Enjoy the game, knowing that those stronger than you have paid their dues and you, too, will grow stronger each day❤️
1 Like
Yeah we had a match like that last war … except we were the newbs 
War match-ups have not been based on cups since the very first matchmaking algorithm. Since then we have had matchups based on:
- Alliance Titan Score (the part of alliance score that has nothing to do with cups)
- Simple sum of team power of top 5 heroes of each player in the alliance
- Simple sum of team power of top 30 heroes of each player in alliance
- Weighted score based on top 30 heroes with extra weight to top 5
- Derived score based on top 30 heroes of each player in the alliance but accounting for non-linearity of Total Power as a strength measure.
- Derived score based on top 5 heroes of each player in the alliance but accounting for non-linearity of Total Power as a strength measure.
Some of these have been run in tandem with random A/B splits to gauge effectiveness. The criteria used to judge fairness has been war outcomes - specifically margin of victory.
Don’t get me wrong. I love the ladder idea. I agree it’s the only solution to this widespread problem. And I also think it brings significant benefit of a fun new game mechanic. (Honestly, I had not even considered the idea of an individual ladder but I like that too.)
That said, I’m firmly in the ‘we’re all in this together’ camp. If you don’t understand why the algorithm are so bad, at least entertain the possibility that you also don’t understand the difficulties inherent in making a good one. There are a TON of factors that go into whether an AW will be competitive or not. An algorithmic solution will need to ignore some of these factors (at it’s peril) or try to account for all of them - perfectly balancing the impact of each on the other. It’s not going to be easy. As evidenced by the multitude of algorithms tried (getting more complex all the time), it’s not a straight forward problem.
1 Like
I completely agree on your assertion. But that is also the biggest problem. Why try coming up with algorithm that would work rather than just making a ladder system?
Cup matching - I was referring to the Raids matchups, not AWs.
It defeats the purpose having cups that are meaningless with regards to Raids experience and fair matches, and at the same time algorithms are not capable of incorporating all needed factors to have a fair AW matching. Some of them cannot be even incorporated, like participation, defense teams and all. That’s my biggest problem - devs are doing things the other way around 
1 Like
We’re continuing to get badly outmatched in war half the time (and slightly outmatched the other half of the time) since the top-30 matchups.
I’d like to point to my other topic on the flaws of team power as a metric to use:
My alliance has enough people who have a decent number of unleveled 4* and 5* heroes that really bump our team power score beyond what the heroes are actually capable of…giving us bad matches if we are matched up against players who have fewer heroes but more of them leveled up.
@mhalttu this might help you guys sort out problems with your team power based top 5 / next 25 matching, if it still doesn’t seem to be working as well as you hoped overall.
3 Likes
11th straight time getting slaughtered, our benches are deep… With unleveled 4s and 5s
Fair match for me. I lost humiliatingly because the first three boards seemed to be on the other side and dispensed gems in every color I didn’t use in that particular team, but the match itself was pretty good.
If I’d had the boards that matched my teams when actually using those teams instead of before and afterwards, I’d probably have won.
I’ll tag @hazard as well, considering our previous discussions.
And I’ll takd @Petri because I think this discussion I’m starting could yield some results…
So, @hazard, I’m changing my mind after giving it a more thought 
Since it ended having a head and a tail, I’ll do it more formally.
Intro:
I think raids cup system has a sense but I think we’re seeing it a bit wrong. It doesn’t show ‘how strong you are’ directly, it shows ‘how competitive you are’ and how active you are. Leave it be, you’ll drop, work on it, it’ll raise. Yes, your firepower and your defence will determine how high you can rise or fall, but still, it’s mostly about your activity.
From that, I came to mind that similar cups system for AW would definitely be good - something which measure competitiveness and activity. Firepower will show itself by time.
Let’s call it WAR CUPS, so different cups.
Conclusion:
I’ll lay down a few examples of usage of the formula I just invented. I think going along that route could be good answer for good AW matchmaking. Good doesn’t mean ‘we’re all winners’ because there couldn’t be two winners.
Notes:
-
War result and number of flags used shows us something and should be used for the cup system.
-
It’s important that players carry value of war cup score with themselves (for easy calculation for leaderboards and matching) but also that they gain it depending on whole alliance participation and success, because wars are a team effort.
-
Winner alliance should significantly jump if there’s so high difference in first match, and non winner should fall a bit, in order to have better chance for more appropriate match next time.
-
All numbers here can be adjusted, I chose them because they so far have some sense and were behaving ok in calculations and edge cases.
How to do it
Initial facts
- alliance war cups = the sum of war cups for each member
- it’s only for people who are opt in for wars, when that option comes
- for simplification here I assume members in each alliance have the same starting war cups along the alliance (we always can calculate the average first and just use that number)
- I’ll round up all numbers, so don’t hold for values here too tight
- for each team we have war cup per player (wc/p), size, war score, flags used
Step zero
we have to start somehow, it doesn’t need to be perfect, it’s just done once, one proposal:
- calculated sometime before war preparation time (0)
- pick top 5 heroes per power for each player, sum them, divide by 10 to lower down numbers → that’s war cups per player (wc/p)
- alliance wc is sum of wc for each player, and you match teams that are roughly in same level, eg in some 200 / 300 points bracket
step for each war
- match alliances few minutes before preparation starts using their wc alliance score (1)
- you gain or lose points by using resulting score as a measure of success and flags as a measure of activity/participation
- score each team won is compared relatively to 2000 (2)
- there’s penalty of 10 for unused flags (3)
formula:
- wc for alliance = size * wc/p
- war points = war score - 2000
- penalty = 10 * (6 * size - flags used)
- gain per player = (war points - penalty) / size // it can be negative b/c penalty
- new wc/p = old wc/p + gain per player
- new team wc = new wc/p * size
Few examples
blue always wins, so it’s easier to check
red highlights some underperformance, bold is just ‘not max’
More clarifications:
-
(0) of course, when exactly it depends how long it will take, that’s on devs to figure it out, but it doesn’t need to be totally latest state
-
(1) all players who enter during the preparation phase should not participate in war to avoid manipulation, and chosen opponent stays, even if alliance kicks out many players - this alliance had to fight with they have now - so maybe some few hours before preparation to have ‘matching time’ (depends how much devs need) and that leaders now that what’s done is done by that time
-
(2) 2000, since that’s roughly how much any team is worth at the beginning, so we assume that decent game is when you manage to kill whole team on the opposite side - otherwise you’re either great or you chewed too big bite and it would be wise to try something smaller
-
(3) if you don’t participate, that’s 180 flags = -1800, so you’ll fall down and if you decide to participate - you won’t be matched with high players immediately, you need to work a bit your way up
-
(3.1) if war loot follows raid arena principle that won’t happen to often - but devs can always put minimal war cup per player value if player’s top 5 heroes team strength is above 1000, 2000, 3000, 4000 etc and mitigate cup drop issue
(btw, they could implement this block in raids as well)
1 Like
I greatly appreciate you hard work and hours you put in this post.
However, I’d like to point out a few things.
Cups are “now” a measurement of devotion and participation but they are considered garbage stat for 99% of the players who drop/climb for the daily quest and don’t give a flying saucer about them.
A simple change of the quest to be 8(10) battles rather than 40 heroes will immediately stop cup drop/climb, give meaning to cups, and at the same time retain them as a pointer for success. Will not break neither the game nor the economy. Paired with significant loot tier upgrades will both incentivize raiding and standings.
It is essential AWs are detached from cups whatsoever. Even for calculating baseline.
AWs are group efforts and should reward group efforts. Meaning alliances should have a different measurement for wars. Wins and loses are so randomly determined (as of current gameplay mechanics) that these stats should be kept only at the Alliance tab (just like raid wins on Player tab).
Instead, it would greatly move things forward, if we had goals in war instead of pure win/lose outcome.
Like:
Wipeout the enemy - grants bonus AM reward chance (personal)
Kill 3 opponents - grants bonus AM reward chance (personal)
Win the war - spawns rare titan next (team)
To make people not cheat for the rewards, defense teams power should be tied to your combined 5 best heroes power (+/- with proper adjustment).
AWs can have War Bonds (or other name) as a measurement and ladder purposes. Those bonds would be raised or lowered based on the teams performance in the war.
All alliances start with 0. Each kill grants 40 War Bonds for the Alliance.
The next war you again start with 0 (!)
Retain Bronze(0+), Silver(600+), Gold(1200+), Platinum(1800+) and Diamond (2400+) brackets.
Make appropriate loot tables for the brackets.
And there you have it all. Competitve Alliances will have to fight hard every time to to keep their standings. Simple.
Match alliances on War Bonds. Give it 5 wars and everything will be set.
3 Likes
I was up there complaining about our latest matchup appearing to be quite unfair, but as it turns out they lost due to participation. This isn’t an inactivity issue either as far as I can see, people were logging in.
Wars (other than for the top tiers, where participation is more consistent) will be lopsided nonsense until opt-in is implemented.
Edit: and yes I know most people here are talking about higher level alliances but I think the solution has to factor in the casual alliances as well.
I was introducing NEW cup system for wars. Just to clear confusion about it. What you call war bonds, I just named war cups. And calculated it differently than you. As I said, I think devs should come, because we could really offer some interesting ideas. I think yours is really good as well.
Comparison with raid cups was just that it shows participation for those who don’t drop, and that probably it’s easier to implement war cup system with reusing some stuff from raid cup system.
BTW, I totally agree with you that significantly different loot per arena + modifying hero kill count chest would stop people from dropping cup in raids. Implementing that would be huge game change and improvement!
1 Like
Horrible always have despite their “patches”. Must have used a new programmer who is intern or graduated. I have learned this game is about gems and money spending which i will not give into anymore! I have the year membership and it is worthless for amount you paid for it. This game algorithm is horrible as well. Winning is not great.
We usually score a fairly respectable 3500-4000. We did the same this war, but were completely outgunned like never before. We were one member away from getting reset twice.
How many eligible players (level 12 and above) in each alliance? What is the total defense team scores of each alliance (eligible members)? I bet the difference is around 10k.
Wow. Just a quick glance at the two: They have only 1 player below 3k and they’re still like 2800-2900. Your alliance drops out of the 3ks at around 20. You might be closer to 15k defense team point difference. Maybe 20k.
Believe it or not, it’s only around 4000. So math wise, it doesn’t look nearly as bad as reality. But I think they must have had a lot more depth. We have very, very few players with a decent team of 30, it sure seemed like just about everyone on their side had 5-6 strong teams. They were just about able to reset us in the 2nd set of flags too.
AW Defense:
Them: 4113, 3696. 3679, 3677, 3573, 3535, 3526, 3497, 3492, 3484, 3472, 3468, 3466, 3466, 3462, 3458, 3457, 3451, 3436, 3428, 3410, 3383, 3354, 3340, 3301, 3254, 3112, 3023
28 members averaging 3465
Us: 4122, 3984, 3930, 3789, 3696, 3682, 3679, 3651, 3608, 3592, 3498, 3464, 3461, 3457, 3457, 3444, 3423, 3414, 3328, 3327, 3243, 3189, 3022, 2904, 2545, 1995, 1756, 1501, 1072
29 members averaging 3215
141 attacks them, 129 attacks us. So we could do better there, but those 12 attacks weren’t going to close that point gap much.
The alliances I looked at (and I think I had the right two) showed both had 30 eligible players? New recruits or players sat out?
We’ve had a couple of new players join in the past day. Perhaps they have as well. But you are correct that both teams now have 30 members.
I still think 4k difference is too much. With as many alliances there are. I think they could get # of elig. play. equal and get within 2k difference. You’re probably right about depth but your larger amount of below 3ks hurt too. They are almost useless no matter their depth against strong defense and the BS defense perks.
1 Like