Upcoming Matchmaking Changes

[Wimbleton]

1. I don't want to win more. I want to stomp less drastically. The intention is to make such stomps closer to impossible for either side.

 

2. The simple win-only matchmaking algorithm can break any stomp streak unless a player plays each of his characters less than 10 times in a lifetime. Do not underestimate the power of very simple math. (But they probably complicated things, in which case, it might easily miss some important factor ir measure it wrong, fixing nothing).

 

Will be very interesting to come back to this post once lifetime win ratios are accessible. Hopefully this doesn't violate my NDA.

[Rafiknoll]

Will be very interesting to come back to this post once lifetime win ratios are accessible. Hopefully this doesn't violate my NDA.

 

I would love to see lifetime win ratio as long as it also shows objective points and the number of players one constantly premade with and their average win ratio. Can't have carried noobs look better than unlucky solo queuing vets

 

And again, I could be even not playing this game at all and still manage to design a simple matchmaking algorithm which works quite efficiently. It just requires some little common sense.

[Grxsr]

Yes it does!

 

-eric

 

 

Yes, Yes, And Yes

 

cross server too.. Yes And Yes

[Ayushman]

Yes, Yes, And Yes

 

cross server too.. Yes And Yes

 

High and Hyped already mate? good for you. I am too.

If my team doesn't go for win instead number farming, trolling, afking etc I would go to annoy the opponent (which I usually do) and go for kill (read BLOOD!). As a mara I will get to say 'I cant healz/ I cant guard/ I cant tauntz/ imma-go-over-there-and-watch-how-it-goes'-

Seriously, If ppl don't play to win then I would get to say ' Wow dude / Git gud / Play2Win / Lest-Everything-In-Vain'

[TrixxieTriss]

Will be very interesting to come back to this post once lifetime win ratios are accessible. Hopefully this doesn't violate my NDA.

 

I dont see adding ratios as proving anything. If you get heaps of noob teams, then it will be skewed and isn’t a reflection of player skill.

I actually hope this so called ratio isn’t made public because all it will do is encourage toxicity in the game. Which if ranked is any indication, it is the last thing we need in regs.

You can already see how many wins you’ve had, but even if they include losses, as long as you aren’t constantly in premades or requeuing or queue dodging, the law of averages should show a rough split of 50/50.

Edited July 16, 2018 by TrixxieTriss

[Rafiknoll]

I dont see adding ratios as proving anything. If you get heaps of noob teams, then it will be skewed and isn’t a reflection of player skill.

I actually hope this so called ratio isn’t made public because all it will do is encourage toxicity in the game. Which if ranked is any indication, it is the last thing we need in regs.

You can already see how many wins you’ve had, but even if they include losses, as long as you aren’t constantly in premades or requeuing or queue dodging, the law of averages should show a rough split of 50/50.

 

But... but... playing premades exclusively and stomping is pure indication of SKILL!!

 

Very good point in general. Didn't consider how toxic will several people be if their records would be public. Might as well just add rating, and premptively auto-ignore the whole team in case of a loss

[LD_Little_Dragon]

You can already see how many wins you’ve had, but even if they include losses, as long as you aren’t constantly in premades or requeuing or queue dodging, the law of averages should show a rough split of 50/50.

 

It doesn't work that way if you don't run in premades. When you solo que at busy times of the day you get to replace a lot of the players who drop mid-game or just before the gate opens. They need to track the ratio of wins, losses, and dropouts.

[TrixxieTriss]

It doesn't work that way if you don't run in premades. When you solo que at busy times of the day you get to replace a lot of the players who drop mid-game or just before the gate opens. They need to track the ratio of wins, losses, and dropouts.

 

Sorry, but how does the law of averages not apply?

[LD_Little_Dragon]

Sorry, but how does the law of averages not apply?

 

You don't win 50% of the game you backfill into.

 

For the games where I don't count as a backfill because someone drops 10seconds before the match starts, then by the time I finish loading and get out of the spawn zone the match has been fought with my team outnumbered. It's possible to win those games, but not 50% of the time. The team that starts a match one or two men down is always at a disadvantage.

[TrixxieTriss]

You don't win 50% of the game you backfill into.

 

For the games where I don't count as a backfill because someone drops 10seconds before the match starts, then by the time I finish loading and get out of the spawn zone the match has been fought with my team outnumbered. It's possible to win those games, but not 50% of the time. The team that starts a match one or two men down is always at a disadvantage.

 

Sorry, I’m not talking about back filling. But let’s look at that for a second. How many matches are we talking here in the big scheme of things. Not very many.

Over a period of time, even they average out to 50/50 or maybe 40/60. But the numbers of backfills vs how many standard matches you play will be insignificant in the long term.

 

I have a feeling I know the point you are trying to make, you just aren’t explaining it properly.

 

Let me see if I can have a go at it.

 

People will quite before the end of matches if the team is losing so it doesn’t spoil their ratio. The poor person who back fills will end up with loss on their ratio. (Have I got it right?)

If that’s the case, then yes, that’s a ledgitmate concern, but it still doesn’t negate the law of averages when discussing normal matches where people don’t quit or back fill, which was the point I was making and you told me I was wrong. When in fact I wasn’t wrong.

 

I can see how exploiting a system like this could cause some headaches and I have to wonder if Bioware have even thought of the consequences of implementing. It’s not like they’ve ever seemed to think about the consequences of past changes Which their poor track record shows most of the time they haven’t thought past getting the idea of the whiteboard. Bioware in their thinking are like “checkers” players, not “chess” players. They rarely look at the cause and effect to its conclusion. It’s probably why they keep making so many fundamentally bad decisions that drive players away.

 

I really hope they don’t go ahead with this. It will cause more problems than its worth. But if they do, they better be ready for the back lash and the exploitation of the system.

[AlrikFassbauer]

People will quite before the end of matches if the team is losing so it doesn’t spoil their ratio.

 

Ah, so all the "quitters" people rant about in this forum are actually "don't-want-my-rating-spoiled-quitters" ?

 

I can see how exploiting a system like this could cause some headaches and I have to wonder if Bioware have even thought of the consequences of implementing. It’s not like they’ve ever seemed to think about the consequences of past changes Which their poor track record shows most of the time they haven’t thought past getting the idea of the whiteboard. Bioware in their thinking are like “checkers” players, not “chess” players. They rarely look at the cause and effect to its conclusion. It’s probably why they keep making so many fundamentally bad decisions that drive players away.

 

I have written for a few years already that Bioware was never good at Player Psychology.

What they do is rather like what a mathematician would do. And Mathematics seriously isn't about psychology at all.

Edited July 16, 2018 by AlrikFassbauer

[Wimbleton]

Early leave once game has started = loss. What about people who solo/duo queue for a win ratio of >.8? Is that possible from your perspective? Edited July 16, 2018 by Wimbleton

[TrixxieTriss]

Ah, so all the "quitters" people rant about in this forum are actually "don't-want-my-rating-spoiled-quitters" ?

 

I have written for a few years already that Bioware was never good at Player Psychology.

What they do is rather like what a mathematician would do. And Mathematics seriously isn't about psychology at all.

 

Honeslty, I’m hypothesising what could happen with people quitting in the future if this ratio system was made available for people to see..

Obviously as the system is in regs at the moment, quitting means nothing to anything. If you’re talking about ranked, I’ll leave that for you guys to discuss because I’m not touching that bucket of crap.

 

I think if Bioware were mathematicians, they would be able to follow cause and effect. Which they do t seem to be able to.

[Wimbleton]

Before you begin a raid you can decide who you will bring and whether or not you want to go into said raid with those people, removing them or yourself if you do not wish to continue. The same goes for regs pregame. Since you have no control over who joins (besides premade) your only option is to remove yourself. This is also presently the only way to deal with maps you do not like, whereas in PVE you simply play what you want, no rng.

 

A person with 10000 games of experience, perfect gear and a win ratio of >.8 shouldn’t be made to play with the guy who trolls and purposely loses or the guy who has horrible gear and no experience. If they wish to, then great, but if not, leave during the pregame with impunity. If they decide to stay, then leave later after the match starts is considered a loss.

 

Backfillers get screwed and should maybe get the opportunity to leave so long as they do not leave the spawn area. Maybe give considerably increased component rewards so they have a reason to stay and (probably lose).

[Seterade]

for those who havent seen it yet

 

I am going to triple check with the team but I believe that what you are describing is still possible, just very unlikely. Basically, we hard enforce the rules I outlined, to a point. After a certain amount of time has passed the matchmaking will start to loosen rules to make a match happen.

 

Although we want matches to be as even as possible within the rules, at a certain point just having a match pop is more important. This is being compounded by a small population queuing on PTS and the makeup of those players. We are reviewing the data for the queues and matches to ensure we are seeing the intended behavior.

 

These reports are super helpful though as it may highlight an issue. The team is looking at everything!

 

-eric

Edited July 16, 2018 by Seterade

[AlrikFassbauer]

A person with 10000 games of experience, perfect gear and a win ratio of >.8 shouldn’t be made to play with the guy who trolls and purposely loses or the guy who has horrible gear and no experience.

 

But ... but ... farming inexperienced people - and using them as tra9ining tools - is so easy !

[Grxsr]

But ... but ... farming inexperienced people - and using them as tra9ining tools - is so easy !

 

Wish that would be the case, but the reality is that experience players get queue with bads, went it should be a mirror image queue, experience players v. experience players and novice v. novice.. Cross Server can at least bring some kind of balance to the Solo queues.. However I perfectly understand, Novice in Urwz shouldn't be used as target practice by experience players, and You're absolutely right.

Edited July 17, 2018 by Grxsr

[Rafiknoll]

A person with 10000 games of experience, perfect gear and a win ratio of >.8 shouldn’t be made to play with the guy who trolls and purposely loses or the guy who has horrible gear and no experience. If they wish to, then great, but if not, leave during the pregame with impunity. If they decide to stay, then leave later after the match starts is considered a loss.

 

It is annoying but the alternative is the opposite of matchmaking, and we can't have that. Randomness, which is the current situation, isn't a good solution either. And matchmaking with bads is not as terrible as it sounds. At least if they use the win-loss based matchmaking, then either the good player carries a victory and then the trolls and bads get a boost to their "matchmaking rating", and you are less likely to see them in your next game, or you lose because of the trolls and bads, your matchmaking rating is reduced and therefore you will get better players next time. Since nobody can see the matchmaking rating, there shouldn't be toxicity about it. In fact, low matchmaking rating is actually good for the player. It is also a good incentive not to quit in the final quarter of a losing match. You won't get the matchmaking rating reduced if you quit, which means you will get a team just as bad in the next time. If you suffered that far, might as well finish, take the loss and get a better position in the next match.

[Wimbleton]

Matchmaking will never be successful in this game due to low population which will never change. Cross faction will help the under-represented faction on each server by diluting themselves into the games of the other faction but it changes nothing in the larger picture because total pop doesn’t change.

[Rafiknoll]

Matchmaking will never be successful in this game due to low population which will never change. Cross faction will help the under-represented faction on each server by diluting themselves into the games of the other faction but it changes nothing in the larger picture because total pop doesn’t change.

 

Population levels are not a factor in the balance of the match. Lack of players in the queue might result in matches with too much support roles on both sides, but as long as they are divided equally, and distributed within their roles according to their skill (which is hopefully measured in a reasonable way. I am hoping a win-loss counter), then any group of 16 players can be split into a fairly balanced match, even if they are the only guys online in the whole server.

[Wimbleton]

There won’t ever be enough players in queue to get a proper match (based off win/loss or anything else). Matchmaking classes does nothing but remove one more excuse for bads blaming losses on something other than themselves.

[Rafiknoll]

There won’t ever be enough players in queue to get a proper match (based off win/loss or anything else). Matchmaking classes does nothing but remove one more excuse for bads blaming losses on something other than themselves.

 

Let's even assume there are only 16 players in queue, which is just enough to make a match. Let's say X of them are healers and Y of them are tanks, then the matchmaking would, first of all, decide that each team gets X/2 healers and Y/2 tanks. If both numbers are odd, the team with the fewer healers gets the more tanks. If only one of the numbers is ODD, the team who gets less of it get the better ones according to whatever intra-role matchmaking factor they implement (which can be win/loss counter or something else. Hopefully something which is a good win-probability measurement). As for the distribution of DPS, they get distributed according to their own intra-role matchmaking factor so that the eventual sum of each team is as close to similar as possible.

 

The teams will not be exactly equal, but the greatest difference in the number of healers between the teams is 1, the greatest difference in the number of tanks between the teams is also 1, and the greatest difference in the number of healers AND tanks between the teams is ALSO 1. DPS-wise, the teams should be very close to equal, so unless the best DPS (which is in group A) is not significantly better than the 2nd best DPS (which should in 90% of the cases be in group B, to make the distribution equal) then the DPS power of both teams is very close to equal.

 

There you go, the match closest possible to "proper" has been set with the minimal possible amount of players. The only thing which is caused by the lack of numbers in the queue is the possibility of too many healers but at least they will be distributed equally.

Edited July 18, 2018 by Rafiknoll

[Wimbleton]

Let's even assume there are only 16 players in queue, which is just enough to make a match. Let's say X of them are healers and Y of them are tanks, then the matchmaking would, first of all, decide that each team gets X/2 healers and Y/2 tanks. If both numbers are odd, the team with the fewer healers gets the more tanks. If only one of the numbers is ODD, the team who gets less of it get the better ones according to whatever intra-role matchmaking factor they implement (which can be win/loss counter or something else. Hopefully something which is a good win-probability measurement). As for the distribution of DPS, they get distributed according to their own intra-role matchmaking factor so that the eventual sum of each team is as close to similar as possible.

 

The teams will not be exactly equal, but the greatest difference in the number of healers between the teams is 1, the greatest difference in the number of tanks between the teams is also 1, and the greatest difference in the number of healers AND tanks between the teams is ALSO 1. DPS-wise, the teams should be very close to equal, so unless the best DPS (which is in group A) is not significantly better than the 2nd best DPS (which should in 90% of the cases be in group B, to make the distribution equal) then the DPS power of both teams is very close to equal.

 

There you go, the match closest possible to "proper" has been set with the minimal possible amount of players. The only thing which is caused by the lack of numbers in the queue is the possibility of too many healers but at least they will be distributed equally.

 

Yeah, thats called class match making, and that only matters when equal skill is represented on both sides. There's not enough of a population to try and balance off of skill. So you typed 15 paragraphs to say exactly what I already said or you're one of those special cases who think that its the other team having an extra healer that caused you to lose.

[Rafiknoll]

Yeah, thats called class match making, and that only matters when equal skill is represented on both sides. There's not enough of a population to try and balance off of skill. So you typed 15 paragraphs to say exactly what I already said or you're one of those special cases who think that its the other team having an extra healer that caused you to lose.

 

You failed to understand what I said then, or you gave up on reading a medium length post and missed the:

Let's even assume there are only 16 players in queue, which is just enough to make a match. Let's say X of them are healers and Y of them are tanks, then the matchmaking would, first of all, decide that each team gets X/2 healers and Y/2 tanks. If both numbers are odd, the team with the fewer healers gets the more tanks. If only one of the numbers is ODD, the team who gets less of it get the better ones according to whatever intra-role matchmaking factor they implement (which can be win/loss counter or something else. Hopefully something which is a good win-probability measurement). As for the distribution of DPS, they get distributed according to their own intra-role matchmaking factor so that the eventual sum of each team is as close to similar as possible.

 

The teams will not be exactly equal, but the greatest difference in the number of healers between the teams is 1, the greatest difference in the number of tanks between the teams is also 1, and the greatest difference in the number of healers AND tanks between the teams is ALSO 1. DPS-wise, the teams should be very close to equal, so unless the best DPS (which is in group A) is not significantly better than the 2nd best DPS (which should in 90% of the cases be in group B, to make the distribution equal) then the DPS power of both teams is very close to equal.

 

The green part is how there is a skill balance between the supports in both teams, and the red is how there is a skill balance between the DPS.

Do tell why do you insist you need the population to make skill balanced matchmaking, and why is it impossible, according to you, to balance any given 16 people.

I will even dumb it down more for you, for the following example, we will take teams of 8 DPS versus 8 DPS, to make the example simplest: Let's rate all 16 players and give each a serial number while 1 is the best and 16 is the worst. Then matchmaking would put in team A:

1, 3, 5, 7, 10, 12, 14, 16 = 68

And in team B:

2, 4, 6, 8, 9, 11, 13, 15 = 68

(note the lower sum means the better team)

There you go. Unless the gap between number 1 and 2 is insurmountable, neither team can be so much better than the other team. Simple math...

[Wimbleton]

You need to learn not to WOT, no one reads that ****. Your simpleton understanding of matchmaking also does not account for premades, or the fact that whatever values they use to assess "skill" can be manipulated easily or simply not actually be a good measure of skill, or the fact that between the 15 and 16th most skilled person could be a gap as wide as the distance from here and reality. Edited July 19, 2018 by Wimbleton