Alliance Wars Matchmaking (Discussion & Developer Response)


#310

The sudden jump ahead this morning could be explained by a number of things. Perhaps timezones were advantageous to you, so you landed a lot more attacks than them? Or your alliance has better roster depth than them, enabling you to get more out of the second leg?

Or your alliance has better defence teams than them: having better defence teams tends to express itself as a particular advantage in the second leg of the war, when most players will have spent their 3 strongest teams and are now facing your defences with their 3 weakest ones.

Adding higher weight to the defence team however, I expect will improve matchmaking.


#311

19 Respawns for us. Not really helping that they had only 8 defense teams (their alliance had 30 players fyi), and we are still building on titans, but yeah.


#312

2018.05.24 War Report: :skull_and_crossbones: -783

  • Titan Hunters Italy

    • Titan’s Score: 42248
    • 27 man alliance
      • (2 not attacking + 3 empty spots)
    • AW def power: 78611
  • The Bearded Rooster

    • Titan’s Score: 43147
    • 30 man alliance
      • (100% all active members)
    • AW def power: 82900


#313

Miss-matched wars are due to alliances matching on the particular metric used to make the match, but a difference in some other key metric that the developers did not take into account.

1st war matching method, “Alliance Score” : works fine if every alliance member is raiding at their full potential and the alliance is taking down the strongest titans possible for them. If anyone drops cups the matching could be thrown off. If the alliance new and they are working their way up to stronger titans, the matching could be thrown off. This also doesn’t take into account that a player may have 30 strong heroes, as raids and titans only use 5 heroes at a time.

2nd (and current) method “Titan score” : ignores cup droppers and individual raid performance… completely based on how well an alliance kills titans. Newly formed alliances with strong members could be able to down 8-star titans, but if they are still on 3 or 4 star titans because they haven’t worked their way up, they’d be matched up against much weaker alliances who can only take out 3 or 4 star titans. This method also fails to account for players with deep strong rosters as 5 strong heroes can contribute as much to titan score as a roster of 30 strong heroes.

3rd war matching method, “Top 30” : This matching method does ensure that each matched alliance has very similar total attack power across all 6 potential teams of each alliance member over the course of a war. Unfortunately power is an inaccurate measure (unlevelled 5 star has almost the same power as a maxed 3 star, but the 3 star has higher stats and will perform better). This method also fails to account for defensive team power… alliances with members who have 6 good attack teams will likely lose against alliances sporting 1-2 strong teams (and a strong defense) due to the balanced alliance’s inability to overcome the strong defenses. (defenses don’t rotate out, strongest defense is always on while strongest attack can only be used once).


Each of those three methods have ways in which improper matches occur. Best thinking I can come up with is that a balanced war match up should take into account offensive AND defensive capabilities. I don’t know how those would be accurately balanced when an alliance has a range of player levels and strength levels… it would always be easier if everyone was at the same maxed strength level, but that is not how it is in the game.


#314

Thanks for having a look into this.

Our alliance: Prometheus Heirs, 26 people, Trophies: 49395, Titan 52589
Opponent: Wolves of Winterfell, 30 people, Trophies 58868 (over 60k at time of match), Titanscore 51046.
Result 2617 (us) - 5011 (them).
We had this 2000+ difference in all wars based on titan score (I think we lost 9 in a row before version1.12).
The two wars with the new 30 Heroes system very very nice and close (200 points difference).
Now back to the pointless matching again.
Additional info: no we are not using mercs for the titans, but a significant number of people playing Wu with Mana potions and Tornadoes, giving us a good titan record.

Hope this helps to fix it. You may want to consider my proposal to offer a switch to not participate in war, to at least save us the embarrassment of being slaughter all the time.

Thanks and good luck.


#315

Another status update.

We can see from our logs that using Titan Score performs clearly worse than using the team power of the top 30 heroes. The median experience is quite similar between the two modes, but on the 90th and 99th percentile the number of total massacres was much higher when using the titan score.

Please keep in mind that this whole thread was about how the new matchmaking (based on team power) was worse than the old one (based on titan score). The latest war verified that this is not the case.

We’ll bring the team power based matchmaking back for the next 1-2 wars. Hopefully after that, we’ll have the next iteration available. When it is, the current plan is to try using only the top five heroes and compare the results to earlier wars. We are planning to do an AB test where half of the alliances get matched using a simple sum and the other half will get matched based on a formula that takes the non-linearity of the team power into account.

After we’ve digested those learnings, we’ll probably combine the top five heroes to the remaining 25 heroes with some weighting.


Alliance Wars Imbalance
Alliance wars new setup
#316

@mhalttu - you have to use multiple factors. Nobody said Titan score alone was a good predictor. Titan Score and average level of the team would work. I posted above the comparison for our last war and predicted we would slaughter them. I was right. Because the other team was much younger even though the Titan score was similar. Throw in the heroes as a third comparison. Fine because of depth but average level of the team should also be a factor along with Titan score.


#317

Think you cannot plan it better.


#318

These are the type of games we want!


#319

For us this worked better. We have a titan score of 48795 them 45785. When we started their score was higher so looks like they let one go while we took ours.
They went up by 1000 in the first few hours.
We took a 300 point lead with 6 hours to go. They rallied at the end but not by enough:


They had one non hitter we had 4 so we could have scored a bit more.


#320

We are now #11 and our opponent is now #18. Don’t know ranks before, but our titan score is very close.


#321

Why not just put in an opt out feature so that those of us who hate AW don’t have to participate without hurting the alliance. I personally could care less how you match us up, I don’t like it and don’t want to participate but I am compelled to cause I want to remain in my alliance with people I like. You would probably have a lot less whining here on forums as well if you just let us who don’t want to participate choose not to.


#322

@chuck - while I understand your not wanting to participate - but I think it would either necessitate even more complicated procedure for matchups - or if not taken into account, would result in even worse matchups… that would generate even more complaints.


#323

@Witch I don’t think it will make procedure more complicated. Alliance scores won’t match but that’s already the case so… Those AW matching will only work once based on previous AW results. So far my team is more engaged in titans as we have a target: TOP 100 :slight_smile:
I hope they start working on Leaderboard. They have all the data for that.


#324

I hope this will work but i am a bit sceptic what about all the people wo arent participating in war anymore what will it do to the stats when they come back in?


#325

Thanks you for bringing the team power based matchmaking back for the next War.
We have just restarted an alliance and this War was Very bored.


#326

How bout finding a alliance that as a whole don’t wanna do AW? Opt out cause of whatever reason you have for disliking AW takes a spot of a player that would obviously contribute more to your alliance. If your alliance as a whole dislikes AW you could all remove your defense teams and problem solved. I find titans boring but if i didn’t hit them i would be removed in no time, id do the same thing to a member if they didn’t wanna do AW…bye :wave:


#327

So - we did cream them but I just finished calculating things and we had 54 more attacks than they did. If they had hit more, it would have been much closer.


#328

The differences between us are really surprising. For us the titan matching is definitively better than the 30 best teams. We loosed that one but because they played more than us, the matching itself was fair.
If I was able to explain why ! :slight_smile:


#329

Ummm, I actually contribute plenty to my alliance…just cause I don’t like it doesn’t mean I don’t contribute, but thank you for assuming just cause I don’t like AW I’m not doing my part in my group…:roll_eyes::roll_eyes::roll_eyes: