Alliance Wars Matchmaking (Discussion & Developer Response) MASTER

I finally found the #’s I’ve been looking for, so I can place them side by side without having to estimate how things stack up with each 10 levels played.

I will say seeing the #’s paints the picture far worse than I ever claimed them to be… I was being conservative intentionally to be fair, and remove any bias I might have been accused of…

It actually does match what I saw in my personal experience in adding 5* hero’s, and looking forward as to at what level I should double that #. Not withstanding my 4* comple bench finishes sooner as I must only create 5 less.

:rofl: :joy: :rofl: I’m totally stealing “transitive properties of $@# kicking.”

If it were 1 v 1, it might not actively hurt (although it’s unlikely to help much either, given the wide variation in what the heroes and troops people have within a given level).

But in AW, it would either be uselessly broad (as you point out) or would create worse matches because it would dramatically limit the pool of available alliances to match with. There’s no rhyme or reason to the levels of alliance members. My alliance has members spanning levels 23-58. If we apply a gate that’s at all tight around each of them, there will be very few alliances we could possibly match up with. And if those happen to be unbalanced matches (as they likely will be), we’ll be stuck in unbalanced fights until our members level us into a new matching clique.

@Balusticballsac

How did it turn out?

This discussion got me thinking about how player levels match up in a war. Is there a correlation to blow outs, who ‘should’ win, etc.? So I took a look at our last war. I’m interested to find out everyones’ thoughts on this data:

  1. Are the player levels close enough to be considered a good match?
  2. If not, should my alliance (We Don’t Kneel) have been paired with a stronger or easier alliance?
  3. Who should be the winner on this match-up?
  4. What is the predicted win/loss variance (Even, 100 points, 500 points, etc.)?
  5. How would adding player level into the equation change how these alliances were matched?

Other info:

  1. The War Scores have us within 46 points of each other, so there’s no big mismatch there.
  2. I’m looking at this from the perspective of should these alliances have been matched when looking at the player levels.
  3. I realize flag usage, etc. plays a big role on the final scores. Again, just looking at if these alliances should have been matched.

I love to color-code, so here it is:
Green - Player level is higher than similar opponent
Orange - Player level is lower than similar opponent
Yellow - Player level matches similar opponent

1 Like

We did not crush them… Yet we did win by a safe margin… But turnout changed the score by about 900 point swing. .

Meaning they had 4 of their bottom 5 players sit out. While we did not but will next war…

We offered the lower fruit. While they did not. So they attacked ours to the max allowed… We could not attack theirs as they got pulled.

I will also add we flipped them… They did not flip us… Had they flipped us… We could not have mathematically won…

I Will add the people they sat was also level specific and consisted of their lowest available.

That’s great stuff… That to me does not include any of the mismatch possibly I’m referring to though.
Meaning nobody has outplayed the top matches to double the team and consolidated power…

I will add a link that gives you the experience needed to rank…

At some point between levels you can see how experience doubles another level…

So an example is from 40 to 50, is almost 2x the experience… Allowing the possibility that the 50 has 2x the 5* hero’s in the final tier to the first 40 match… this opens the possibility to all but negate the 40 completely if done in a couple examples.

This mismatch is what I believe is a catalyst for a blowout,
Yet if seen at the bottom can happen as well…

So one team has a load of really low, vs a team of many medium missing the low… as that same doubling effect will be seen in the 4* maxed scenario…

Another thing to look at would be how often a certain rank group is attacked per war… 50’s zero? 40’s 1-2? 30, 2-3? 20’s and below always attached the max available, increased and attacked extra every flip?

Question…

I’m not in either of these alliances but was asked how this matchup occured and the friend says it’s “getting old” so I’m assuming it’s not the first time. And yes “war score” is the number they are looking at. Already ruled out opt outs since opt outs are reflected in the war score. @Petri any ideas or insight? How did this matchup occur?

But can you tell us how many actually opted in on both sides? The team with more players opted in should have a penalty applied to their war core for matching purposes. That snapshot would make it look like the team with the lower war score had fewer players as well…counter intuitive unless the number of members isn’t reflected of the number of opt ins.

1 Like

Fig leaf has 19/19 opted in

Other side idk. But opt outs are typically reflected in war score. They could have had new additions after matchmaking(which is what I’m thinking) but if that’s the case then it would be nice if war scores locked when matchmaking started and didnt change until after war. This isn’t the first time I’ve seen this sort of confusion and definitely not the first time it’s been shown on the forum.

Devs, want players to stop screaming about matchmaking? Preventing this sort of confusion and misinformation would be 1 step in the right direction

1 Like

I don’t think the penalty for a difference in number of opted in players is reflected in war score, but it does impact matching. Or at least that is how I understand it based on the description. Yes, the total point of those opted in currently are reflected, but I don’t think the penalty is.

Also - I agree it would be nice to have a snapshot of the war score used to make the match. It would help us to understand things a bit better.

1 Like

Interesting discussion. Haven’t posted in a while bcs in general matchups have been better the more they refine the system. I used to post after every war, especially in the early days of hero matching.
I think using experience level would be a step backwards.
A level 40 who spends can quite possibly have a much stronger team than a 50 who doesn’t. So if even one data point says “match 50s” and 50 is a spender and the other isn’t, it could be disastrous.
Examples from my own alliance…
Me first, a spender


A non-spender:

In addition to what you can’t see:
I have 11 5* maxed. Lazlo has 6 5* total and none in blue. I have both critical and fast mana troops at level 11+. Lazlo does not have both sets what you see is what you get.
Here is a last shot of an equal level. This person is also a spender and came from a top 100 alliance. In no way does Lazlo have the roster to be part of a top 100, even tho they are at the same “level”

As someone else said… Weighting the top 5 heavier but then also the next 5, and then the next after that might make for further improvements

5 Likes

Inacurrate information has been the leading cause of complaints about matchmaking

They gave us “War Scores” to see how we matchup with opponents so know if it’s a close match

But if the number we reference to determine that is inaccurate or at the very least misinforming, then it’s a waste of code and should be redone

4 Likes

Absolutely agreed thanks for showing visual evidence of that

Hopefully it killed the horse that has been severely beaten

3 Likes

Most everyone is on holiday. Besides, we generally don’t see dev responses. They read and make notes, but a written response is rare.

2 Likes

nevarmaor maybe asking specifically which posts in this thread have the developers response…its gotten long and is hard to find unless you know how to search.

1 Like

Should be a way for mods to pin posts or even make a locked thread showing only the OP, developers responses, and only the closest related posts to those responses.

Not that there is a way, idk if there is, but i have seen instances it would be useful

Go to the top of the thread, then hit “Summarize This Thread”. I get 100 posts, including all the staff posts.

2 Likes

I like my idea better lol

The 40 that doesn’t buy wouid if they grind like me have 4-5 5* like me leveled.

The 40 that buys wouid have 4-5 leveled like me…

The 50 that didn’t buy could/should have 8-10

The 50 that buys wouid likely have an opportunity by that jump to add perhaps 2 more than a 50, that did not…

You can see on the link I gave some of why I’m saying what I am as far as how the actual play time doubles…At a certain point you can assume that down time feeding lesser hero’s is what accounts for the feeding of better hero’s, only when the non buyer has no targets to feed…

But despite all that, items still dictate who actually ascends… So buying hero’s is not enough… most must also buy items, at a very costly rate to keep the cycle moving… or they will level.
That figure quickly moves into the thousands of dollars… Not negating all, but safe to assume most…

If, would, could, should…but no levels are created equal. The days of discussion on it I think we should agree to disagree. Player level does not correlate to hero strength near as much as…just using hero strength. If it was put into the equation, how would it even be calculated… a lvl 50 and a lvl 52 should multiply out and be far apart?? At those levels…good chance good RNG and player skill result in the 50 having a much better roster…yeah very well could.

1 Like

Cookie Settings