War Matchmaking and player point values

Apologies if this has been discussed, I searched and didn’t see anything that addressed our particular problem.

I’m looking for some guidence on how this stuff works b/c it has looked really out of wack the last 4 wars my Alliance has been a part of. I understand that Matchmaking is (likely) done by size of alliance and War Score, is that correct?

What I don’t understand is how we get matched up with Alliances that are WAY MORE powerful than we are. In the example I’m going to give… we have one member with a team over 4000 and we faced a team with 5 members over 4000.

And that’s not the only issue we are seeing… The 4000 member of my alliance right now is me… the point value for defeating my team was 13 points higher than the point value of their highest player who was 167 power higher than I was.

This is what I saw…
Opposing team                            My Team            
Power  defeat pts                          Power  defeat pts
        
4322    33           > 4000        5        4155    38          > 4000        1
4222    32           35-4000      6        3919    34          35-4000      8
4175    32           3000-3500  4        3761    33          3000-3500  4
4049    33           3000-2000  1        3750    36          3000-2000  3
4026    31           < 2000        3        3687    24          < 2000        2
3947    31           Average     3352   3633    33          Average     3194
3858    28           Median       3573   3620    34          Median      3507.5
3673    28                                         3582    33        
3649    28           point sum   509     3516    30          point sum  509
3573    27                                         3499    31        
3531    26                                         3496    32        
3474    28                                         3173    28        
3464    30                                         3168    30        
3434    26                                         2579    22        
3104    25                                         2181    20        
2606    19                                         2038    18        
1747    29                                         1890    17        
1437    12                                         1840    16
1391    11

*defeat pts is the value of “Bonus points for victory.” I did not put the total as it would have taken too much time to go through the battle log and get them all as well as it may not have been accurate due to some 0 point attacks where healing was involved.

As you can see, the player that matches up points-wise with the highest player on the opposing team is at 3761… 561 pionts lower than the opposing player. This does not seem fair at all. If I take out their top 3 guys, I get 97 bonus points while their opposing player would get 105 and not have to work nearly as hard.
This causes all of our battles to need to be against people significantly higher than us just to get the same amount of points. Now I understand that the lower end teams are worth more on their end than ours… (even though their lowest is still far lower than our lowest), But when the higher point totals are so weighted in their favor, and they have 5 above 4000 as opposed to our 1, this puts us at a severe disadvantage. We got completely decimated in this war. We lost by 1000 points and I had the highest point total in the war.

The same thing happened in our previous war. My point total at 4160 was 111
The highest on the opposing side at 4135 was 92
How does 25 power equate to 19 points.

Something is inherently wrong with the way this is working. This does not make for winnable wars and we have seen this happen the last 4 or so wars.

Any info would be helpful. My players are getting extremely frustrated with this repeatedly happening.

EDIT: Additional info: I just realized that we have a player opted into wars that does not have a team set up. Though they are very low level (< 2000). I’m sure that contributes, but it’s hard to believe that someone that low a level could throw off the numbers this much.

Let’s start with the official knowledge we have:

Nevarmaor and I tried to figure out how points are distributed. We had seen different explanations but these never satisfied me as they just didn’t corroborate with the data I’d collected for over half a year.

After some calculating what we can tentatively conclude is points are calculated as 1000 x health contribution to alliance, and victory points as 500 x health contribution to alliance. This post of mine includes all the data I used to determine this. It also works well for Nevarmaor’s set of data in the same thread except for a very odd line which we never managed to account for.

With that out of the way…

It seems you’re trying to make sense of how Team Power matches up to the points worth of the team on the battlefield. Referring back to what I started with, these are two completely different metrics. Team Power isn’t a measure of the amount of health the team has. Effectively, the higher the points the larger the health bar that team has.

That said, given the large spread of Team Power of what you show I can only imagine that your alliance mates fields teams of 3* characters while you’re fielding 4/5*s. This sort of discrepancy will lead to large points variance for the alliance on the battlefield. If you were in an alliance of approximate equals, then you won’t see that much variance in points for a team.

As for the fairness of matchmaking… as I understand it the algorithm is perhaps better when it’s a full 30 versus 30 situation. Anything less seems to lead to discrepancies, which I personally assume is a result of a best fit scenario (even though it doesn’t feel like it). I also suspect alliances with large variance in Team Power between members also introduce matching difficulties, and further think to reduce this discrepancy would require minimising this on an alliance level.

2 Likes

Discrepancies is putting it lightly. In this case, it’s 21 vs. 21 and I’m still trying to figure out how some of these point values were assigned, unless HP was the only metric being used.

Our side…(I guess Boril > Heimdall?)

And their side…

As I explained, and linked, points values are health based. There’s more data in that thread to corroborate this understanding including my tentative understanding of the formula to calculate the points value. And you’d need the health values of every team of the alliance to calculate it properly.

But let’s look at those teams:
Heimdall team (TP4443): HP = 7045
Kunchen team (TP4389): HP = 7000
Aeron team (TP4065): HP = 7118
Boril team (TP4195): HP = 7274
Telluria team (TP3992): HP = 6688

In terms of health, it’s Boril team, Aeron team, Heimdall team, Kunchen team then Telluria team. I see nothing wrong there in terms of points distribution.

The war matchmaking algorithm and perceived fairness is a separate matter.

The war matchmaking algorithm and perceived fairness is a separate matter.

Very true.
And I think the fact that it only uses HP to determine the point values makes it skewed.
I booted the member that wasn’t setting up his team, and the next war was better, but still pretty lopsided.

This should be reevaluated at the very least. If the system is going to use HP to distribute the points, then the pool should come from both teams. So rank everyone involved in the war in order and distribute the points. This may make it so that one team’s point sum will end up more than the others, but if they are that much more powerful, then it should be. But even with that, there are so many other pieces that go into a hero… what good is HP if they have no D?

What’s also frustrating… we never seem to be on the opposite side of this algoithm. We always face teams that have multiple in the 4000s even though we only have 1. And in the past 4 wars, I’ve been the highest point total on either team, even though I’m most definitely not the highest in team strength.

In my example… a 4322 team had the same point total as a 3761 team and less than a 3750 team. IMO, that alone should show that the algorithm seems like it needs some work.

One of the problems with the matchmaking process is we only see one small part of it (potentially one side of it). I’ve been tracking our alliance’s war for, maybe, nine months or so. And have yet to participate in a full 30 v 30 war. We know what metrics are used, according to SGG, which is what I linked above.

Assuming no war shuffling to exploit the obvious flaw of the system, it is possible (with a lot of cooperation from alliance team mates) to know all the values of the heroes and troops which affect the algorithm (the top 30 heroes, and top five troops). Crucially, we don’t know the opposition’s.

Assume further there is no deliberate ‘sand bagging’ of their war defence (from what you provided above), and they put out the strongest team according to whatever defensive strategy they are using, then there’s a wider power spread through their alliance: higher highs, and lower lows.

This only applies to their war defence. Even within my own bench there’s a large power difference between what would be my most powerful team (in accordance to TP) and lowest team (in accordance to TP and the top 30 metric). This is likely to be the case for a lot of players.

The exception is to set minimum terms for alliance entry/membership. We can see some top alliances’ requirements like 30 fully maxed out 5* characters, and level 30 fast mana troops. This would prevent a wide discrepancy of TP spread within an alliance. So in terms of the matching process there aren’t as many anomalies (as far as I’m concerned).

That’s an interesting idea; should consider posting on the idea and requests section if it hasn’t already been mentioned (I haven’t seen it proposed but haven’t searched for it). It might actually be a solution to some particular issues.

One small point is that the character ‘power’ (that is the actual figure used on the character description) does technically factor in the defence, but it is a part of it. It goes quite in-depth as to how ‘power’ is calculated.

One piece of advice I would give is don’t get too caught up in assessing an opponent just on their TP alone; assess the troops they use, assess the characters’ specials and mana speeds. I think there’s evidence on the forums of someone using a team of 3* to take down a team of 5* (at least I’ve seen mention of such a feat).

Someone I know has told me about their luck with one-shots on TP4400+ teams in recent wars. They’re lucky to put together a team around TP3700. Even with my 5* I have trouble getting one-shots on TP4400+ teams.

A lot of info here… thank you. This has sparked some interesting things.

Sandbagging (which I’m assuming is referring to putting in a crappy defense team to keep their point values down) is definitely something to be aware of… But doesn’t the algorithm account for that now?
I know I have seen sandbaggers in the past, but I think there has been a change that accounted for this and that’s why we are where we are… The way it is now, even if a team is sandbagging their point values will still be high b/c it seems, based on your info, like it distributes the point values based on the ratio of your HP to your teams total hp. So sandbagging now would likely hurt the alliance as a whole. Does that seem correct?

Most definitely… My teams range from 3500 to 4100 in a war. For the record, I’m usually not afraid to take on any team with my teams regardless of power. I have, in the past, taken on 4000 teams with a 3500 team, as long as I get a healer (sometimes 2) in there and the board is 1/2 decent, I’m usually fine. Unless it’s Telluria… Sorry… unpopular opinion I know… but she’s still broken and needs to be “slow.”

But that’s another thing they could be doing… running metadata statistics as to what Defense teams that win have in them and adjust point values for “meta” heroes… b/c Telluria definitely seems to be the meta right now. If the statistics show that defense teams with a particular hero end up winning more often (within certain parameters like “first attack” and against a similar power team), then teams with that hero should be worth more. This should not be hard to do from a logic perspective and the change would be gradual over time so there would be little to no benefit of “throwing” a fight to try and manipulate the metadata.

I’ll do that. Thanks.

Right I understand. And it is also sometimes misleading b/c obviously some heroes are better than others even though there fully ranked up status maybe close in Power, doesn’t mean they are actually close in usefulness.

OH most defintiely… Sometimes it feels like the boards are created for disadvantage to you. I usually run 2 colors and get no useful matches far more often than I have “overly” useful matches (multiple well positioned matches or a 5oak).

But I have taken out a 3900 team with a team around 3000 that wasn’t even tailored to the defense… it’s all about how good your luck is when the board is created. (mine’s usually bad :wink: )

Generally it is thought sand bagging does more harm than good. It’s thought it is best to make things as difficult as possible for the opponents. I can think of one scenario where something of this nature might be worth considering: if the opponents’ weakest offensive team is still stronger than your weakest member’s strongest defence team.

That is to say, this player’s team will be one-shot no matter what the circumstances. And even then it assumes the rest of the alliance’s defence teams are much stronger to justify this redistribution of points. Generally it only makes sense if this one player is so much weaker than their alliance mates.

Telluria is still a menace (despite the complaints of being rebalanced).

As a comparative metric Power is an odd one, and can be misleading. Some players manage to make the best of the characters they have, and this can include characters not considered ‘good’. While this is something the system presumably uses to assess our benches as a player it’s not something I really use (except roster sorting). However, from raiding experience there feels like there are power tiers which makes the difficulty more pronounced. But due to characters, specials and troops not all are equal in difficulty. Also depends on your own roster of characters.

And yes, board luck pretty much does trump most other considerations. :laughing:

What I think drives it is that some alliances form, earn a war chest, and disband. Only the same people just start a new alliance and effectively reset the numbers that drive their rating.

So, keep an eye on how long the alliance has been around. You can check by seeing how long ago members joined their alliance. Teams doing what I’m saying here will tend to be high powered teams and they all joined 30 days ago. Something like that.

Oh yeah we’ve been together a while. My alliance is a 0 point entry and open one… but as you can see from the stats above, we have a pretty broad range of players. Most of the people in the alliance came in as a beginner and worked their way up to where they are now. There are a few higher level players that joined and stayed around, but most are here from the start. Many over a year.