should reflect that.

One of my complaints about the point system is that it doesn't account for a disconnection after the game starts. You can lose any of the ranked games pretty quickly, whether it be luck or simply being outplayed. So even if you lose a player 3 minutes into the game you're still at a potential game-losing disadvantage. But the system doesn't appear to account for losing a player even 10 seconds into the match!

I would leave the -12 for people who DC/Rage Quit

Rage quit, definitely. But DC? I know Nintendo can't tell the difference, but there has to be some concession.

...and set it to -8 for 3v4, -6 for 2v3 and -4 for 2v4. I wouldn't deduct points for 1v4.

Do you feel like it's fair to expect the 2 in a 2vs4 to win? By deducting points you're saying the game is still partially fair, and by losing 40% points (based on 10) you're effectively saying they are expected to win 40% as often as they would have if they had 4 players. Is that how you see it? That seems unrealistic to me.

It's unsafe to assume circumstances in a game like Splatoon, and it's more consistently fair to look at the circumstances at face value—a 2vs4 is all but a guaranteed loss and a 3vs4 is an unlikely win, among equally skilled players at least, and the points should reflect that. It's an embarrassment to Nintendo that a 2vs4 could result in a extra-penalised point loss.

 

[BlackZero]

One of my complaints about the point system is that it doesn't account for a disconnection after the game starts. You can lose any of the ranked games pretty quickly, whether it be luck or simply being outplayed. So even if you lose a player 3 minutes into the game you're still at a potential game-losing disadvantage. But the system doesn't appear to account for losing a player even 10 seconds into the match!

It can tell when someone is no longer connected and in the match, even if it happens in the middle I think the Devs didn't bother to accommodate this or forgot about it.

Do you feel like it's fair to expect the 2 in a 2vs4 to win? By deducting points you're saying the game is still partially fair, and by losing 40% points (based on 10) you're effectively saying they are expected to win 40% as often as they would have if they had 4 players. Is that how you see it? That seems unrealistic to me.

The game could consider these matches a draw, but I don't know that this would be fair either. In that case, the team with more players would actually lose points as if they lost and the team with less would gain as if they won. If the difference was minor, that wouldn't be a big deal. Problem is, by losing 1/4 of their team, the 3 man team would benefit greatly and the 4 player team would lose points as if they were actually beaten.

Spoiler: Warning: Maths

Here's how the K factor works. I'll keep it simple for the sake of example, but bear in mind the actual numbers involved will probably be a lot more complicated.

Let's say the matchmaker builds a lobby of two teams that are an even matchup based on team Elo rating (not in game rank).

Team A: 4 players with a player value of 600, for a team matchmaking value of 2400.
Team B: 4 players with a player value of 600 for a team value of 2400

These teams have a 50/50 chance of winning based entirely on the formula the computer uses and NOT their actual skill. Let's say Team B loses one player. their combined value now becomes 1800. This is what happens:

First the computer determines a transformed rating for each team: 10^(elo rating/400)

Team A: 10^(2400/400) = 1,000,000
Team B: 10^(1800/400) = 31,623

Then it calculates the expected outcome using the ratings we just calculated: transformed rating/(team a rating + team b rating)

Team A: 1,000,000/(1,000,000 + 31,623) = 0.97
Team B: 31,623/(1,000,000 + 31,623) = 0.03

Now, it calculates how much the player value will change after the match: new rating = elo rating + 32 (x - expected outcome).
32 is the K factor that chess uses. I suspect Splatoon's is much higher because of the dramatic shifts opponent difficulty after wins and losses, but this is just an example of how it all works.
Win: x = 1
Loss: x = 0
Draw: x = 0.5

Here's what a draw looks like:

Team A: 2400 + 32 (0.5 - 0.97) = 2385
Team B: 1800 + 32 (0.5 - 0.03) = 1815

Divide these by teammates, and everyone on Team A has their Elo value drop by 4 points in what was supposed to be considered a draw
Team B does well though. All teammates have their value increase by 5 just because they were a man down. If both teams had full teams and played a legit match...

Team A: 2400 + 32 (1 - 0.5) = 2416
Team B: 2400 + 32 (0 - 0.5) = 2384

...we see about the same 15 point change in team value, and 4 point change in each player's Elo rating. So even though we're trying to be fair and say "no one wins," someone is still getting screwed. That's just how the numbers work out. We can't even let the game use the initial 50/50 figure it had before someone DCed because the actual match ups aren't going to be a perfect 50/50, so even the draw formula will award and deduct points from someone.

The best solution would be for the devs to set a much lower k factor if the game detects that it lost connection with someone. I don't know if the devs will do this because they will have to program the game to switch in a really low K factor so the change is practically nonexistant when it detects that it lost connection with a player. Since they've discontinued updates, I really don't see this happening. So the easy/lazy way to fix it is to just lower the change in rank points.

 

[HypernovaSoul]

Ugh it seriously sickens me when that happens. It makes no sense either, because sometimes I'll lose a smaller amount of points if some of my teammates disconnected, and yet sometimes I'll lose the full whopping 10-12 when I was at a team count disadvantage....? What the absolute frick Nintendo. Have some consistency there. Frankly if someone disconnects, regardless of whether it was a connection loss or rage quit, their teammates should not suffer the full brunt of the consequences. It's so unfair. I really hope this is patched ASAP. And frankly if a team is missing more than 1 of its players, the match should absolutely end as a no contest (I don't mean for every rage quit, I mean if the match is 4v2 starting out). Seriously don't make people trudge through that BS...

 

[Redx115]

What about the times when you lose someone at the very start and only lose 1 point when you eventually lose the match? I've had that happen to me. It recognizes the fact that you played down a player then yet if you lose a teammate 30 seconds into the match you'll still lose the normal amount of points.

 

[Reala]

My perspective is that point deduction should be calculated from two things.

1. Number of players on your team. If someone else disconnects or rage quits, you shouldn't be punished for their actions. They already get their deduction, so why give the teammates a deduction too? Their actions do not reflect on your skill in the game.

2. K/D ratio. If you get 16/2 and everyone else on your team got, for example, 4/6 or worse, and your team loses, you should not have to lose points for this. Obviously, you have portrayed higher level play than the rest and it's most likely their fault your team lost. Let them get higher punishment and not you.

 

[Award]

2. K/D ratio. If you get 16/2 and everyone else on your team got, for example, 4/6 or worse, and your team loses, you should not have to lose points for this. Obviously, you have portrayed higher level play than the rest and it's most likely their fault your team lost. Let them get higher punishment and not you.

I agree with your points, however, in the ranked modes, there's sometimes no good way to use k/d as a metric. Sometimes players with a high k/d were off playing team deathmatch or spawncamping but never assisted in pushing the objective. Sometimes players with negative k/d or parity have that k/d because they were non-stop protecting the objective. RM for example, I'll often grab the RM on kamikaze runs if I think I might have a chance to push it a good distance. Setting it closer to the enemy goal is always beneficial. TC often requires sacrificial runs on the tower. Often the highest death counts will go to the players that were most focused on the objective. I'll also often prioritize opponents, self-sacrificing to take out a high priority opponent (player that whipped out an inkzooka or bomb rush, a trade is acceptable to prevent the widespread chaos they're going to cause, an enemy sniper plaguing the team, a trade is useful. And I'll always accept a sacrificial run to escort the RM carrier and paint a trail and clear the way. The high K/D guys are sometimes great, but sometimes they look so much better on the stat screen, while I was the one running the RM alone 4 times with no one around.

In TW that probably makes sense. SZ too. But in RM and TC it's so much more complicated. You need to balance k/d against time spent with/near the objective. It's usually easy to tell when the players with the bad k/d's were just playing bad versus choosing tactical losses. But there's plenty of time where the team with the great k/d's is the losing team and the team with all negative k/ds won because they focused on pushing the objective instead of on killing, while the team deathmatch team outclassed them in combat, but ignored the objective too often to do so.

It's so rage inducing when I see 3/4 of my team with double digit kills, single digit deaths, and the enemy team with single digit kills, double digit deaths. And they WON. Negative k/d doesn't always mean worse play.

I generally have a sense of when my negative k/d's are because I was taking risks on gambits that didn't pay off versus just plain suckage. It's 50/50 :D

 

[Dessgeega]

Well, one thing's certain: Splatoon 2 will probably handle rank points at least a little differently, and we'll surely still be here discussing this :P

 

[Nero86]

If it's still impossible for Nintendo to detect ragequits, I would accept this punishment, angry, but would accept.
I did win many matches against teams with 3 players who did got disconnected on the last minute and we could turn the tides. The contrary also ocurred, so other people were happy too.

 

[Award]

If it's still impossible for Nintendo to detect ragequits, I would accept this punishment, angry, but would accept.
I did win many matches against teams with 3 players who did got disconnected on the last minute and we could turn the tides. The contrary also ocurred, so other people were happy too.

Technically without hacking, they could detect ragequits (or at least ragequits and power outages) because the WiiU itself could report to the servers if it was powered off or not. That code would probably be hackable, and they don't have the infrastructure for it, but since the only way to RQ would be a power event, it would be simple enough to use that data (in theory.) I guess some people would power down their router (yikes) as well though.

But I agree, despite my two unfortunate disconnects last night ( Stupid extreme weather! :( ) the penalty needs to remain in place for the drop-out because we all know people would "ragescum" in losing matches (not that they don't rankscum anyway) The rest of their team should not be penalized for them dropping out though.

Then again all this comes right back to the primary crux of ALL the ranking problems: You shouldn't be scored as a team but on individual performance or some combination of the two. The team scoring is precisely what leads to the disconnect penalties, the carrying of poor players, and the dropping of good players and the overall broken mashup that is the ranks. If you were scored, or mostly scored on your own accomplishments in the match, you wouldn't lose so many points for going 20/2 regardless of if you had 2 or 4 players on your team. Maybe even a threshold where you lose 0 points for losing a match if you played well. (not gain points but not lose points - a wash.)

If I were personally redesigning the ladder system I'd get rid of the ludicrously ridiculous floating ladder entirely. Even if I was forbidden from tracking individual performance for the score and HAD to continue scoring the team by decree from high atop Mt. Myamotomius, I'd design a purely progressive ladder system that resets monthly or quarterly. You know, like MOST other ladder based games. You climb the ranks as high as you can, earn your way to that bracket, then it resets and you can try to earn it again in the current meta. Not good enough in this month's meta? Too bad, try again next month! It lets you always strive for that next level and/or to defend your prior title without this crazy sliding back and forth based on who dropped out and who happens to be online during your usual playtime and what C+'s were carried to A- by an S+ friend that creates such polluted player pools in the ranks over months.

 

[jsilva]

It can tell when someone is no longer connected and in the match, even if it happens in the middle I think the Devs didn't bother to accommodate this or forgot about it.



The game could consider these matches a draw, but I don't know that this would be fair either. In that case, the team with more players would actually lose points as if they lost and the team with less would gain as if they won. If the difference was minor, that wouldn't be a big deal. Problem is, by losing 1/4 of their team, the 3 man team would benefit greatly and the 4 player team would lose points as if they were actually beaten.

Spoiler: Warning: Maths

Here's how the K factor works. I'll keep it simple for the sake of example, but bear in mind the actual numbers involved will probably be a lot more complicated.

Let's say the matchmaker builds a lobby of two teams that are an even matchup based on team Elo rating (not in game rank).

Team A: 4 players with a player value of 600, for a team matchmaking value of 2400.
Team B: 4 players with a player value of 600 for a team value of 2400

These teams have a 50/50 chance of winning based entirely on the formula the computer uses and NOT their actual skill. Let's say Team B loses one player. their combined value now becomes 1800. This is what happens:

First the computer determines a transformed rating for each team: 10^(elo rating/400)

Team A: 10^(2400/400) = 1,000,000
Team B: 10^(1800/400) = 31,623

Then it calculates the expected outcome using the ratings we just calculated: transformed rating/(team a rating + team b rating)

Team A: 1,000,000/(1,000,000 + 31,623) = 0.97
Team B: 31,623/(1,000,000 + 31,623) = 0.03

Now, it calculates how much the player value will change after the match: new rating = elo rating + 32 (x - expected outcome).
32 is the K factor that chess uses. I suspect Splatoon's is much higher because of the dramatic shifts opponent difficulty after wins and losses, but this is just an example of how it all works.
Win: x = 1
Loss: x = 0
Draw: x = 0.5

Here's what a draw looks like:

Team A: 2400 + 32 (0.5 - 0.97) = 2385
Team B: 1800 + 32 (0.5 - 0.03) = 1815

Divide these by teammates, and everyone on Team A has their Elo value drop by 4 points in what was supposed to be considered a draw
Team B does well though. All teammates have their value increase by 5 just because they were a man down. If both teams had full teams and played a legit match...

Team A: 2400 + 32 (1 - 0.5) = 2416
Team B: 2400 + 32 (0 - 0.5) = 2384

...we see about the same 15 point change in team value, and 4 point change in each player's Elo rating. So even though we're trying to be fair and say "no one wins," someone is still getting screwed. That's just how the numbers work out. We can't even let the game use the initial 50/50 figure it had before someone DCed because the actual match ups aren't going to be a perfect 50/50, so even the draw formula will award and deduct points from someone.

The best solution would be for the devs to set a much lower k factor if the game detects that it lost connection with someone. I don't know if the devs will do this because they will have to program the game to switch in a really low K factor so the change is practically nonexistant when it detects that it lost connection with a player. Since they've discontinued updates, I really don't see this happening. So the easy/lazy way to fix it is to just lower the change in rank points.

It's certainly interesting to consider, although we of course don't know that Splatoon uses ELO. Also, what's stopping the system from making exceptions for point award/loss based on player disconnections? It seems to me the gain from a more accurate end result of point award/losses would offset any 'miscalculation' that might arise from the point award/loss not following the ELO player value. It seems to me the ELO player value would overall be more accurate.

 

[jsilva]

Technically without hacking, they could detect ragequits (or at least ragequits and power outages) because the WiiU itself could report to the servers if it was powered off or not. That code would probably be hackable, and they don't have the infrastructure for it, but since the only way to RQ would be a power event, it would be simple enough to use that data (in theory.) I guess some people would power down their router (yikes) as well though.

But I agree, despite my two unfortunate disconnects last night ( Stupid extreme weather! :( ) the penalty needs to remain in place for the drop-out because we all know people would "ragescum" in losing matches (not that they don't rankscum anyway) The rest of their team should not be penalized for them dropping out though.

Then again all this comes right back to the primary crux of ALL the ranking problems: You shouldn't be scored as a team but on individual performance or some combination of the two. The team scoring is precisely what leads to the disconnect penalties, the carrying of poor players, and the dropping of good players and the overall broken mashup that is the ranks. If you were scored, or mostly scored on your own accomplishments in the match, you wouldn't lose so many points for going 20/2 regardless of if you had 2 or 4 players on your team. Maybe even a threshold where you lose 0 points for losing a match if you played well. (not gain points but not lose points - a wash.)

If I were personally redesigning the ladder system I'd get rid of the ludicrously ridiculous floating ladder entirely. Even if I was forbidden from tracking individual performance for the score and HAD to continue scoring the team by decree from high atop Mt. Myamotomius, I'd design a purely progressive ladder system that resets monthly or quarterly. You know, like MOST other ladder based games. You climb the ranks as high as you can, earn your way to that bracket, then it resets and you can try to earn it again in the current meta. Not good enough in this month's meta? Too bad, try again next month! It lets you always strive for that next level and/or to defend your prior title without this crazy sliding back and forth based on who dropped out and who happens to be online during your usual playtime and what C+'s were carried to A- by an S+ friend that creates such polluted player pools in the ranks over months.

I'm fine with being scored as a team. That seems more real life and is good for us spoiled and self-entitled Americans to experience :) It is a virtue to be able to graciously be valued with those who potentially ruin things for us.

Interesting idea with resetting the ranks periodically. Maybe for casual gamers that would be upsetting. I'm not sure how I feel about it. There's got to be a better way. Whatever is there is ridiculous...

 

[binx]

About the disconnecting things, you can just unplug your Ethernet wire if you're using it, there is not only the power off option. So it's not that easy to know if it's rage quit or pure problem.

About ratio, I agree with Award, this can't be used as a metric. Right, if your ratio is awesome, you have more chance to have done well, but when I look at the game, lots of players are aiming at the ratio and not doing well even when they do get it. I remembr having played with a Tentatek who had the best ratio of my team two times in a row, and we were all negative. He was probably thinking "what a bad team I had these two last games" or something, but he was totally responsible for our losses from my point of view, as we were in facts playing a 3v3 RM while he was away, picking up some players that had no weight on the action. Of course, Tentatek kinda have this role of killing people, playing hero and so on. But he had each time a whole 4 minutes to realise this wasn't what we needed at all.

On the other side, I won with a 0-7 N-ZAP with a KO on Bridge TC. It was quite fast, the N-ZAP took the tower very well, I took this opportunity to kill and we won easily. He was first on our team, showing he was winning a lot. Another time I lost a game mainly because a Dynamo was taking the zones everytime he approached it on Mahi. He ended at 2-9, first on his team, because we couldn't prevent it well enough. These are two examples but I saw many more too.

It would be wrong to reward the best ratio, as it's not always the best thing to go after: it would push the players to play this way. By the way, it's a good decision to put the players with more wins in the upper list, than the one with the most kills. It goes in this direction, showing the players there is more than just a team death match going on.