War Matchmaking Iteration

[Dale s]

Very well put. I concur.

 

[Chadwicke]

Above he said that war seach increases by 3.5 percentage every 30 mins by only spinning 25 mins it's too short to Match you need to at the least let it run for an hour and a half that's 11.5 % variance so will still be a good war , new system looks like as stop and start every 1 and half hours to 2 and a half . Longer will be a bigger mis match by 3.5% every half hour so even a 4 hour spin is 24% which is way less than the almost 100% previous war seach was

 

[LordStark263AC]

So you saying we should cancel and restart War search every hour and a half? Chadwicke

 

[Xabar]

I can totally agree that it made the matching worse at it ever was. Currently we got a opponent which was so much stronger that we decided not to fight at all.
It would be perfect if there would be stronger rules to avoid mixing a atomic player with a level1 player or any other way to make it fair (like decreasing the strength if you fight against a player with x-Level below). One additional idea would be that you can decide in the search how strong the matching should be.
Everthing would be better than the current situation.

I really would like to continue playing this great game but if they do not change something I will search for something better.

The sandbags are a problem In the MM. Because Allows compensate the high weight of the 10 maxed bases, with 10 Iron or clasic Agw in a 20vs20WW.

 

[BHG_Muet]

I'm going to copy/paste my comment from the other thread with Dradis as I think it applies to several conversations here about stopping/starting your search.

*tl;dr : stop cancelling your search if you want to not wait as long.*

----
I was asking about stopping/starting because each individual queue entry is a refresh or new search. If you wait in queue for awhile, cancel, and then join again, the system doesn't care how long you waited previously. It's going to start the search over again with tight restrictions around who you can match. So, the "total" wait time across all queues/cancels isn't really a stat that the system uses or that we balance around. We only evaluate individual queue entries.

As for suggestions in how to use the system, that's kind of up to each Alliance. If you're unwilling to accept matches that aren't extremely even, then you may prefer a strategy of cancelling the queue after awhile and re-entering. Doing this is effectively manipulating the system to stop expanding your search to less even matches. As you can imagine, that also means your potential "total" wait time could be very long as you're now only going to match if another Alliance that would be a good fit for you enters the queue. But, if you just want to play, don't cancel and let the system do its thing. It'll try to find you a high quality match and then slowly favor letting you play over match quality. You can kind of think of it this way: The system is trying to find you a fair match from the pool of available Alliances. If you're not matching it's because there aren't any other Alliances in the queue that would be even matches. Eventually, the system begins to favor letting you play so it lightens the restrictions (reducing the potential quality of matches) slowly over time until it finds a match.
----

LordStark, stopping every 25min is a very aggressive strategy. You're effectively telling the system you're unwilling to accept anything but a "perfect" match with that short of a time. I would wait longer as the expansion is slow.

 

[LordStark263AC]

Thanks. Understood. We'll be cancelling and restarting War search every hour or hour and a half.

 

[Xabar]

I'm going to copy/paste my comment from the other thread with Dradis as I think it applies to several conversations here about stopping/starting your search.

*tl;dr : stop cancelling your search if you want to not wait as long.*

----
I was asking about stopping/starting because each individual queue entry is a refresh or new search. If you wait in queue for awhile, cancel, and then join again, the system doesn't care how long you waited previously. It's going to start the search over again with tight restrictions around who you can match. So, the "total" wait time across all queues/cancels isn't really a stat that the system uses or that we balance around. We only evaluate individual queue entries.

As for suggestions in how to use the system, that's kind of up to each Alliance. If you're unwilling to accept matches that aren't extremely even, then you may prefer a strategy of cancelling the queue after awhile and re-entering. Doing this is effectively manipulating the system to stop expanding your search to less even matches. As you can imagine, that also means your potential "total" wait time could be very long as you're now only going to match if another Alliance that would be a good fit for you enters the queue. But, if you just want to play, don't cancel and let the system do its thing. It'll try to find you a high quality match and then slowly favor letting you play over match quality. You can kind of think of it this way: The system is trying to find you a fair match from the pool of available Alliances. If you're not matching it's because there aren't any other Alliances in the queue that would be even matches. Eventually, the system begins to favor letting you play so it lightens the restrictions (reducing the potential quality of matches) slowly over time until it finds a match.
----

LordStark, stopping every 25min is a very aggressive strategy. You're effectively telling the system you're unwilling to accept anything but a "perfect" match with that short of a time. I would wait longer as the expansion is slow.

BHG_Muet. Our last search took 27 min, just one attempt. 30vs30. Very uneven in top list players. The alliance with more SA players, that certainly won, have more low level players, sandbags.
The sandbags must be allowed. But not take them to calculate the war weigths.
also note that the differences in glory were very large in the last 6 wars. 14.000 VS 16.000 16.000 VS 18.000

 

[BHG_Muet]

Hm. a 2000 Glory difference should have been denied at 27min and there should have been a minimal difference between the top/bottom of both of your Alliances. I'll pass this along to QA to look into. Thanks.

 

[Chadwicke]

Yes this new system is alot slower to take the filters off than the previous on e at even 4 hours that's only a 2*% variance as it was before that would be 80+%

 

[Player Killer]

Any chance you can implement a way for for coleaders to see the current queue sizes. This will help with search times and quality matches as leaders can plan accordingly.

 

[BHG_Muet]

The nature of it being harder to get more active players willing to queue together makes it linear. 10v10 is always the most full followed by 15, 20, and then 30.

 

[Cannibals]

You could get people to queue together if you only ran the MM algorithm at fixed times, say every 4 or 6 hours.

 

[FrostMr]

BHG_Muet we just matched against 11 digital and 9 space. It was a very quick match and our lineup is 3 digital, 7 space, 2 cold, 7 atomic and 1 gunpowder.

It doesn’t matter what the algorithm does if war weight is bonkers every time a new age comes out. Surely someone checks this stuff?

Theres no point even to bother filling war bases, wasting blessings and oil or NTGs for the next 2 days.

 

[LordStark263AC]

So we let War search spin for 3 hours continuously without cancel. Now we have an easy War.

 

[Theoneandonly]

Over two wars I maxed two bunkers. I dropped from 11 via 14 to 18 in line up. Definitely the calculation on line up is broken again and pretty sure also the war weight.

 

[Spaceboy]

Very very bad match we have 15v15..completely useless to do the fight, except for the XP

we: 1 CWA, 1 AA, 4 GA, ...
they: 2 SA, 5 AA, ... (all high levels).

We loose 200 against 400.

I understand that finding a perfect match is not possible, but presenting a fight that is just impossible to win from the first second, is useless also. We are loosing 48h of our time, and the other alliance too I guess.

The extended margins are too wide. I do not mind to wait, but only for an interesting match.

 

[boomboomboom]

BHG_Muet
It doesn’t matter what the algorithm does if war weight is bonkers every time a new age comes out. Surely someone checks this stuff?

Yes of course, we are doing this for them each and every time.

 

[Xabar]

Hm. a 2000 Glory difference should have been denied at 27min and there should have been a minimal difference between the top/bottom of both of your Alliances. I'll pass this along to QA to look into. Thanks.

BHG_Muet
In fact, one of out last wars, in one of the alliances where I am, was against the LordStark263AC alliance, 19139 glory, against 17100 of ours (Difference 2039) the search takes 20 min. We dont have Digital players, and just 1 SA beginner.

2 DA, 4 SA, 4 CWA, 4AA, 6 GA, 4 IA, 2 Enlightment, 2 GunPowder, 2 medieval. Against our: 0 DA, 1 SA, 2 CWA, 3 AA, 5 GA, 5 IA, 2 Enlightment, 6 GunPowder, 4 medieval, 2 clasical

 

[LordStark263AC]

That's right. It was a mismatch. 166 Glory. We've won 148 - 136.

 

[Xabar]

BHG_Muet. OBJECTIVELY. Without complaining, just to mark what to improve, the MM has several problems:
-The glory cutoff seams not to work, or at the least 2000 is to much in less than 12 hr. Pair an alliance ranked 300-500vs1-100 or 1000-1500vs300-500, is not a funny war
-The offensive/defensive war weight, is minimizing undervaluing the hihger level players, DA and SA.
-The offensive/defensive war weight is taking the Sand Bags
-The Ages distribution, is not detected correctly.
-When the MM is bad, the glory lost is too much. In a war where one alliance is ranked 300-500 vs another ranked 1000-1500, the 2nd alliance could los 200-300 of glory