Alliance Wars Matchmaking (Discussion & Developer Response)


#502

We were matched with a solid opponent this time around.

We had a bit of an upper hand on the defence, our teams were a bit stronger. They had the upper hand in activity, we had 4 players not participating, they only 2.

Both alliances were made up of fairly old members, this could be indicative of a somewhat equal roster depth. Based on the assumption that the longer you play, the deeper your roster grows.

We ended up with a narrow victory: 3.184 against 2.962.

What made this war particularly interesting was that our alliances were partially from different timezones. We woke up this morning with only 4 hours left and over 200 points behind. We did our best to catch up, and they did what they could to stop us.

Needless to say the only thing -not- exciting about this war was the victory reward :wink:


#503

Another loss of 1k for us. Our full team attacked as the start but with over 10 players left to bat, pre opponent had a lead of 400 points. We posted some personal best scores but wasn’t enough. Sick of being annihilate now. Login once a day to say hi and hit the titan but fairly done with this game now


#504

We lost by 750 or so, we just couldn’t close the gap. We never really got that close (except when they were asleep LOL). Was it a bad match? I dunno. Probably on the outside of where I want it to be (300-400 loss feels fine to me) - but I still don’t think we’ve had as bad a war as some have.


#505

Let’s see…We lost by 1,154 points.

Normally I’d say we were slaughtered, but after I pulled the numbers, we were down four people (1 awol, 1 vamoose and 2 vaycays) and had the zero bug to contend with.

4000+ points on a bad day? Hell, that’s a win. :grin:


#506

Another one from the kitchen (posts 192, 314, 407).
Same basic theme for us here. They had overall better defensive teams that in the end we couldn’t overcome. We were basically behind by 500 points the whole way, just like all other top 30 matches.
Their alliance score roughly 15k above ours. Since the new top 30 matching we have never been ranked close to their score. I know that doesnt mean everything but it’s an odd imbalance. Why arent we closer? How is it that we are never paired with someone 15k lower?


They attacked 143 us 146.
4100+ they 1 us 0
39+ they 1 us 0
37+ 1 and 1
36+ 3 and 1
35+ 2 and 1
So they had 8 above 3500 and us 3. Their top 3 were higher than our top.
Similar through the rest of the field
Our benches were not deeper.
They had 3 players that were able to 1 shot 8 of our top 10 guys; not so for us. They 1 shot 25+ of us we about 18, but they could take out stronger teams than us.
We are reviewing the attack logs to see off there is something we could do better strategy-wise We are kicking members who don’t contribute. Hopefully latest update improves power imbalance.


#507

Things are proceeding as planned. Tomorrow’s matchmaking should use the new logic, which is only based on the best five heroes (and the member count). Half of the alliances will 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.

Next week we’ll analyze the results and decide the next steps. You can keep posting your results here if you want to, but we can actually get a more comprehensive picture straight from the database after the wars have ended.


#508

So you are testing two approaches this time? Top 5 only in half and the other half based on the top 30 with a weighting towards top 5?

Thanks in advance for clarifying.


#509

thanks for the reply.

i’m still bit unsure if only-top-5 heroes is the method to use.

since it is pretty clear that getting quite solid first 5 4*/5* heroes is easy (to smash titan and have success in raids), but having solid bottom-25 heroes to do even something noteworthy damage to the other fellows top-5 defense just requires far more.

just comparing top-5 per player just causes your method to fail hard when alliance has players with large variation in game time i.e. how much they lack in solid builds in the bottom-25 heroes.

add: maybe just account the last-25 heroes as negative values to match making strength based on how much are below owning players top-5 average.

sounds good, if i have understood correctly, that you use the full live game server db results, not just beta.

anyhow sounds pretty neat numbers crunching what you’re doing there. :slight_smile:


#510

I m going to quit the game. Got enough off those wars


#511

Thanks for the update.

I’m a bit confused by that, are you saying half will get matched based on a simple sum of the top 5 only and not at all heroes 6-30? And the other half is based on the top 30 but with higher weight given to the top 5?

Thanks.


#512

The initiative sounds good. Last war the other alliance e had a front line of all 3k+ players and we do not have one person in the alliance that is over 3k


#513

I can only say that we we’re mismatched again…lost by about 2000pts. Very funny these wars :frowning:


#514

They aren’t any fun. It’s miserable trying to keep your team motivated when it’s so uneven a match.


#515

No. They are saying all will be matched up based on top 5.

Half will get matched up based on the simple sum of the total Power of the top 5 heroes for the alliance.

Half will get matched up based on a more complicated formula that attempts to balance the fact that the difference between say 300-400 Power is way different than the difference between 400-500 Power even though they are both equal to 100 points difference.

The reason they will split it this way is to evaluate the effectiveness of each method and then decide which formula is better for ensuring even match-ups. It will likely (hopefully) be the first of several A/B splits to determine the best possible approach.

@mhalttu feel free to correct me if any of this is incorrect.


#516

Ugh. it does not need to be that complicated. Just match us with an alliance that is ranked within 20 or 25 up or down from us. It will be a fair fight at least 9 out of 10 times. I freakin’ promise you. The vast, vast majority of players are using some combination of their top 5 to 10 players in raids and titans, right? That’s how alliance ranks are based. It’s essentially the same thing. Will you occasionally get a bad matchup? Probably. It can’t be right every time, but over the long term it will produce more good matchups than bad.


#517

We lost again… As we knew we would.
We had less than 2000 points at the end and they had almost 3700.
Its not a surprise though. Their strength was close to 74700 where their titanscore was more than 41700 of them. Ours is almost 57500, with 30500 being titanscore.

When it comes to members then their 6 best members had more strength than our best player. And out 5 worst players had less trophies than theirs worst…

I really want a button saying “this alliance won’t participate in wars” as long as this happens… It’s really no fun or motivating when we know we don’t stand a chance.


#518

Top 5 is probably better than top 30. I think top 10 would work as well, maybe better.

People are filling out depth pretty quickly, the first month caught a lot of people off guard as far as a hero roster.


#519

Shouldn’t there be more weight towards the 5 heroes on defense, not top 5? My defense team and top 5 heroes are not the same.

Anyways, keep testing. Thanks,


#520

I wonder how complicated it’s going to get by comparing heroes power. Our newer members with still growing teams will naturally want to use their strongest heroes. Our stronger players may choose to use 2 or 3 healers, and not their extra-special heroes. Unfortunately i don’t have any suggestions or solutions I’m just adding my voice to the chorus of… it’s complicated!


#521

Just FYI, we were creamed this past war:
------------------------Trophy Count------Titan Score------Avg Level------Total points
Apocalipse Tuga- -53727 ---------------53666 ----------------31.47 ------ 4143
Us --------------------43563 ----------------45641 ----------------30.85 ------ 2771