If things are how I understand it in the post above about how matchmaking works, then the determining factor is the war score. What happens is the individual war score is modified by a certain amount. I was never able to calculate the intricacies of how the modification works, but another forum member claims to have worked it out (pre-limit break).
From my own calculations, a player should have a base war score determined by their roster and troops. The modifier then changes this war score. If the record remains at 20, then there should be a theoretical maximum a player can achieve. I’m not certain if the war score can be lower than the base score if a player kept losing.
Keep in mind, between wars a player is also likely to strengthen their roster through leveling, emblems and possibly limit breaks. The top five do account for more. E.g. an extra emblem node on a top five hero was worth 3 points, whereas it would be 1 point for others.
So with the modifiers in play, capable of changing the war score up and down, it’s entirely possible for a ‘weak’ team on a winning streak to face a ‘strong’ team on a losing streak. My alliance often hits a point where the opponent alliance is ‘obviously stronger’. What usually happens is we then bounce from wins and losses, until we can string a few wins again.
What many claim is the algorithm is essentially geared towards a 50/50 split for wins and losses. It would take a more strategic (and powerful, I suspect) alliance to overcome those odds.
The issue with matchmaking is its hidden behind a number not everyone understands.
In the Top 100 we collate far more information than SGG give us and essentially we have a Top 100 leaderboard. What it allows us to do is have an idea who we might get next (assuming it’s a 30 v 30) and maybe 5/10 times we get the opponent we thought we would.
However the war block so you don’t get the same opponent again for 5 wars can make it a bit funkier and we still get mismatches at this level too.
Wheat we notice is the more you have under 30 the more likely for a mismatch. There are specialist alliances that war at 29 purposely to avoid playing the Top alliances and then you have others that cap at 24/25 for the same reason but even more skewed in their favour.
The mismatches there are very noticeable. We war at 30 and sit around a 60% win rate. There’s an alliance at 29 players who have just lost (by 1 pt!) and ended a 25 win streak…
Matchmaking is designed for full alliances and on the balance of play you’ll get like matches most of the time. Eventually you rise too high and you get a bigger alliance coming off a loss due to the 2000pt drop they get from that. It’s feels pretty horrible when it happens and it’s designed that at some point it will.
The Top 3 only really compete with each other generally. We just had 3rd vs 12th and 12th got absolutely hosed. That’s the difference in alliances in the top 50. If 30th gets a top 5 it’s a massive mismatch and 30th gets properly hammered.
We are in a 3 match streak so would have footed a tougher opponent but one of the team is on holiday so we are at 29 which means all bets are off.
There’s no magic wand to make it better but generally it’s not working horribly these days. People just think they’ve been mismatched unfairly when they’ve been on a streak and hit an alliance who’ve had a loss or 2. It designed that at some point you get too close to the sun and you get put back in your place.
I’ve seen many claim the algorithm only really works, so to speak, for a full 30 alliance. I have some doubts, though I’ve never really found answers here on the forum.
I thought the likelihood for the full alliance observation is more to do with stability of the alliance and war score range. Also possible is those top alliances are more likely to track data which lends credence to it, but for casual alliances who also happen to be fully active with 30 players but not fully developed rosters. This is more me diving into thought experiment territory here. More practically, I suspect there’s an inadvertent form of selection bias as well when it comes to player roster and activity for a full 30 member war focused alliance.
Dropping down in alliance member numbers, I suspect the war score range to be higher (again on a practical basis rather than theoretical). Perhaps more diversity in activity levels leading to difference in war score ranges and player membership and rosters.
On that note, always nice to see proper discussion of topics and questions.
Light blue is Warcapped which is to say that’s the highest our score can be. A win only adds 200 each time, a loss drops you circa 2000. These are all based on a full alliance of 30.
We can predict our opponents from this but the algorithm still does funky things at times. We got matched against an alliance with a 2k warscore difference a few back, not great for them as at this level 2k is a lot and we roundly beat them up.
Likewise we got a mismatch with Gipsy Danger a while back as they came off a loss to our highest and we got crushed. Utterly crushed.
At this level we see the matchmaking work “ok” which is to say it’s not the best but better than it was years back.
How that warscore is individually made up still has an element of mystery to it though. We’ve best guessed it pretty accurately but the score of an individual is still nearly impossible to calculate due to the hidden war result ranking we each have
I believe Rigs mentioned on the forum that they figured out how the war score was calculated, but I don’t think they ever shared how (not from my searching anyway).
If I could be bothered I’d probably have another go, but it would require a new account to do so and extensive logging.
In my experience its more difficult with alliances with a broad distribution of roster strength. To make it simple, if a new member adds a 5100 defense to a team consisting mainly out of 4500 teams, they can face an enemy with an average of 4700 or 4800 defenses.
The opponents may need 3-4 flags to bring the 5100 team down, but it is overcompensated, if nearly every member with a 4400 team needs 2 flags to bring down the opponents 4700 teams. This effect is stronger with smaller alliances, and alliances with less than 30 members sre more likely to have a broader roster distribution.
It’s why I suspect the data gathered has a sort of selection bias due to the nature of play involved. Roster strength discrepancy is probably a lot smaller at the top of any particular ‘number of participation league’ (i.e. have fully leveled, emblemed, limit broken 5*s) than those towards the bottom.
If some ‘hardcore’ players with fully developed rosters are in alliances who cap at, eg. 25 members, then they’re likely to remain near the top of that ‘participation’ league.
Always interesting to hear more about this stuff. Boy, I miss this sort of discussion.
But this time my complaint, it’s up to war
in my understanding
the pairing of wars
, counts heroes , troops and scores from previous wars .
or am i wrong?
after 3 defeats in a row following the opponents
I see that our opponents have a team of heroes far superior to that of the allies
Oh, not counting the yropas
I will forward a link and if someone could guide me if this is correct
notice in this video the difference in troops
and heroes of the ally that I’m in the first case
and the opponent and the second
I am mid high level player liv 96, mi ally have best spot in top 306 place top 500 now
, after the infamous relase of second limit brake the war is very frustrating moment!!
Mi alliance have “only 6 5*” 2th limit brake hero on 30 player teams,
the enemies * 16 5* 2th limit brake" hero in 30 war team, 10 more!!!
Matchmaking chose too very strong alliance after one defeat whit 1000 pt of distance, seme destiny of this second war
You thinking that a single 2LB on the opposing tank makes a significant difference to your performance tells me all I need to know about your prowess as an alliance
I have 4 one shot on +5400 team and open two +5600, is mid performance for my normaly perform 5 6 one shot in all war,
my team is most powerful of my allyance 4554 power 1 LB hero Aramis,i am only +5400 defense in alliance
I am really curious on how they figure your opponent in alliance wars. I am really getting tired of getting beat 2 a week. We might win once in awhile if they dont attack. Like today my team is 5120, 4124, 4872 and 4152. Their team is 4794 5065, 5245 and 2500. They killed us 1534 and still have 5 attacks left. We usually get beat 1500-2000 pts every fight. Not worth the time. Its as bad as raid attacks today that my 4124 pts team got beat by a 2256 team…really really…wish i had their luck and boards…
Just confused and tired
The Alliance Score is a combination of the Trophies of the members, the points earned from fighting Titans, and the strength and victories in Wars.
The War Score is calculated from the most powerful Heroes and Troops of each Alliance member participating in the Wars, adjusted with the past performance of the Alliance.
The Titan Score decreases over time.
Trophy Score: #####
Titan Score: #####
War Score: #####
Pop out notes attached to each Alliance’s Alliance Score.
Alliance scores are listed in on the alliance page and present the combined totals of an Alliance’s Trophy Score, Titan Score and War Score. The alliance score is used to determine the alliance’s spot on the ranking page.
Alliance War scores are based on the power level of the top 30 heroes for each alliance member participating in wars, then adjusted for past war participation. It is used to help match up alliances for wars.
Roughly it seems to work this way: the alliances are divided into groups of the same size. Then the groups are sorted by war score. And then, the pairs are made of neighboring alliances going from the highest score down. If the number of alliances of the size is odd, then the lowest scoring alliance is paired with highest scoring alliance 1 member smaller. And so it goes until all alliances are paired.
First: look at your defences. The TP number is not as important as the synergy between heroes. Take notes from several wars how many attacks which def team takes to be killed and then adapt your defences so they will survive better.
Second: learn how to compose your attack teams for best results. To win the war you need that more than half of your attacks result in a one-shot kill. The “friendly attacks” can help here - ask your allies to set defences with different structures and try what team works best with it.
You cannot beat bad boards with this, they will still happen, but they will have to be worse for you to fail the attack.
Also: don’t invest all your resources in def team. The attack is more important than defence. For best results you should be able to build 6 attack teams with TP about 300-400 less than your def team. Otherwise your attacks will fail too often.
Third: think about timing your attacks in a war. The best results you get when you attack the weakest teams more times than the strong ones. If you kill the weak ones, wait 6 hours until they revive, kill them again, wait next 8 hours ror them to revive, and only then attack also strong ones, you will probably have better attack success rate.
Speaking from experience, it can end horribly! 