More evidence for broken Alliance War Score algorithm ... post-update


#1

We have just been through two totally imbalanced Alliance Wars. Here are the stats of the two aliances in the current war:

Guardians Ascending:
Alliance score: 69754 [consisting of Trophy score: 34294 plus Titan score: 35460]
War Score: 204508
Heroes 3000+: 7

Slovakia Vikings:
Alliance Score: 79796 [consisting of Trophy score: 40492 plus Titan score: 39304]
War Score: 212305
Heroes 3000+: 13 (with numerous at 2900+)

War outcome so far [07:45h remaining]
Guardians Ascending: 1261 [100 Current War Energy]
Slovakia Vikings: 2368 [77 Current War Energy]

Discussion:

  1. The relatively minor War Score difference explains the matching, but it is totally non-reflective of the relative stengths of the two teams
  2. The components of the Alliance Score [trophies and titans] are in complete mismatch with the War Score
  3. The % 3000+ heroes is in complete mismatch with the War Score, 2) and 3) being mutually confirmatory

Possible Explanation (Discarded):
It could be that Guardians Ascending have been so spectacularly successful in wars that this component results in the War Scores being similar to that of a much more powerful team. This explanation is fatally flawed for two reasons
a) The GAs have recently been in back-to-back wars with two much more powerful teams, having also lost the previous match
b) It is highly unlikely that the SVs are so pathetic at war that their War Score drops to be similar to a much weaker team, because that is not compatible with their ability as reflected in their Titan score
c) The SV is currently winning the war by a margin of over 1100 points

Previous Wars:
a) Although I did not record the stats and the results, the immediately preceding war had similar figures. The opposing alliance had 13 heroes of 3000+ while the GA had 5. The GA lost that preceding war by over 800 points
b) Prior to that, when War scores PLUS Alliance scores were similarly alligned, the GA won three wars on the trot demonstrating that they were at least competent players

Consequences:
Younger players with teams below 2500, plus even the older players, are becoming demotivated by two back-to-back losses of such massive magnitudes. This can be seen from the high level of outstanding war energy which is mostly made up of more junior players not bothering to hit with their weaker teams because of their earlier single digit damage with their stronger teams.

While we appreciate the sincere offort that has been put into improving the Alliance Wars, I cannot but conclude that (assuming that the calcs are indeed accurate) the War Score algorithm is seriously flawed.

I am confident that all of the members of our alliance, which span from 2000 to 3400, would appreciae assistance in avoiding such disappointments in the future.

Many thanks for your attention,
Bedu the Dwarf


#2

Well reasoned through most, but on this one I think its a huge assumption. You have no idea who they have been up against and how they have fared.


#3

Accepted … but enough to drop them to matching a tiddler team in War Score … most unlikely in my admittedly n0000bie opinion


#4

That’s not a high war score. Do you guys have a lot of unleveled 4* and 5* deep in your war rosters as an alliance?

3/50 3* vs 1/1 5* at that level will be comparable on the effects of a war score, but the 3/50 3* will be way more useful.

Put another way, it sounds plausible that one alliance has better developed defense teams without much fluff in their war roster.

Just a thing to look into.


#5

Obviously not … in a war you need 6 teams … no further place to hide high noobie heroes


#6

It’s not really obvious, and it’s December with deals all month. I’d expect a lot of rosters in that range have lots of 1/1 5* causing problems with war scores.

If you have buyers with 23 D teams under 3k, that’s a likely problem.


#7

And the rest of the 3vidence, given that we are mostly n000bs trying to level our A and B teams?


#8

Looking at your alliance, most of your players are below lvl 30. You probably have very few players that can sport more than 2 teams deep of 2/60 5* 3/60 4*.

At this level, 190 points for a 2*, 250 points for a 3*, 310 points for a 4*, and 400 points for a 5* will massively effect your war scores deep in your rosters. Without knowing the full war rosters of both alliances, it’s hard to say. But a solid alliance of leveled 3* heroes would have a huge advantage in that range as far as the algorithm cares.


#9

We all know that the AW matchmaking is still not good. That’s why I do not participate…


#10

Thank you for explaining how it might be possible for a low level alliance to get a high W ar score. It is indeed a
possible explanation for the algorithm’s fatal flaw in our case. So how do we fix it without destroying useful future heroes and yet remedying the reality that you observed of our alliance being weak … and hence mismatched. Seems like way too much weight is being given to unleveled heroes.

What is certain is that we cannot keep on suffering these huge defeats and still maintain player motivation.

Any devs willing to take up a minnow’s cause?


#11

You weather it out, don’t push for leveling 5* early nobody has the mats to ascend. Leveling 5-10 3*, in particular the Atlantis 3* is worthwhile.

It gets better. Next weeks opponent could be easier.

A long time ago we lost like 15 in a row or something.


#12

I’m going to predict you will get easier matchups.

War Matchmaking still isn’t very good at figuring out how strong different alliances are in war.

But the performance adjustment means that if it errs on the side of making you lose more, you will get thrown into easier matches later. Eventually.


#13

It smooths out a bit around 300k


#14

I hope you are right Barry and I appreciate your responses Cruz. I actually posted several weeks ago that I thought the matchmaking system had improved. We lost one war by 2 points and won another by 20. Several others were all close too. War was down to the wire and exciting. We won a few more than we lost during this time but either way it was fun. As my alliance mate Bedu said, the past couple have just been ridiculous. When half their team is a strong as our top 5 there just isn’t much to get the newer players motivated. I appreciate that they are willing to tweak things to make them better as has been demonstrated in the past. Hopefully they make this better soon.


#15

The most recent match was indeed more equitable along the lines that Phelan described. Which of course begs the question: What caused the matching to go so waaaay offline for those two back-to-back wars - half their teams as strong as our top 5. Those two matches were so inefficient that a fundamental flaw must be lurking, waiting to be triggered and to cause mayhem.


#16

Perhaps they were a newly formed alliance, consisting of more experienced players that had migrated from more experienced alliance?
It is possible that there are not an infinite amount of alliances that can be matched at your level and this would probably cause some mismatches to occur.
But you will, of course, not stay at this level, as you and your alliance continue to grow.

Wishing you great matches and exciting wars!


#17

Sad to say, after two close war matches and exciting finishes the same old problem of inexplicable mismatch has returned.

Yes, the war scores were similar - Guardians Ascending (G.A. - us) - 214986; them (NHKBN3NTOP2 - non-Roman) 216175. And, no, they were not a newly formed alliance, neither were they not a dedicated group, as shown by only 10 flags being unused at the end. Our outstanding number (mostly lower level players) was 25 which is not normal, and displays the demoralizing effect of being overwhelmed by an opponent. I presume that such a reaction is NOT what the developers are intending, hence I am posting these details with the constructive intention of providing useful data for their consideration, and am not just having a knee-jerk whinge.

Both teams had 28 players. Quality of defence teams (usually a reflection of overall roster quality): 3600+ (G.A./N-TOP2) 1/2; 3400+ 3/8; 3000+ 9/12 (each segment includes the foregoing). This skewness towards the higher levels amongst our opponent is fairly reflected in the final score of 2893/3503.

Analysis: Given the penultimate wars, with similar war scores, demostrates that G.A. are at worst moderately competent players. This war’s final score shows that the matching war score cannot be due to our opponents being rather useless in past wars. There is therefore only one conclusion, the method of deriving the war score is still flawed, since it appears to given two separate clusters of outcomes - one as desired (close - when G.A. won the past two wars the difference was 5 and about 30), and the other cluster containing differences of many hundreds, with seemingly little in between (but from a small sample).

I will keep on monitoring.


#18

For completeness sake, the Alliance scores were as follows:
G.A.: Total - 76158 of which Trophy was 38766 and Titan was 37392
N…TOP2: Total 76154 of which Trophy = 36985 (???) and Titans 39169


#19

We have seen in other threads that some alliances are taking a dive for a certain number of wars to be able to have an easier time with the next few wars. With that happening it makes it really hard to create a balanced war.

I also have noticed that some teams with similar TP and warscore end up not using 30-40 flags sometimes. The only way they could have a similar score at similar TP and that low level of participation is if they are taking a dive, or if they have players who hit somethings and not others. This also makes it hard to balance.

For sure there could be other issues - and we have identified some, but alliances doing the above makes it really hard to determine when its a system matching problem or people just playing games.


#20

Just went through almost exactly the same story during the last 10 days… The war score does not reflect the real war capabilities of an alliance… not by far…