Page 2 of 2 FirstFirst 1 2
Results 21 to 37 of 37

  Click here to go to the first staff post in this thread.   Thread: Matchmaking: What do we want it to do?

  1. #21
    Junior Member epfrndz's Avatar
    Join Date
    Mar 2013
    Posts
    5
    I like bwoneill's proposed matchmaking formula (I'm a fan of applied statistics, even though I could only understand the gist of bwoneill's equation) but my only concern for this is that wouldn't it take up server load if it had to compute every single match?

    And wouldn't basing underdog bonuses on the probability of winning the match make it complicated for some users to understand how it is given to them. (I like that method though, I think it rewards the player to atleast stand and fight)

  2. #22
    Superbacker netnazgul's Avatar
    Join Date
    May 2012
    Location
    Belarus
    Posts
    456
    Dropping 5c here - I don't understand, why even in a match "high elo player, lower team power" vs "low elo player, higher team power" the former should still get elo gain based solely on Elo and not together with team power handicap. Cause in such situation Elo shows that the latter player has very low chances of winning the match due to skill difference, and we artificially raise his chances without anything to be given for more skillful player. In the whole, it then leads to higher ranked players being forced to play much tougher uneven matches without any reward for that.
    If you don't know where to put it - put it in the pillage

    Steelhammer Tribune issues collected here
    Some of my Factions games can be observed here
    Also possible streaming at http://www.twitch.tv/netnazgul

  3. #23
    Senior Member loveboof's Avatar
    Join Date
    Mar 2013
    Posts
    137
    Would this issue be solved by simply allowing us to alter our teams directly before a match?

    Exactly like XCOM - have us click 'ready' or something...

  4. #24
    Junior Member Hölgarn's Avatar
    Join Date
    Mar 2013
    Posts
    3
    Quote Originally Posted by raven2134 View Post
    Underdog bonus is awarded win or lose.
    Maybe its just me but I didnt think 1 point of renown is enough comensation for the constans losing streak and the failing statistics. I meet constantly 2+ higher teams

  5. #25
    Developer raven2134's Avatar
    Join Date
    May 2012
    Location
    Manila, Philippines
    Posts
    1,061
    I am leaning toward underdog awarding 1 renown per 1 power gap. But I would also think the win bonus needs to scale up a bit if the losing player can earn a big amount from being outmatched.

    Net, another concern, what you say would assume that skill-power difference behaves that way. But as with the discussion on unit/build balance, just because a situation or setup is winnable, doesn't make it balanced or fair if the other player can win much more easily with it. I.e. the power gap at a point we do not really know may outpace skill level and for that matter, we don't know how much skill level is ideal for this kind of offsetting.

    This is also why basing Elo gain on % to win (or lose) is a good idea rather than hard coding it to a set difference in power or skill. I do believe the rating/matching was based on % to win.

  6. #26
    Superbacker netnazgul's Avatar
    Join Date
    May 2012
    Location
    Belarus
    Posts
    456
    Quote Originally Posted by raven2134 View Post
    Net, another concern, what you say would assume that skill-power difference behaves that way. But as with the discussion on unit/build balance, just because a situation or setup is winnable, doesn't make it balanced or fair if the other player can win much more easily with it. I.e. the power gap at a point we do not really know may outpace skill level and for that matter, we don't know how much skill level is ideal for this kind of offsetting.

    This is also why basing Elo gain on % to win (or lose) is a good idea rather than hard coding it to a set difference in power or skill. I do believe the rating/matching was based on % to win.
    What I'm saying is that John clearly states, that team power CAN be a substitute for Elo difference and thus is taken into matchmaking consideration. But the opposite - elo gain/loss should be based also on team power - is not considered. So either team power should be taken out of matchmaker (thus resulting in equal team power matches) or player should gain more Elo when having less team power than his opponent.

    EDIT: Putting it straight - when I'm playing vs someone having 600+ less Elo than me, I have a fairly big chance to win the match, but I'll gain 1 Elo. Then if the same opponent has, for example, twice the team power (12 vs my 6), my chances of winning are severely cut down, but I'll still gain 1 Elo for a win and lose 31 Elo after defeat.
    Last edited by netnazgul; 03-11-2013 at 08:05 AM.
    If you don't know where to put it - put it in the pillage

    Steelhammer Tribune issues collected here
    Some of my Factions games can be observed here
    Also possible streaming at http://www.twitch.tv/netnazgul

  7. #27
    I think one issue that can be addressed fairly easily is the leveling new units problem. Right now you need kills. If we changed it to something else it would alleviate the whole problem. I understand you want people to take the time to level units etc..., so there is a sense of accomplishment and to slow the progress down a bit. Especially since you can buy the in game currency for real $$. You want to avoid the P2W trap.

    So how to fix this problem?:

    - Allow players to change classes with their units. I would say give a few free times out (say 6) a week. Then charge renown after that to change classes. Say 25 renown ( it should be a fairly cheap amount of renown to encourage changing rather than grinding, which no one likes to do). This avoids the whole need for leveling a bunch of different types of units. Right now it is not too much of an issue, but when there are 4 classes for each unit and there are more than 4 types of units it will become more of an issue. You can already buy a rank 1 unit so imo there is no need for this grinding aspect of the game. What this also does is allow players who have higher ranked units to try different configs out. It will take an awful long time to get a full rank 3 team. Once you get there you do not want to have to grind 6 more people to rank 3 and then maybe a few more for alternate configs. Instead All you you will have to do is grind 3-4 of each unit type to level 3 and then switch between classes. It is a better system all around imo. This would get rid of the need for high skilled players to play lower ranked match-ups

    - Change how units level:
    I already have a thread on this and feel this should go hand in hand with the above. Getting x kills on a unit to level is not a very good mechanic in a strategy based game. There is more detail in this thread on the subject.

    Some other things that will help with match-making/playing.

    - Have # of games played also be a factor in the match making process. Since someone who has 100 games is going to be generally better than someone with 10 games for the most part anyway.

    - Have a play same rank bg button in match-up. Besides elo and games played rank is the determining factor in what is an even fight. This button would allow a player to make sure that at least one of these factors is even.

    - Have a "friendly quit" button. This would be a sort of surrender, but at no penalty to either party and no elo or renowned gained and the game would not count for any stats. The opponent has to agree to the FQ. In this way if you are really mismatched or you both agree that you do not want to play each other after getting a match you could just "FQ" and the game would be over.

    - Have a draw button. This has been brought up already so I will not go into discussion, but I feel it should happen.

    - Have a training grounds area. This is where people that have, say sub rank 6 bgs and newer players could go and find a match. Make it rank 5 bgs and under, as well as no more than 40 total games played only. No elo is gained, but the fear of playing a much higher ranked and skilled opponent would be gone. The max games played is to prevent players from farming new players for renown. Would be perfect.

  8. #28
    Superbacker netnazgul's Avatar
    Join Date
    May 2012
    Location
    Belarus
    Posts
    456
    Quote Originally Posted by Vexbane View Post
    - Have a training grounds area. This is where people that have, say sub rank 6 bgs and newer players could go and find a match. Make it rank 5 bgs and under, as well as no more than 40 total games played only. No elo is gained, but the fear of playing a much higher ranked and skilled opponent would be gone. The max games played is to prevent players from farming new players for renown. Would be perfect.
    Not addressing the whole bunch of features mentioned - this one is single best idea to eliminate farming.
    If you don't know where to put it - put it in the pillage

    Steelhammer Tribune issues collected here
    Some of my Factions games can be observed here
    Also possible streaming at http://www.twitch.tv/netnazgul

  9.   This is the last staff post in this thread.   #29
    Art Director Arnie's Avatar
    Join Date
    Apr 2012
    Posts
    319
    John will be looking into this when he arrives. We just got back from a long, tiring show.
    Thanks a ton for all the feedback.

  10. #30
    Superbacker gripho's Avatar
    Join Date
    May 2012
    Location
    France
    Posts
    62
    Quote Originally Posted by epfrndz View Post
    I like bwoneill's proposed matchmaking formula (I'm a fan of applied statistics, even though I could only understand the gist of bwoneill's equation) but my only concern for this is that wouldn't it take up server load if it had to compute every single match?
    I like it too, but I'm not sure power difference translates to Elo linearly... Besides, power difference in team power below 6 and above 6 don't have the same effect at all in my opinion: having 1 more stat point and the possibility to use higher level abilities is a clear advantage, but much less than having the possibility to re-stat and to use active abilities compared to a base unit.

  11. #31
    Junior Member caine1138's Avatar
    Join Date
    Nov 2012
    Posts
    10
    Quote Originally Posted by netnazgul View Post
    Not addressing the whole bunch of features mentioned - this one is single best idea to eliminate farming.
    Im afraid that that would split the playerbase, and as a result... I feel that that is unwise.
    Runes from the Web...

  12. #32
    Quote Originally Posted by caine1138 View Post
    Im afraid that that would split the playerbase, and as a result... I feel that that is unwise.
    You may be right caine in that it would split the player base. In this case that is a good thing. You want new players to have a good experience. If they get a bad 1st impression they will leave and not come back. Especially in such a new game like BS:F. You want the low level player base to be able to feel "safe" and play the game and enjoy it.

    At the same time you want the established players to be able to have a good time as they are more apt to spend $$ on the game. A lot of players dislike the mismatches. This includes the player that has the stronger team. They may get the win, but it was not an enjoyable game as there was little skill or tactics involved when you have such a mismatch.

    With the game being new it is the best time to implement a training grounds type of area. This way there is a transition to rank 6 and the pvp ladder. Right now new players are thrown in the deep end and are left to sink or swim. My idea is to give them some shallow water to wade in. Will it create longer que times? Maybe, but I think most people will not mind as imo it will bring more players to the game ultimately.

  13. #33
    Backer Slimsy Platypus's Avatar
    Join Date
    Nov 2012
    Location
    Northeast US
    Posts
    368
    Regarding the new player experience:

    I threw this idea around a while back in a discussion in the chat box. In the future when the AI gets released, perhaps you could collect your first 5 kills on basic units against the AI. That way, new players can both learn the game and promote their team up prior to entering the lions den of multiplayer matchmaking. Alternatively they could just queue right in and let fate decide...

    Just a thought

  14. #34
    could there be a system that allows for players with stronger teams to 'nerf' toons of their choosing before the match begins so the teams are more power balanced, even if the players have different levels of experience?

    I have played lots of Blood Bowl Online, and matchmaking on that game has similar issues of low power teams being matched against high power teams. Its frustrating, and would be great to finally see a game company put some serious effort into resolving the issue. The devs of Blood Bowl Online could give 2 sheeits about this issue.

    In BBO, if you have a much lower team power than your opponent, then you get temporary gold so you can buy stronger players just for that one game. But really i think it should work in reverse, with the higher power team having to nerf some of their strongest toons.
    Last edited by masterblaster; 03-11-2013 at 09:37 PM.

  15. #35
    Thanks for the feedback, guys.

    Quote Originally Posted by epfrndz View Post
    I like bwoneill's proposed matchmaking formula (I'm a fan of applied statistics, even though I could only understand the gist of bwoneill's equation) but my only concern for this is that wouldn't it take up server load if it had to compute every single match?

    And wouldn't basing underdog bonuses on the probability of winning the match make it complicated for some users to understand how it is given to them. (I like that method though, I think it rewards the player to atleast stand and fight)
    My algorithm would find a match in O(N) time where N is the size of the queue, the queue could a queue should be emptied in less than O(N^2) time. It is possible that a very large queue could increase the load, however this could be mitigated by limiting the queue size. For example, if a queue grows to 200 players, then you could split it into two 100 player queues. That would cut the CPU time by nearly a factor of 4.
    However, I don't think the additional CPU time would be significant. I just did a quick check on a C interpreter using all the functions necessary to implement my algorithm. My laptop went through 100,000 iterations in about 0.51s.
    Determining players' chances of winning may seem complicated to the average player, but I don't think it's necessary that they understand. The way I see the underdog bonus working is that a pop-up message will be shown to the underdog at the beginning of the match. Something like: "The spectators agree you are the underdog here and are impressed by your gall for even showing up. If you don't flee, you will receive X bonus renown." The player need not know how that bonus is calculated, just how much that bonus is.

    Quote Originally Posted by gripho View Post
    I like it too, but I'm not sure power difference translates to Elo linearly... Besides, power difference in team power below 6 and above 6 don't have the same effect at all in my opinion: having 1 more stat point and the possibility to use higher level abilities is a clear advantage, but much less than having the possibility to re-stat and to use active abilities compared to a base unit.
    I agree that the effective Elo difference is probably not linear. Most likely, we would need a table of Elo adjustments for all possible power levels (0-18). At the very worst, I can imagine that we may need a table of all possible power combinations (0-18 x 0-18). Also, these values would also have to be adjusted periodically as the average build changes. However, a linear approximation would still be helpful as a starting point as these values are adjusted.

  16. #36
    Superbacker netnazgul's Avatar
    Join Date
    May 2012
    Location
    Belarus
    Posts
    456
    Adding to the current matchmaker issue:

    was playing against current rankings #1 (he is obviously higher than me in rankings, I'm around 1400ies now). He had all rank1 units except Warmaster, who was rank3, so I suppose team power equals 8. And this was matched against my pow6 team in 1-2 minutes.

    Also off-topic - I've lost but haven't got any underdog bonus. Shouldn't it be given in any non-equal game win or loss?
    If you don't know where to put it - put it in the pillage

    Steelhammer Tribune issues collected here
    Some of my Factions games can be observed here
    Also possible streaming at http://www.twitch.tv/netnazgul

  17. #37
    Junior Member caine1138's Avatar
    Join Date
    Nov 2012
    Posts
    10
    Quote Originally Posted by Vexbane View Post
    You may be right caine in that it would split the player base. In this case that is a good thing. You want new players to have a good experience. If they get a bad 1st impression they will leave and not come back. Especially in such a new game like BS:F. You want the low level player base to be able to feel "safe" and play the game and enjoy it.

    At the same time you want the established players to be able to have a good time as they are more apt to spend $$ on the game. A lot of players dislike the mismatches. This includes the player that has the stronger team. They may get the win, but it was not an enjoyable game as there was little skill or tactics involved when you have such a mismatch.

    With the game being new it is the best time to implement a training grounds type of area. This way there is a transition to rank 6 and the pvp ladder. Right now new players are thrown in the deep end and are left to sink or swim. My idea is to give them some shallow water to wade in. Will it create longer que times? Maybe, but I think most people will not mind as imo it will bring more players to the game ultimately.

    I think your priorities are out of whack, on this.

    Its more important to keep the 400-500 players going head to head. Until we reach a higher population. Im guessing a 2nd room would be doable when we reach a couple thousand players actively playing daily.

    There are many ways to address the issues you are siting, that do not seperate the players into rooms... even some mentioned in this thread.
    Last edited by caine1138; 03-12-2013 at 07:04 PM.
    Runes from the Web...

Page 2 of 2 FirstFirst 1 2

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •