It’s amazing how little people read into the changes of GAC and GAC matchmaking isn’t it. Every GAC matchup since the changes there have been probably more than a dozen posts complaining
GAC, since it’s inception, has seen the forums awash with people who only lose when matchmaking is unfair and only win when they are strategic masterminds.
Some people only want to fight rosters that essentially match their own. Even then some of them would complain that RNG went against them in mirror matches.
I have 2 GL's (SLKR and SEE) and everyone in my GA has 5 or 4 GL and they all have JMK. I had zero chance at all and with crystals being moved to GA not even having a chance at all to win a single match seems like a huge error.
I have 2 GL's (SLKR and SEE) and everyone in my GA has 5 or 4 GL and they all have JMK. I had zero chance at all and with crystals being moved to GA not even having a chance at all to win a single match seems like a huge error.
It is my understanding that the current matchmaking mechanic will take some time to level out and give us more “even” matches. Short term pain for long term gains.
Matchmaking needs to fixed, still a bit off on power vs power overall and so does this kind of ****.
Should be a penalty for not setting full defense, sorry to those who disagree but if they are going to have requirements then make them mandatory, i.e. you don’t set a full defense before the match begins means it’s an instant loss for you. I want to play the game as intended and not against lazy players looking for loopholes.
Matchmaking needs to fixed, still a bit off on power vs power overall and so does this kind of ****.
Should be a penalty for not setting full defense, sorry to those who disagree but if they are going to have requirements then make them mandatory, i.e. you don’t set a full defense before the match begins means it’s an instant loss for you. I want to play the game as intended and not against lazy players looking for loopholes.
So you couldn't get past a phoenix team? That is...sad.
Yeah. That one is confusing. They're in Kyber. His opponent basically gave him a win if he cleared. He seems to not understand how the scoring works vs incomplete defenses.
Typical for forums, I guess. Person A was given a huge advantage. Person A doesn't understand how scoring works. Person A rushes to forums to proudly display their ignorance rather than playing to see what happens or doing some light reading.
It will definitely even out over time. There are absolutely some short term "features" from the skill rating system that are undesirable. So far so good, and take that as someone who started 5-0 then subsequently lost 4 straight, including all 3 last week.
The only thing I would like to see is a bit smaller difference in winner vs. loser crystal comp for each round. I understand they're trying to incentivize participation, but the unevenness with the current paucity of skill data can lead to some pretty uneven matches (as expected). Paying the crystal penalty for that is a bit hard to swallow given how much of your crystal income is now tied to wins.
As skill rating becomes more "accurate", I'm good with the higher difference, but easing us in would have been helfpul helpful.
I broadly understand the new matchmaking & I quite like it. I’m currently matched against an opponent with 3 times as many GLs and c 600k GP advantage but I take that as evidence I’ve done relatively ok in the previous round to earn my **** whupping this time. However I do have two genuine questions that I’d like an informed response to:
1. Is there anywhere an authorative description of how the skill rating is generated - ie points per win, points deducted per loss etc?
2. If you are in division 4 of any league during round 4 and you get demoted at the end of that round, does that mean that, since you end in div 5 you get demoted further to the lower league? That’s one reading of the rules but it feels a bit unfair if it’s correct
If anyone knows where the definitive answers to these are posted, please share as I haven’t spotted that
@Wed_Santa
In answer..
1, nope, not that I’ve seen
2, if your skill level is above the lower threshold of that bracket at the end of a season, you stay in that division, someone demoted from k4 to k5 wouldn’t then be demoted to the next division
@Wed_Santa
In answer..
1, nope, not that I’ve seen
2, if your skill level is above the lower threshold of that bracket at the end of a season, you stay in that division, someone demoted from k4 to k5 wouldn’t then be demoted to the next division
Not sure about your answer for number 2 there. Everyone in division 5 of a league at the end of the season gets demoted as they are the lowest 10% in that league. So from my understanding if you get relegated to div 5 in the last week you will then get demoted to the next league down for the next GAC season
@Wed_Santa
In answer..
1, nope, not that I’ve seen
2, if your skill level is above the lower threshold of that bracket at the end of a season, you stay in that division, someone demoted from k4 to k5 wouldn’t then be demoted to the next division
Not sure about your answer for number 2 there. Everyone in division 5 of a league at the end of the season gets demoted as they are the lowest 10% in that league. So from my understanding if you get relegated to div 5 in the last week you will then get demoted to the next league down for the next GAC season
Thanks both. My reading of 2 was the latter but this does feel counterintuitive. So for example any player starting out in d5 in any league above carbonite could have a reasonable season with around a 50:50 win rate and still end up being demoted to a lower league while players who just don’t bother or otherwise have a very poor win rate will end up in the same place (or even better placed in some cases). I know devs say it will resolve eventually but if season 1’s result is unfair for some and if that translates to league placement in season 2, it really could take a while to sort out.
Not that I’m losing sleep about it but maybe a bit more transparency about the system would help
Please stop defending the new matchmaking system that is based on a "skill". What kind of skill is required to win if you outmatch your opponent by two or more GLs? Screw the daily rewards. Way to take fun out of another game mode in the game.
Please stop defending the new matchmaking system that is based on a "skill". What kind of skill is required to win if you outmatch your opponent by two or more GLs? Screw the daily rewards. Way to take fun out of another game mode in the game.
Isn’t the fundamental factor that’s making things less fun just that, post-nerf, there are just far fewer ways to beat a GL and therefore fewer ways to demonstrate skill? The principle of a skill (or win) based ladder is fine if there are genuine opportunities to play. I think my problems with it remain that the definition of ‘skill’ is not clear and that the promotion system seems to have an inherent unfairness for the double droppers from d4 in the final round
I think my problems with it remain that the definition of ‘skill’ is not clear
Skill: the ability to do something well; expertise.
"difficult work, taking great skill"
So by definition and following the example google gladly gave me, farming a GL is a difficult work so it takes great skill.
Always happy to resolve a problem.
It’s always good to see people can use Google but in this case, as I suspect you know, I was referring to the lack of a specific definition of how CG defines skill in GAC. So we know from their post that skill rating is based on both wins and losses (so it is n +/- x ) but as far as I know, x hasn’t been defined, nor has any info been communicated on other factors that might affect it. If you could Google that and provide a useful response it would be very much appreciated
I think my problems with it remain that the definition of ‘skill’ is not clear
Skill: the ability to do something well; expertise.
"difficult work, taking great skill"
So by definition and following the example google gladly gave me, farming a GL is a difficult work so it takes great skill.
Always happy to resolve a problem.
It’s always good to see people can use Google but in this case, as I suspect you know, I was referring to the lack of a specific definition of how CG defines skill in GAC. So we know from their post that skill rating is based on both wins and losses (so it is n +/- x ) but as far as I know, x hasn’t been defined, nor has any info been communicated on other factors that might affect it. If you could Google that and provide a useful response it would be very much appreciated
Thanks, that’s an interesting paper. Unfortunately it doesn’t answer my question as it’s about creating a workable ranking system for multiple leagues (eg if you wanted to rank football teams across different countries that have different domestic leagues). In the case of GAC there is just one ladder so the arithmetic doesn’t need to be complicated. All that’s really needed is the quantum of progression or regression and a set of simple rules to govern them. I’m surprised that CG haven’t found the time to share that as it seems self evident that any competition would only benefit if the competitors knew what the basic premise of it is. That probably needs fewer pages and footnotes than the paper you shared. It is however a nice thing to see so thanks for sharing it.
I think my problems with it remain that the definition of ‘skill’ is not clear
Skill: the ability to do something well; expertise.
"difficult work, taking great skill"
So by definition and following the example google gladly gave me, farming a GL is a difficult work so it takes great skill.
Always happy to resolve a problem.
It’s always good to see people can use Google but in this case, as I suspect you know, I was referring to the lack of a specific definition of how CG defines skill in GAC. So we know from their post that skill rating is based on both wins and losses (so it is n +/- x ) but as far as I know, x hasn’t been defined, nor has any info been communicated on other factors that might affect it. If you could Google that and provide a useful response it would be very much appreciated
CG isn't likely to tell us the method they are using outside of what they told us in the Q&A which is that it's a "bayesian matchmaking system" which is to say it's a mathematical system expanded beyond just an elo system. As such there is no 'x' - its not a fixed value and is based on your skill rating versus your opponents and the confidence in the expected outcome. Which means in extreme cases a loss can cause nearly no loss in skill points; you were expected to lose because the match was way beyond your skill ranking, or the inverse where a win gets nothing because the champion was way beyond their opponent's skill level.
Short(ish) answer is all that matters is beating opponents that are equal or greater skill level than you. Beating people rated much lower than you does not increase your score much if at all, beating an opponent much higher skill rating than you will result in a large gain for you and a large decrease for the opponent who was supposed to win. Matches currently don't seem to have an issue with creating matches of near equal skill levels - that may change as the skill levels normalize based on records rather than just GP. At which point matches may vary in skill levels by more than a few points and we'll have larger point gain/loss potential per match.
Thanks- that’s really useful & i suspect the closest I’ll get to an answer to my original question. Personally it feels to me that a Bayesian system would tend to stagnation over time - as the more predictable outcomes assign less movement up or down the ladder - so eddies get created in the flow where players bob about in a small part of the stream like so many discarded plastic bottles. But I suppose that’s a good outcome for CG as it will lead to frustration that prompts roster development (& associated spend). I’m trying to imagine how it would work in the English Premier League if Brentford got more points for beating Liverpool than Liverpool would get vice versa.
If I compare the matches in current setting to let's say soccer. Your opponent team start with the traditional 11 players on the field. You however only get to use 9. Of course you can win, but your chances are very low and all the opposite team has to do is play a mediocre defense.
If I compare the matches in current setting to let's say soccer. Your opponent team start with the traditional 11 players on the field. You however only get to use 9. Of course you can win, but your chances are very low and all the opposite team has to do is play a mediocre defense.
No, you each get 11 but your side has some teachers and doctors.
If I compare the matches in current setting to let's say soccer. Your opponent team start with the traditional 11 players on the field. You however only get to use 9. Of course you can win, but your chances are very low and all the opposite team has to do is play a mediocre defense.
My unsolicited advice would be to field 2 more players.
Please stop defending the new matchmaking system that is based on a "skill". What kind of skill is required to win if you outmatch your opponent by two or more GLs? Screw the daily rewards. Way to take fun out of another game mode in the game.
Meh. I'm 5-1 this GAC against opponents with 2 more GLs than me. So people could make an actual effort to outplay their opponent or they can just throw in the towel from the start. But, if they choose the latter, that's on them. Eventually they'll start getting matched up with bad players and people with bad rosters. It was made very clear from the start that it would take probably 2 full GACs for the matchmaking to settle.
Replies
You should take the time to read how GAC matchmaking has been amended. Eventually weaker players will drop down to face people they can beat.
GAC, since it’s inception, has seen the forums awash with people who only lose when matchmaking is unfair and only win when they are strategic masterminds.
Some people only want to fight rosters that essentially match their own. Even then some of them would complain that RNG went against them in mirror matches.
Its called SKILL rating for a reason, your lack of skill in building a good roster is why you are losing. Git Gud.
|| Grum the Mighty swgoh.gg || OldBaldGuyGaming Youtube
Did you lose both matches this week?
Welcome to galaxy of heroes. Don’t waste your breath petitioning for improvements.
Should be a penalty for not setting full defense, sorry to those who disagree but if they are going to have requirements then make them mandatory, i.e. you don’t set a full defense before the match begins means it’s an instant loss for you. I want to play the game as intended and not against lazy players looking for loopholes.
So you couldn't get past a phoenix team? That is...sad.
Typical for forums, I guess. Person A was given a huge advantage. Person A doesn't understand how scoring works. Person A rushes to forums to proudly display their ignorance rather than playing to see what happens or doing some light reading.
The only thing I would like to see is a bit smaller difference in winner vs. loser crystal comp for each round. I understand they're trying to incentivize participation, but the unevenness with the current paucity of skill data can lead to some pretty uneven matches (as expected). Paying the crystal penalty for that is a bit hard to swallow given how much of your crystal income is now tied to wins.
As skill rating becomes more "accurate", I'm good with the higher difference, but easing us in would have been helfpul helpful.
edit:words
1. Is there anywhere an authorative description of how the skill rating is generated - ie points per win, points deducted per loss etc?
2. If you are in division 4 of any league during round 4 and you get demoted at the end of that round, does that mean that, since you end in div 5 you get demoted further to the lower league? That’s one reading of the rules but it feels a bit unfair if it’s correct
If anyone knows where the definitive answers to these are posted, please share as I haven’t spotted that
Cheers
In answer..
1, nope, not that I’ve seen
2, if your skill level is above the lower threshold of that bracket at the end of a season, you stay in that division, someone demoted from k4 to k5 wouldn’t then be demoted to the next division
Not sure about your answer for number 2 there. Everyone in division 5 of a league at the end of the season gets demoted as they are the lowest 10% in that league. So from my understanding if you get relegated to div 5 in the last week you will then get demoted to the next league down for the next GAC season
Thanks both. My reading of 2 was the latter but this does feel counterintuitive. So for example any player starting out in d5 in any league above carbonite could have a reasonable season with around a 50:50 win rate and still end up being demoted to a lower league while players who just don’t bother or otherwise have a very poor win rate will end up in the same place (or even better placed in some cases). I know devs say it will resolve eventually but if season 1’s result is unfair for some and if that translates to league placement in season 2, it really could take a while to sort out.
Not that I’m losing sleep about it but maybe a bit more transparency about the system would help
Isn’t the fundamental factor that’s making things less fun just that, post-nerf, there are just far fewer ways to beat a GL and therefore fewer ways to demonstrate skill? The principle of a skill (or win) based ladder is fine if there are genuine opportunities to play. I think my problems with it remain that the definition of ‘skill’ is not clear and that the promotion system seems to have an inherent unfairness for the double droppers from d4 in the final round
Skill: the ability to do something well; expertise.
"difficult work, taking great skill"
So by definition and following the example google gladly gave me, farming a GL is a difficult work so it takes great skill.
Always happy to resolve a problem.
It’s always good to see people can use Google but in this case, as I suspect you know, I was referring to the lack of a specific definition of how CG defines skill in GAC. So we know from their post that skill rating is based on both wins and losses (so it is n +/- x ) but as far as I know, x hasn’t been defined, nor has any info been communicated on other factors that might affect it. If you could Google that and provide a useful response it would be very much appreciated
https://jmlr.csail.mit.edu/papers/volume12/weng11a/weng11a.pdf
You’re welcome.
CG isn't likely to tell us the method they are using outside of what they told us in the Q&A which is that it's a "bayesian matchmaking system" which is to say it's a mathematical system expanded beyond just an elo system. As such there is no 'x' - its not a fixed value and is based on your skill rating versus your opponents and the confidence in the expected outcome. Which means in extreme cases a loss can cause nearly no loss in skill points; you were expected to lose because the match was way beyond your skill ranking, or the inverse where a win gets nothing because the champion was way beyond their opponent's skill level.
If you want to take a look at some of the math in these systems I'd suggest Glicko ( https://en.wikipedia.org/wiki/Glicko_rating_system ) which is a small extension of elo, or Trueskill/TrueSkill 2 which are far more involved (https://www.microsoft.com/en-us/research/uploads/prod/2018/03/trueskill2.pdf)
Short(ish) answer is all that matters is beating opponents that are equal or greater skill level than you. Beating people rated much lower than you does not increase your score much if at all, beating an opponent much higher skill rating than you will result in a large gain for you and a large decrease for the opponent who was supposed to win. Matches currently don't seem to have an issue with creating matches of near equal skill levels - that may change as the skill levels normalize based on records rather than just GP. At which point matches may vary in skill levels by more than a few points and we'll have larger point gain/loss potential per match.
No, you each get 11 but your side has some teachers and doctors.
My unsolicited advice would be to field 2 more players.
As a Charlton Athletic supporter, I can confirm we would kill for some teachers and doctors
Meh. I'm 5-1 this GAC against opponents with 2 more GLs than me. So people could make an actual effort to outplay their opponent or they can just throw in the towel from the start. But, if they choose the latter, that's on them. Eventually they'll start getting matched up with bad players and people with bad rosters. It was made very clear from the start that it would take probably 2 full GACs for the matchmaking to settle.