Alliance Wars Matchmaking (Discussion & Developer Response)

I honestly think it just need to NOT give priority to number of players.
I understand that match 2 alliances with the same number of players sounds cool, but not if it is worse then pairing them with an uneven number of players buth overall more similar strenght.

I tested it in a 2 members alliance for some time.
We NEVER get paired with a different number of players (always 2, like us) but with huge difference in strenght from one to another.

We totally annihilated people and totally get annihilated sometimes.
Getting from fighting two lv60 totally P2W players to two lv40 F2P/C2P.

It was absurd, just pair us with a 3 or 4 players alliance if you don’t have anything close to us.

If there’s any “fault” in the matchups, I think it has less to do with the algorithm, are more to do with alliances like his.

The system is required to find an alliance with 24 members (eliminates most alliances), with similar heroes (eliminates more alliances), similar troops (eliminates more alliances), and similar war performance (eliminates more alliances) until its left to find two “equal” alliances of some people having 15 5* troops, some newbies who are just starting their lineups, and a few in between.

It’s the top people in these alliances that are skewing opportunities for these new(er) players by refusing to play against people of their own skill set.

1 Like

I agree, unless it’s too unbalanced to allow for that to happen.
Our one 4K team can’t knock off 8 3900+ teams and get to 2 flips…
Their 8 3900+ teams easily when backed by another slew of 3800+ teams, runs the board…

I will post the last war… we used 20 more flags…
This was a far more even match, where we had targets to attack, that actually matched our average of 3400…
you will see the difference as it shows a 2500 point swing, compared in the other match type…
so the other match is pretty much a 1200 point advantage…
yet the system sees them enough the same as to match us with both?
Clearly that’s a flaw…
notice how even the top is in scoring compared to the others I’ve posted. Yet the only difference in this one is closer levels. Where our players and bench could both attack more evenly matched players, throughout an entire tea, not just the bottom 4 people.
image

Aww you should have let me reply by quoting you. There’s enough material and repeated arguments I think I could get away with it for a while.

1 Like

We’re going to have to find you a hobby that doesn’t involve pot-stirring.

It was pretty stinking funny, though. :rofl:

You’re thinking Rook/Brobb :wink:

1 Like

What I see happening is. We can fight against 15 Boldtusk or boril, Kash…
because that’s what we bring into the battle as well.
But they ask instead to take out 15 elana, Genevieve and gravediggers.

We have 2 of those I think across 25 players. But we must kill 15 or often far far more.
That’s just an example…

One time I counted though and it was 18 full 5*
Vs our 2 5*… now we’re adding more as players approach lvl 40… But that’s not just a tall order, it’s impossible to do more than once if your lucky.

Killing my Boldtusk is easy match after match if they can bring in a few 5* for all 6 flags… and I never can return the favor…
still now we can bring a few 5’s into a few battles… so should not be expected to kill an army of them.

A data point for your discussion with @Kayo, for what it’s worth:

Over the last 100 fights with players 500+ TP above me, I have a 70% win rate (I have been keeping a log for a while now, so this isn’t a guess). With the exception of field aid fights, I have no fears about going up against a 3900 TP team with a 3200TP team in war. I’m more likely than not to one-shot them. And I’m level 30, for however you wish to factor that information in.

Right and that’s because you have true power in the color match. As do I and that’s why I’m at the top in every war… I can match or stack my way up…

The others though can’t… instead they must thin that one super power team they built up, as a 33, across the 6 flags to boost the Tp within a range.

Me at a 40 can just stack up 4 red 4* and now one 5… Them though it’s 2 good teams and they are looking for cleanups…

Guy on our team today who buys lots of draws and is a 32… he has a slew of players… I know he had 5 bolds at one point.
but he says… I don’t have any built up red hero’s… so that’s all he really has is a great 5 team, and a barn full of dreams until he gets some experience behind him.

Here is one for you… me just 5 levels ago at a 38 vs me at a 42…and the power difference.

At a 38, my honest strategy was. I’m going to take Oberon into this battle because he has over 800 hp and can survive the one hit that is going to kill the rest of my team, made up of all 3* hero’s.
Hecwas the best on my entire purple team at the time too because of only that.

Yet by 43 I should have Sartana, Sabina, and cyp maxed and Domitia to 4-70… oboron will not make the voyage unless I need a buff remover because I’m forced to attack far above me. But odds are he’ll get 1 shot…

“So bA data point for your discussion with @Kayo, for what it’s worth:”

And yes, I can easily see that happening…
But… you knew I’d add the but. So don’t act shocked…

I will venture that is with his one main team. That’s not just randoms from a bench fighting 500 points up…

My red team could pull that off, but my yellow purple, and green, would have far different results…

Yet my 3/2 red green could also pull that off… as could my 3/2 red blue.

But in war I can only choose that for one battle.
I’m not dismissing the data, I just struggle when someone thinks that is practical data, within the war system…
Unless the players were discussing have the sort of power needed to make it legitimate.
The below 40’s do not, unless we’re talking about data on a single team, and not a roster of players.

Ok, so what I’m seeing, your argument is this:

Let’s pretend you’re a division I college football player. At some point you decided to play on a high school football team that you were eligible for. You lead the team in stats, and all is great.

Then, 5 other college player dudes say “Hey, we should just play high school football, too. And we’ll be on the same team!”

Then you play their team and you get pissed because they should be in college ball, and it’s not fair!

And had just the high school students just played with their regular teams, instead of dudes like you, then there wouldn’t really be what you deem a mismatch.

1 Like

What I’m saying is actually that it should match the way it currently does, by power…
it should the find all the available matches. At all levels. It should add another check that then matches those teams to the closest match in power and level makeup…

This is not about me or what I should match against… It’s about what should be added to avoid what I’m pointing out is happening in current system.

You may not see the pattern if you have a beefy enough alliance as under the current system, you have the required depth to surpass the depth influx… yet teams of a different makeup are still on the way to building to that bypass point…

I’m not asking for any special match or treatment, that match would work exactly as it does now, it wouid simply take more things into account…
My power ratings would not change one bit, nor wouid my match potential, based on power.

The system I mentioned gives everyone the same consideration, so you too would see closer matches… I thought that is what you wanted to see.

Or is it your take they should keep matching college teams to high schoolers?

Your football analogy is perfect in the way you did not say it. So I will.
That’s why they do divisions as in one or two… It’s based on the available pool they draw from at its heart.
Unlike the matches you can see here… The more the level gap, the more likely that players pool from which to draw has grown…

I personally have far more players at 42 than I did at 20… Don’t you.

It seems you want it both ways… as I’m offering an addition that does not take anything from you, and simply matches your team to a closer potential power and development match, and you keep turning it into me not wanting a better match somehow.

After reading your post again it’s clear you want to have the current advantage of the known mismatch…
You don’t want a fix, you just want to keep the flaw in place, because it only affect your alliance by matching you to lower level teams…

You could have just said that long ago…

Ok, I’ll probably regret this, but going to throw my 2 cents in here after trying to keep up with this thread…

My issue with your argument about using the level in matchmaking is that at best, level is proxy (and a poor one at that) for how many and how powerful of heroes you have… So, instead of using the poor proxy for number and power of heroes… why don’t we just use the number and power of heroes!!!

Yes, we all had more heroes at level 42 than 20. What is by no means guaranteed, is that 2 different level 42 players have similarly powered heroes. You could have big spenders at 42 with a ton of HOTMs and event heroes, or someone completely F2P that has had bad luck from TC20 and barely any 5* at level 42.

I could see an argument for matchmaking to take in to account, for example, the top 35 or 40 heroes to take in to account a team’s depth. Even if you have teams with identical top 30 heroes, the team with stronger heroes 31 through 40 will do better in war because they have more flexibility in choosing colors and specials to attack with.

But instead of using player level as a proxy for depth… Just use these power scores.

The problem is, there are a ton of rabbit hole arguments like this we could go down. There will never be a perfect calculation to match alliances because of differences in depth of players within alliances and heroes within each roster. This is why I’m glad SG implemented past war performance in the score.

5 Likes

@Balusticballsac if what you were suggesting was possible, then I would back you 100%. Unfortunately it’s not mathematically or programmatically possible to factor in levels without increasing the power discrepancy.

I’m willing to put that aside and work with you to implement a mock up of your solution so that we can see how it plays out - if you’re willing to work with me for this experiment. If you accept, I will do my best to code the solution exactly as you direct me. All I need from you is very clear instructions on how it should work. I will make sure you have full control of anything you want, including the sample data and any of the variables.

Note that I will be doing this in my spare time so can’t commit to any timelines.

If you agree, please choose one of the following implementations to start off:
Implementation A - Match based on power first followed by levels
Current matchmaking based on “power” is done exactly as is, followed by additional checks to pair alliances that are closer in power.
Pros: In theory, this would never result in worse matches because the matchmaking is done on power first, levels are just a second step.
Cons: Unfortunately this is actually impossible to actually implement. The power based matchmaking presumably outputs a pair - it’s too late to add in levels. I only put this option here because I saw it mentioned earlier - please don’t pick it or our little experiment is over.

Implementation B - Add levels as a factor to existing matchmaking
Current matchmaking based on “power” now also takes levels into account. For this to work, a weight will need to be assigned to each variable to tell the algorithm what matters more: power or levels and to what degree.
Pros: Actually implementable… this is how matchmaking works already with plenty of other variables. Please pick this option - if you’re right then caring about levels to a certain extent will result in better matches… and you can tweak it all you want until you get results you’re happy with.
Cons: This will result in matches between alliances with a larger power difference than previously, regardless of what weight is given to levels. If more levels matter, the bigger the power difference.

Implementation C - Other
I don’t have an option C… if you don’t pick option B, you’re going to need to tell me how you imagine the matchmaking works. And you’ll need to be very precise, because it needs to actually be possible to implement.

1 Like

@Rob_DI

You’re not alone in this opinion. But I don’t think you’re going to convince Balusticballsac with this argument. Here’s how I know that. From a month ago:

And here we are a month later :slight_smile: By all means enjoy the debate if you want. Just don’t measure your success by whether you convince him, or you’re likely to be disappointed.

2 Likes

I had an account in a two member alliance, and it had a match against five.
I think they have to keep the number of players close if they can. They know their scoring of a 10 player alliance the same as some 20 player alliance is going to leave someone way outmatched.

Man, I don’t know why I keep reading this thread. It lost its entertainment value long ago.

The way I see it is:

Alliance A is being matched against Alliance B1 / B2 / B3 / etc (call it B) that has on average a MUCH higher defense TP. Two scenarios present themselves:

a) the Alliance War scores are comparable
b) the Alliance War score of B is significantly higher than that of A

For case a) the ONLY way this happens is if the bench of A has a MUCH higher TP average than the bench of B. In that case B likely exhausts their high level heroes in 2-3 flags and are trying to fight with the low-level heroes they have left. It will be a nail-biter but the war could go either way.

For case b) there is a serious issue with the matchmaking algorithm that is not doing what it is supposed to be doing and seriously needs to be addressed directly to the devs, for example in a bug report, rather than here where it is getting lost in the noise.

Note I have tried to avoid addressing any specific examples here as I don’t feel like getting dragged back into this debate (and won’t be). It IS after all, only a game.

3 Likes

I actually wrote a while back, on here that I work on the user end, and often get accounting software written for my needs, inside existing systems…

So I figured that’s why you did not go the route I suggested. As the system was exporting specific totals and could not adapt…

But a few questions on what you can, can’t do and you’re tacit of the system.

Can the system break portions and bring back more than one figure to match to on the power end…

Meaning can it separate into groups of say the top 5 in an alliance then the next 5 and match to more than one total… instead of only using one power total each alliance?
20 players wouid yield 4 totals each?

If so can it then be programmed to only match to a certain groping of that total. Be it the 1st, 2,nd… so forth…

Another option may be to purge a certain amount or range of undeveloped hero’s … not allowing them into the total or the battlefield.

I know it weighst the first 5, heavier. I’m guessing because it assumes that as your repeated defense play. So it singles those as if it’s more… perhaps that check extends out a bit further as you increase levels? 5 at lvl 20… 6 at 28 and so on?
Is that viable.

Here is where you prove that you read comments biased towards what you want to see, if you read them at all

which is really very insulting to the person you comment to.
I was NOT agreeing with you. I was agreeing with @Garanwyn.
You twist and turn your arguments, depending on what is being pointed out to you. You continue to ridicule and insult players that are trying to show you where your convictions do not hold true.

Well, you have done yourself quite a disservice. I don’t think the developers will take someone’s suggestions seriously, after following where you have taken this thread. I know I can’t take you seriously.

5 Likes

Cookie Settings