In the current system, rating 1v1 games is borderline rating manipulation

6

Don't get me wrong, 1v1 is great and important and everything, and global is really bad, but hear me out:

  • smaller the team size, the more global rating points you get. I think this is a remnant of the zep days and really needs to be changed.
    • Example: grimpy vs nexus from fall tourny (140 rating difference), grimplex gained 64 points from 2 wins. Average big teamgame: +7-9 points (14 to 18 for two). This means grimpy would have to win 9 teamgames to get the same rating. On ladder 140 rating difference is worth about 15 points.
  • global is basically only used to for teamgames (be it for balance or otherwise). For players with a small deviation this means an 1v1 game can only make their rating less accurate
  • most of the playerbase only cares about global rating
    as they only play custom games (anecdotally)
  • while you can't compare ratings across different types of players, global is still somewhat representative of players' ability within the type of games that they play (and therefore is useful in balancing those games)
  • just because global is bad doesn't mean we should make it even worse
  • rating is fun. If you keep playing the same types of games you can compare your own performance across time, or relative to other people that you know play similar types of games etc.
  • tmm is not an alternative for many groups of people: gappers, setoners, high rating players. Hell, even astro crater players (turns out they are people too!)
  • we have anti rating manipulation rules for a reason (global really matters, believe it or not) and have previously unranked eg. unbalanced teams and ffas

It's totally fair to not personally care about global, but it doesn't mean that it's wrong for other people to care, and it's not an excuse to neglect the health of global rating. So I have 2 requests: 1. 1v1 games should amount the same to global rating as teamgames do and 2. until its fixed, tourny games should be unrated by default.

I feel like I'm putting myself in front of the firing squad by making this post (which is fucked up btw, I should think that the forum should feel like a bit safer of an environment than that), but it had to be done. There's some really messed up anti-global (actually to extend that, anti-tryhard) sentiment in the community that I've not seen anywhere else.

0

Although I don't think global rating is all that serious I agree with the tourny point. I think last tourny I played in (Morax's summer tourny) I gained around 80 points from 3 wins.

1

Tournies are so uncompetitive we need to unrank them to preserve the competitive nature of 5v5 map gen teamgames.

What? If you only play mapgen teamgames and get fat and then enter an environment of 1v1s where the distribution of players is smaller then you dump points in it to accommodate the skill level discrepancy. Who cares?

I win 2v2s and 3v3s twice as often as I win 5v5 and 6v6 games, doesn’t mean I go into the lobbies and demand they get unranked to preserve my fatness from those game modes.

1

Also you gain more rating with less players because the system has more confidence you are the result of the victory. It’s the same reason uncertainty levels out at a different level in ladder vs a pure dual gap player when you compare sigmas.

Wasn’t nexus like 2500 or so and grimplex like 2200-2300? That means in a BO5 the expected result would be either 3-2 or 4-1 in favor of Nexus since a 250 mu discrepancy is a 75% win chance. Since grimplex won his BO3 2-1, you’d expect the deviations to close quickly.

Likewise if you hosted teamgames with that rating difference and the expected losers won, you would see a comparative swap in rating. There is nothing wrong with that, that’s just how trueskill works.

2

Tournament games are ranked in order to facilitate the running of the tournament. Being able to see the status of each game after it finishes is important information and is necessary for a good tournament experience for both players, viewers, and tournament directors. I've been pretty lax about letting people unrank their tournament games recently, but after Fall Invitational Group Stage, I'm finding it too disruptive and I will likely require all tournament games be ranked for future tournaments.

If someone wants to implement seeing win/loss/draw functionality in the replay vault for unranked games, I would probably reconsider.

Global rating is always going to be an unreliable indicator of skill because people can pick the games they play. Someone who is good at 5v5 but relatively bad at 1v1 will gain rating from playing a series of 5v5s and lose rating after a series of 1v1s. There's no reason we should treat 5v5s as more valid than 1v1s.

2

It already takes forever for someone who is overranked (e.g. due to having historically played only 1 map in global) to try and get to a rank reflective of their skill to enable them to play other maps/games. At least with the current situation you can try and play more 1v1 and 2v2s so the process is less painful. Making this take vastly longer (which seems to be the suggestion by having 1v1s only have as much impact as say a 6v6 game) wouldn't be a positive move.

For example, I've already given up on trying to get a sensible global rank for now since I only have time for around 4 games a week (having been a near-full time astro player which took me up to around 1.6k rank, compared to a 4v4 TMM rank of around 1k) and am leaving it for when I have a bit more time or get bored with TMM, but I'd probably just give up altogether if the rate of rating change was slowed down even more.

I don't understand why tournaments would be unranked - aren't they the best indication of how good someone is, since people are far more likely to be trying to win and play competitively in a tournament than in a random global game?

Also making a 6v6 have the same ranking impact as a 1v1 feels like it would lead to ranking being even less reliable an indication of how good someone is. On a 6v6 the impact your performance has on the match outcome is vastly reduced, so you could play worse than your indicated ranking (or significantly better) for a number of games yet the results of those games doesn't align with that because of how your teammates did. For example, if a new player has 5 1v1 games and wins them all that would point to someone who is good at the game and their ranking should quickly increase to reflect it (which it does). If that same new player had 5 6v6 games and won them all, they could be rubbish and just got lucky with teammates, so their rank shouldn't increase as fast.

M27AI developer; Devlog and more general AI development guide:
https://forum.faforever.com/topic/2373/ai-development-guide-and-m27ai-v66-devlog

0

making tourney games unranked by default to satisfy an extremely small amount of players doesn't seem to be fixing the problem with global issues like at all, it literally makes things harder and breaks the concept of rating, toruney should be honestly getting 400% rating change instead of what it has right now.
rating can't be fun as your skill deviation of people with the same rating is insanely big so comparing an average player that hasn't spiked with 600-700 rating is irrelevant in performance tracking

queuing with a newbie to show him the beauty of tmm and meeting tagada be like:
https://www.youtube.com/watch?v=yLcRpdZ0Xb0&ab_channel=Tomoko

1

@ftxcommando said in In the current system, rating 1v1 games is borderline rating manipulation:

Also you gain more rating with less players because the system has more confidence you are the result of the victory. It’s the same reason uncertainty levels out at a different level in ladder vs a pure dual gap player when you compare sigmas.

Wasn’t nexus like 2500 or so and grimplex like 2200-2300? That means in a BO5 the expected result would be either 3-2 or 4-1 in favor of Nexus since a 250 mu discrepancy is a 75% win chance. Since grimplex won his BO3 2-1, you’d expect the deviations to close quickly.

Likewise if you hosted teamgames with that rating difference and the expected losers won, you would see a comparative swap in rating. There is nothing wrong with that, that’s just how trueskill works.

Perhaps you missed the part of the post where I mentioned that on ladder with a similar points difference it's only +15 for a win (rather than +30)? Why the difference between ladder and custom 1v1?

3

I don't want to throw anyone under the bus so I won't give any names, but I have heard from multiple (other) people that they don't want to play tournaments, because of the impact on their global rating it has. This is especially true for players who are not 2k+ and rating is actually super relevant for them to even get into games. If you think rating doesn't matter, then I guess you might want to try playing all welcome games? Or perhaps you'd like to give the random unbalanced setting a shot?

2

As swkoll said its easier for us if they are ranked, I dont mind either way as long as people accurate tell us their scores.
If there could be a Win/loss/draw visible even when unranked it would be perfect.
This is a matter being discussed to be covered in the upcoming FAF Official Tournaments rules

Ras Boi's save lives.

1

@blodir said in In the current system, rating 1v1 games is borderline rating manipulation:

@ftxcommando said in In the current system, rating 1v1 games is borderline rating manipulation:

Also you gain more rating with less players because the system has more confidence you are the result of the victory. It’s the same reason uncertainty levels out at a different level in ladder vs a pure dual gap player when you compare sigmas.

Wasn’t nexus like 2500 or so and grimplex like 2200-2300? That means in a BO5 the expected result would be either 3-2 or 4-1 in favor of Nexus since a 250 mu discrepancy is a 75% win chance. Since grimplex won his BO3 2-1, you’d expect the deviations to close quickly.

Likewise if you hosted teamgames with that rating difference and the expected losers won, you would see a comparative swap in rating. There is nothing wrong with that, that’s just how trueskill works.

Perhaps you missed the part of the post where I mentioned that on ladder with a similar points difference it's only +15 for a win (rather than +30)? Why the difference between ladder and custom 1v1?

It’s what I said. If your last 300 games are purely 5v5+, then you maintain a high sigma value. This is a result of FAF adjusting tau values after complaints about rating taking forever to adjust (and IMO was a good change and one I have tried to advocate other communities like BAR to do). Ladder has a lower sigma value in general because it’s a pure 1v1 environment so the system is more absolute about how settled a person’s rating is.

Sigma = ur general uncertainty
Tau = the level of “stabilized uncertainty” you want people to be at once the system is “confident” of their placement. It basically controls how volatile rating jumps are.

3

@blodir said in In the current system, rating 1v1 games is borderline rating manipulation:

I don't want to throw anyone under the bus so I won't give any names, but I have heard from multiple (other) people that they don't want to play tournaments, because of the impact on their global rating it has. This is especially true for players who are not 2k+ and rating is actually super relevant for them to even get into games. If you think rating doesn't matter, then I guess you might want to try playing all welcome games? Or perhaps you'd like to give the random unbalanced setting a shot?

You can literally entirely reverse this logic. Anybody that is currently 1600 or higher in skill level can immediately become 2200 by farming all welcome games if they care that much about getting into lobbies.

People should take the games as learning experiences, when mid tier players try to actually get better and learn and lose rating from it they don’t get kicked much at all unless it’s an intentionally smaller game and better players joined. In which case, their rating is kinda irrelevant since in those high rated lobbies, everyone knows who can hold their rating and who can’t already.

The way I always operated with tournies is that default is ranked, if both players want to play unranked then by all means. But we got four groups:

  • The guys that want to unrank because they enjoy being low ranked in global even though they got decent ladder/2v2/whatever
  • The guys that want to unrank because they want to be 2600 even though they are 2200 in ladder/2v2/whatever
  • The guys that want to rank because they want to lose global rating and have more chill games
  • The guys that want to rank because they see it as a way to get into the lobbies you describe/prestige/whatever.

Why is group B the most important?

1

@ftxcommando said in In the current system, rating 1v1 games is borderline rating manipulation:

@blodir said in In the current system, rating 1v1 games is borderline rating manipulation:

I don't want to throw anyone under the bus so I won't give any names, but I have heard from multiple (other) people that they don't want to play tournaments, because of the impact on their global rating it has. This is especially true for players who are not 2k+ and rating is actually super relevant for them to even get into games. If you think rating doesn't matter, then I guess you might want to try playing all welcome games? Or perhaps you'd like to give the random unbalanced setting a shot?

You can literally entirely reverse this logic. Anybody that is currently 1600 or higher in skill level can immediately become 2200 by farming all welcome games if they care that much about getting into lobbies.

People should take the games as learning experiences, when mid tier players try to actually get better and learn and lose rating from it they don’t get kicked much at all unless it’s an intentionally smaller game and better players joined. In which case, their rating is kinda irrelevant since in those high rated lobbies, everyone knows who can hold their rating and who can’t already.

The way I always operated with tournies is that default is ranked, if both players want to play unranked then by all means. But we got four groups:

  • The guys that want to unrank because they enjoy being low ranked in global even though they got decent ladder/2v2/whatever
  • The guys that want to unrank because they want to be 2600 even though they are 2200 in ladder/2v2/whatever
  • The guys that want to rank because they want to lose global rating and have more chill games
  • The guys that want to rank because they see it as a way to get into the lobbies you describe/prestige/whatever.

Why is group B the most important?

I described the reasons in the original post. Global is used for teamgames (including balancing them), not 1v1. While 1v1 and teamgames are related, they are not at all the same. We have no shortage of great teamgames players who aren't as experienced in 1v1 and do not perform at their rating level in 1v1. There are 45 1800+ players online right now, out of these players only swkoll, harzer, blast, dro, pepsi, banani and shen have a comparable 1v1 rating to their global rating (within about 150 rating points). That's 7/45. And yes, I've been in the remaining group of 38 teamgamer champions so far in 2022 too. Seriously, it's ok to be a teamgame player. It's ok to be a gapper or astro enjoyer too.

--> any reason that has to do with "I want to raise/lower my rating through tournaments" is the very definition of rating manipulation. Seeing as their tournament performance doesn't reflect their performance in the games that global rating is used for. This response also goes for maudlin's reply. I get that there's a problem with returning players, but trying to "fix" rating through manipulation is not the way.

I don't know what exactly is the best way to fix the system (no small amount due to not having read the paper on trueskill), so I can only talk about the problems, not really on how to fix them. That being said, it seems like there's an assumption in the system that 1v1 and teamgames are comparable ie. 1v1 performance gives a good description of teamgame performance, which might be true for shooter games (idk I don't play them), but it is not true for supcom.

Just to give you an example, can you imagine Foley (2400 global, 1500 ladder) playing a series of 1v1 matches vs a 2k ladder player. Then you get that newly birthed 2400 ladder guy in a teamgame air spot and he gets brutally dominated by now 2k rated Foley. Maybe you feel like it's justified that finally the ladder champions get the recognition they deserve or whatever, but in reality nobody's experience is improved.

I mean I'm sorry to say this guys, but I'd take Foley in a 5v5 mapgen any day over you guys (swkoll 2.2k ladder, noc 1.8k ladder, ftx, srry idk your rating), so I'm really glad he hasn't got any bright ideas about playing any 1v1 tournys recently.

As for making TD job harder (or any replay viewer for that matter), I think this is the only solid argument. The only real fix is that there should be win/loss/draw indicator. However:

  • there's no problem in terms of verifying results. Even if someone falsified results they would be found out and permabanned from tournys for wasting whatever few minutes it took for the TD to 10x through replays in case of conflict in player statements
  • the issue of seeing which game in a series was a win/loss doesn't matter for TDing, only for replay viewing and maybe for casting if the casters have downtime and want to look back. This is a legit issue for sure though
  • I would argue that the legit part of the issue is outweighed by the downsides of 1. people being disincentivized to play tournaments and 2. general experience of people playing custom games
0

That’s how it always goes. I will absurdly dominate Foley in normal teamgames. Foley will absurdly dominate me in sentons. Which one of us wins the 1v1 tourney? Who knows? Which one of these three deserves to be unranked? There is no rational reason for any, really.

0

@ftxcommando said in In the current system, rating 1v1 games is borderline rating manipulation:

That’s how it always goes. I will absurdly dominate Foley in normal teamgames. Foley will absurdly dominate me in sentons. Which one of us wins the 1v1 tourney? Who knows? Which one of these three deserves to be unranked? There is no rational reason for any, really.

I don't want to make this about you vs Foley so I won't say any more, but I definitely do not agree with your assesment.

0

@blodir said in In the current system, rating 1v1 games is borderline rating manipulation:

I don't want to throw anyone under the bus so I won't give any names, but I have heard from multiple (other) people that they don't want to play tournaments, because of the impact on their global rating it has. This is especially true for players who are not 2k+ and rating is actually super relevant for them to even get into games.

Rating for players under 2k is not exactly super relevant to get into games. 90%+ of high lvl teamgames ppl know eachother and if they are competent enough to play.

If it is a very high rated 2k+ ish lobby rating isnt the main thing to look at to balance it but skill is since once again everyone knows eachother.

Aside from that if youre under 2k u tend to fight people in tourneys that are equally or higher rated than you which means even if u get 4-0d youd only lose 50 rating or so.

Due to rng in teamgames its extremely common to balance from like -150 to 150 rating compared to ur average, so a drop of 50 rating is extremely common.

All in all it seems like an extremely niche thing that isnt really that harmfull that only very few ppl care about (since ive never heard about it) which imo more of an issue in their minds than a real issue.

0
This post is deleted!
2

Kinda sounds to me like 1v1 tourneys should use 1v1 ladder rating for entrance barriers and as rating system ingame, without global ever being involved in any way in 1v1 tourneys.

0

@katharsas
When hosting 1v1 Tournaments we do use ladder rating as the entrance barrier.
The issue is we have to host all the matches for these tournies as custom matches which means they affect global rating not ladder rating.

Ras Boi's save lives.

1

@blodir said in In the current system, rating 1v1 games is borderline rating manipulation:

--> any reason that has to do with "I want to raise/lower my rating through tournaments" is the very definition of rating manipulation. Seeing as their tournament performance doesn't reflect their performance in the games that global rating is used for. This response also goes for maudlin's reply. I get that there's a problem with returning players, but trying to "fix" rating through manipulation is not the way.

I don't know what exactly is the best way to fix the system (no small amount due to not having read the paper on trueskill), so I can only talk about the problems, not really on how to fix them. That being said, it seems like there's an assumption in the system that 1v1 and teamgames are comparable ie. 1v1 performance gives a good description of teamgame performance, which might be true for shooter games (idk I don't play them), but it is not true for supcom.

This is the fulcrum of it. You have mistake “the point of global rating” with “what I use global rating for.”

At no point in time was global defined as teamgame rating. At no point in time was global defined as map gen rating. At no point in time was global defined as 5v5 rating. That’s the way you use it.

Someone can theoretically only use global to play custom 1v1 games, then a global that only has 1v1 data is accurate. Someone can occasionally play 1v1 games, then a global where it occasionally accounts for 1v1 games is accurate. Someone joins your 5v5 map gen with 1v1 tourney rating, the game will now be inaccurate. You join with your 5v5 map gen rating, the tourney game will now be inaccurate. The only solution is making a rating for every FAF defined “skill zone” which only accommodates half a dozen high rated dudes at best while drastically increasing and prolonging one of the worst parts of FAF for everyone else (being an unknown entity in the rating environment).

Then you may argue that 1v1 weighs more than 5v5, but of course it does. I have won games where I’ve done literally nothing just because my team outclassed my enemy. That cannot happen in 1v1. And if you actually hold factors equal, a 1v1 is likely going to result in something like double the point swing that a teamgame would in the logic of the system. The bigger issue is you cannot hide a 2600 with 1600s against 1800s and a 2200 in a 1v1. It’s easy to hit 95% balance quality or whatever even if you’re 500 rating (read: smth like a 95% win probability) higher than the next best player.

Arguing the 1v1 rating is inaccurate but the 5v5 rating is accurate is also strange when there is also the rationale that the hard part of global is actually rising in rating. Keeping a rating is much easier. Precisely because of the dynamic I said above, it’s just a less extreme version of what Sid is currently doing.

The convo reminds me of the problems Chess has with massive rating discrepancies in their Elo system that FIDE tries to resolve. They first made an addition that caps any rating discrepancy as though there was a 400 rating discrepancy in order to not make it a waste of time for high rated players to play anyone less (this results in .8 points per win). They recently passed a new rule that only lets you utilize this rule once per tournament and beyond that, the games are counted as the original Elo calculation intends, which is essentially a <.01 point gain. That was to stop people artificially increasing Elo against way worse players due to the original rule.

Rather that fuck with 1v1, one can make the entirely different argument to unrank high rating discrepancy teamgames (sorta similar to the situation FIDE found itself in) since 1v1 is closer to an accurate metric of skill in general and the logic of teamgame rating calculations is just derived from the 1v1 system. All rating systems tend to suck when dealing with ridiculous rating range discrepancy, hard for a person to actually win every game except 1 in 300 games against somebody 800ish rating less, particularly due to how humans aren’t programs that stop learning.

Because I feel like some dude is going to point out that I contradict myself by saying that this task of consistently beating 800 rating worse players is hard but gaining rating in teamgames against them is easy, I will reiterate that the problem is you can make a teamgame where the odds of beating them is 50/50 or 60/40 rather than the .3/99.7 of a 1v1.