Jump to content
Scr@tcH

Metal League Feedbacks and Improvement suggestions

Recommended Posts

Hello everyone, veterans, newcomers, metal gladiators alike !

Yesterday, the last phase of the 3rd Metal League Qualifier ended, and I felt like it's the perfect time to start a huge and complete topic about pros/cons and how to improve this system !

You will find below my own feedback and vision about how the next metal league should improve : do not hesitate to comment and give your own here !


 

As an introduction, there are some points about HMM on which the ML system is highly depending on :

 

  1. Number of teams participating to each stages, and by extension the amount of competitive players/teams in the game :

    We were a bit disappointed when we learned that the system is designed with 100+ teams being simulated at the same time to balance the system as we barely reach 20/30 teams when considering the whole ML competition. This criteria leads to 2 distinct scenarios depending on if the game reach a player-base big enough : one would be to stick to the Swiss-round (or Round-robin if there are even less teams) format that was used during first ESL EU & SA tournaments and worked pretty well ; the other one would be to improve the current system (which is the optimistic scenario about reaching a great number of competitive teams I am keeping in this example).


     
  2. Technical aspects about MatchMaking algorithm and Steam Friends system :

    Steam friends integration is clunky since Valve modified it (even bugging on it's own in other games in my personal experience) and the implementation in the game is a technical aspect on which we have to wait for the dev-team fixes or stepping out of Steam Friends system at least for MatchMaking and profiles purposes.
    But this led to issues allowing sniping mechanics (checking which team is currently in the queue to optimize encounters and thus ranking points, or even enter the "wrong queue" in casual to bait teams doing this ?) and the way to counter this was to use Steam's invisible mode so there would be less bias and snipings (when you go invisible, you can't either check who is entering the queue). Plus, if you have a friend or the 5th player of your team checking the observer mode, you can know which team you are encountering while you are still in the pick phase (if you have good information about their playstyles/picks, it gives you a nice advantage).

    Proposed solutions :

    -> Hide which teams are encountering each other in the spectator mode until the pick phase has ended.
    -> Find a way to hide/not display every player status when you enter the Metal League queue.

    The Matchmaking algorithm is matching teams too fast even if the teams queueing are the 1st and the last of the current phase, which is not a good scenario to earn points/have competitively interesting matches, which can be a problem in such ranking system.
    Proposed solution :
    -> It would be a solution for the RANKED MODE as well : impose a threshold of your current ranking points +/- X inside which you would be able to match other teams. The longer you wait in the queue, the bigger the X (inclusion) gets (you have 1400 points, threshold X is initially at 100 so you get matched with teams around 1300 to 1500 points first and it increases with time to 1200-1600, 1000-1800 and so on ...). Something like waiting for 10 minutes to get to situations where you would be able to get matched 1st against last team of the league could be nice.


     
  3. Overall balance between all characters of the game :

    During the Combat Test, the overall balance between machines was really difficult to adjust and was still not very good during the EU ML (leading to normalized picks in draft and the same compositions during qualifiers phase). It's better now, but still not enough in my opinion for the competitive E-sport scene we want to build. I won't discuss balance issues here, specially when uncertainties about garage remain.
    Spoiler

    (in my opinion, Vulture is OP, LM PM and WF a bit too strong, Arti really needs a buff, Photon and FMJ needs a small boost)

    It's not a huge problem during qualifiers, but it is for final phases where there are drafts. We also have different playstyles between SA and EU, and the balance is not felt the same way. If you ask me, I would say SA players have more knowledge about overall teams compositions and EU players know more about precise machine capabilities/picks/counterpicks.
    Proposed solutions :

    -> Create a "balance team" in both SA and EU servers which would work with testers and devs.
    -> Balance characters accordingly to their "maximum capabilities" first (maybe with an insanely good AI ?) and then tune some machines controls to make them more "newplayers-friendly"/easier to use.

     

About the system itself : Here is the main part of the discussion, and I will try to provide usefull propositions to bias I found in the ML ranking system :


 

  • We figured out that the value of each matches are strangely calculated as it seems the first matches you participate to gets you more (positive or negative) points even if the differential between you and the other team is small (so it values participating to 1/2 matches fully and leaving after reaching 1450 points, it happened a lot during ML SA).
    Proposed solution :
    -> All matches regardless of the occurence # number should earn as much point (But the difference between winning a "high-ranked" team or a lower ranked one should subsist, so it would be better to re-calculate entirely the tie-breaks point bonuses for the phase after each match until the end of the phase).

     
  • By extension, the actual number of points of each teams should not be displayed, but only the current ranking of the phase, so the proposition I made above is doable and relevant + it does not discourage teams from participating and compete to reach higher ranks because "we will never earn enough".
    Proposed solution :
    -> Hiding phase points

     
  • Having a system where you start at 1000 points and can fall below is bad because it does not encourage players to compete (we had a lot of cases where teams entered the queue only to get the 1000 points and according stage points) and if your team is not one of the favourites/loses 2/3times in a row at start, it discourages a lot to continue the phase.
    Proposed solution :
    -> Starting at 0 points and only winning points according to the team performance would be better (and applying proportional factor if you win or lose, like 75% of a match worth/difficulty if you win and 25% if you lose).

     
  • The average time and scores for each phases were barely taken into account as tie-breaks in case of complete equality in stage points (which almost never happened). It was a nice idea but it should go further (and refine the previous proposition) : about scores, it's not the same to win 3-0 in 2 minutes as to win 3-2 in 1h matches, so there should be a differential of earned points depending on match result and match length. Coefficients are adjustable, but it would be fairer to take into account the score difference of each matches.
    Proposed solution :
    -> If I take into account previous propositions : let's admit a match worth "100 points" won 3-2 by one team : the winning team gets flat 25% of the match worth and the losing team 0% (25-0). Then you apply ratios about the result with a formula like "score number"/"sum of scores of the match" : winning team gets 3/5 of the remaining 75 points and losing team still gets 2/5 of the remaining -> (25+45=70 for the winning team and 30 for the losing team in total). I think having the winners of matches progress 2 to 3 times faster than losers is a good ratio.

     
  • About the time dedicated to the league and match lengths :  there is a huge difference between having Black Lotus ending the match in 2 minutes and an IceBringer duel of 1h.
    About this precise point, I think we should take the elapsed match time into account as a global coefficient to calcul earned points, but not without adding a flat participation bonus for each match you participated to (overwise, it would encourage abuses like winning 3-0 with a defensive team and keeping the bomb for 6 hours long, or even more). Let's take our previous example : your 3-2 match of 30 minutes should get you as much points as 3 matches 3-2 of 10 minutes against the same team, but the difference would be the flat participation points you earn regardless of the result.


    Proposed solution :
    -> Flat bonus for each participated match :
    Case 1 : 1 match 3-2 30 mins : Win Team : 70*100% (match time) + 20 (flat participation) = 90  - Lost Team : 30*100%(match time) + 20 (flat participation) = 50
    Case 2 : 3 matches 3-2 10 mins : Win Team : (70*33% (match time) + 20 (flat participation))*3 = 130  - Lost Team : (30*33%(match time) + 20 (flat participation))*3 = 90

    Basically, the more matches you participate to (and the less resistance you encounter), the faster you earn points, but it's difficult to balance it well with the dispersion of each match length and without having any data myself ?

     
  • Also, as we do not have a lot of teams participating yet, I think we should add points for elapsed time in the ML queue or try to adjust it because it often happens that you have an odd number of teams and you have to wait 10 minutes without matches (while others end their own matches) which impacts your earning points capabilities. However as we still experience group issues and crashes, this bonus should not be too strong until these technical issues are resolved.

 

Let's step out of the technical aspects : I always felt that HMM compstitive scene is about "Winners takes all" rather than an equally scattered prizes for each ranks (apart from ML money prizes).

  • I think there should be an incentive to participate to the ML for everyone without considering the results of the teams : like a huge fame bonus for each ML match (winning 150 instead of 25), a flat prize once you participated to a set number of matches (like earning 2000 Fame if your team participates to 6 matches : 1 match for every hour in the ML seems a good value).
  • And another incentive to do it's best during each phase (to reach top1-top3) rather than the huge bonus ratio in ranking points for being 1st/top3 of the phase : give a heavy metal skin to each players or such rewards meaningful to veterans.
  • (and keep the lotery, it's a nice community event !)
     

As a veteran, I can keep going for 6 hours without major tiredness -even with 250 ping while I was participating to SA ML- but it's not the case of newly made teams : having 4h ML phases instead of 6h would be better, but the system format in which you can still reach top8 when participating only to the last phase should be kept all the same. I am unsure about the required number of phases though, but I think it would be nice to reduce it to 4 qualifier stages to give more room to other unofficial tournaments.

 

There is also another aspect to take into account : the only balanced and competitive map is Metal Gods Arena : I would not mind having different maps during qualifiers or even final phases if all maps were as balanced (Legacy ToS has too narrow parts although the global design was great, Cursed Necropolis has too large main road/lack obstacles and zig-zags, new ToS is too straightforward -coming back to the enemy transporter is too slow/difficult-, the last acid-vat part is too close to the score zone and some walls make the bomb bounce strangely)

 

I will try to make this more reader-friendly AND update this post regularly, but be kind as it's my first post on the HMM forums :D

Edited by Scr@tcH
improved readabiliy
  • Upvote 1

Share this post


Link to post
Share on other sites

Okay, I just thought about something that might be better to balance the "earned points" previous idea I had :

 

The won points for each teams during each matches should be chosen with some kind of ELO system (except, we have to adapt it so you only win points even if you lose) and still being asymetric. But the bonus points for scoring should be shared as such : Doing the average of how many points each teams would win against each others and share with the previous formula I proposed.

 

For instance, let's consider Team 1 being matched higher than Team 2 :

Scenario 1 : Close win

- If T1 wins 3-2, the "ELO system" for winning gives 25 points to the T1 (or 75 to the T2) -> 50 points in average (a % can be applied here to balance further) -> +3/5*50 for T1 and +2/5*50 for T2

->T1 earns 55 points and T2 earns 20

- If T2 wins 3-2, the "ELO system" for winning gives 75 points to the T2 (or 25 to the T1) -> 50 points in average (a % can be applied here to balance further) -> +2/5*50 for T1 and +3/5*50 for T2

T1 earns 20 points and T2 earns 105

 

Scenario 2 : Absolute victory

- If T1 wins 3-0, the "ELO system" for winning gives 25 points to the T1 (or 75 to the T2) -> 50 points in average (a % can be applied here to balance further) -> +50 for T1 and +0 for T2

->T1 earns 75 points and T2 earns 0

- If T2 wins 3-0, the "ELO system" for winning gives 75 points to the T2 (or 25 to the T1) -> 50 points in average (a % can be applied here to balance further) -> +0 for T1 and +50 for T2

->T1 earns 0 points and T2 earns 130

(Match time % and flat participation bonus are to be added like I proposed in my previous message)

 

This way, you conserves asymetric earnings if you manage to win against a "stronger team" (or win a lot if you beat a "stronger team") in terms of rank. Maybe the average value should be set as a reference for the won/lost points and adjusted depending on the current difference of ranks between encountering teams rather than an "ELO system" (and as a bonus it should favor "catching-up" with higher-ranked teams). For instance, if we take "50" as the reference, two teams having the same rank would get the same "win bonus" of 50, and if we match 1st and last team in the rankings, the difference should be something like +/-85 (if 1st team wins, it earns 15 as a bonus, if it's the last team, it earns 185) -> This could even allow late teams to catch-up their delay.

Edited by Scr@tcH
mistyping
  • Upvote 1

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...