Alliance Wars Matchmaking (Discussion & Developer Response)

One of the biggest flaws in war is using hero power to determine war score.

Panther - Gravemaker - Guinevere - Kage - Zeline

Counts about the same as

Obakan - Elkanen - Khagan - Thorne - Vivica

And I tried to find a better way to say it, but it’s completely stupid. A hero’s war score should not be the same as it’s “power” score. Or power scores need to be revisited among 4 and 5 stars.

3 Likes

Thank you for sharing the results. That does look like a valliant effort. I’d have taken a 139pt loss in that situation too. I’d agree that level looks to have played a relatively minor role in predicting the outcomes.

My guess on war score is uneven benches, plus your opponent is very erratic in their flag use. They used 4.6 flags per player, so they obviously don’t have a mandatory use requirement. I get the feeling the algorithm struggles with a high war-to-war variance in flag use by an alliance.

That’s a fascinating point. It certainly explains a lot of the thin bench high-level players out there. Of course, if one raided wisely back in the day, one will have at least had 3-5 of each color on the bench. But that’s still only 15-25 heroes. Still a steep building job.

1 Like

I agree wholeheartedly with your point about inequality in 5* heros at the conference same nominal power level. Way up this thread, there was some discussion about different ways to try to numerically differentiate these heroes. My thought was a points-based rating system for different powers, where the most raid-useful powers got more points than weaker powers.

If you have some thoughts as to how to differentiate 5* heroes effectively so that it could be used in the raid matching algorithm, I for one would be very interested to hear them.

I am sorry, but that argument about bench strength is flawed. Are you seriously suggesting that a low or medium leveled player could possibly have a stronger bench strength than someone who had built teams before the alliance wars? What you are saying may make sense for variations for players of similar levels, but not when there is a difference of say 10 levels. When the alliance wars format was announced higher level players would already have an advantage as they would have already have at least 5 fully ascended heroes whereas lower level players are struggling with putting a decent defense team. Higher level players have just the same opportunity to build up depth as any new player who started when the wars were implemented. If they decided not to spend their energy on doing so is on them.

There already is some variance in fully leveled 5 wrt power rating. I’m not sure anything we did would be perfect, but we started giving even more points to:

  1. Very Fast Mana
  2. Debuffs of any kind
  3. More than 3 special effects when triggered (including link)

I’m sure there are others we could add.

But I’m not sure this addresses some of the more common mismatches. There is a common theme that weak teams just can’t do as much damage as they are rated for.

E&P is a game of thresholds. For example, if I can get my attack greater than your defense, then the odds of an easy victory goes way up. If I can’t, I’ll be lucky to get a win.

If weaker teams can’t get enough attack going to beat at least a reasonable number of the other side, they will be destined to waste many flags to take down one opponent. Even in cleanup, the alliance has wasted multiple flags to take down one. If the other side doesn’t have this issue, then they win (assuming equal participation).

2 Likes

What I’m saying is that a level 35 player who basically started competing with Alliance Wars could easily have a similar strength bench to a level 50 who was basically cannibalizing his or her bench for levelling purposes because they didn’t need the extra heroes.

Levels just simply doesn’t equal heroes. There are people who’ve been playing for a year who don’t have 5 fully ascended 5*.

For instance:

2 Likes

Sure it’s possible 30 got the jump but only as a straw argument not reflect in anything at all that has been show in this thread as of yet.

Look at the reality of what you’re saying.

A player level 40 vs a level 30 still has an advantage in many tangibles as they both got started building a better team on x date. They did not get later notice, at worst, they had a head start…

So 40 would have more energy on any extended log on period… Also the 40 would have on average more ascension items on hand, at the given date… Also 40 would have more hero’s waiting in the wings. 40 would be further along in base building. 40 would have more daily resource generation.

All of these things are natural progression. AKA normal play.
Not one single example of normal play, suggests what you’re saying is accurate or seen in normal play.

I’m not sure how both get the news on the same day and the 30 has an advantage over the 40.

What Balusticballsac is saying echoes my point. How can you say that lower level players would have a higher probability of a deeper bench? Before the wars were announced, unless people made individual choices to have a spread of heroes, all levels of players would be cannibalizing all their heroes to feed their main heroes.
If you are a new player joining after the wars, you may have a scattering of 2 and 3 stars for depth, but you can’t tell me that higher level players can’t match that by the time the new players reach level 12. Not to mention the fact that new players are also concentrating on leveling up their main team to progress on the storylines and events, so their depth would be limited to 1 or 2 stars.

You seem to want the median case to be magic. But random systems like this that demand high numbers of successful outcomes to advance produce very high variances. The middle case of a high variance system isn’t useful for characterizing that system.

Please read the link in my post. I think you’ll find it interesting re: an example of normal play.

Incidentally, if you read the threads on the forum on “the XXX probabilities are too low,” you’ll see myriad posts from people who’ve been playing for a long time and just can’t get their heroes ascended. That’s what I’m talking about in terms of a high variance system.

And if you want examples of level being a poor delineator for breaking players apart? They’re super easy to find.

Are you arguing that these two players need to be several divisions apart?

1 Like

Arguing about what level 12 players do or don’t have on average isn’t relevant to the discussion. How often do you see a level 12 matched against a level 40+?

Gar… consider this we had 25 teams each gives up 300 points in power at the start… over 150 energy. We give up 45,000 points in power…

If 1/2 of your team has a team as good as you… we give up 15 of those 3500 power teams… 15 of your team… for our bench to make up. Do you honestly believe level 20 players have that type of bench?

Your whole staring team doesn’t have 1/3 of that showing… but me as a 40 and my 36 and 34 sidekick have that in bench?

Your team is 6k higher in war score too. And y’all don’t come close.
I may be referring Lady or confused team above? But that puts it in perspective, how off it is.

There are two separate debates:

  1. Does the war matchmaking algorithm need work?

Yes! Amen, brother! I would love to see SG improve it. We’re in complete agreement here.

  1. Does going to a level based system fix the problem?

No. The variance on what level translates to in terms of real resources in this game is just too large for it to be reliable for matchmaking. Level means chances to have gotten resources. It doesn’t tell you how lucky you were with those chances…

2 Likes

It is quite relevant and we see it all the time as our team has a bunch of intermediate and beginner players and we see this type of matchup every war. Also, your argument it predicated on the idea that players joining after the alliance wars were implemented build up their teams differently than those before and therefore have a deeper bench.

I have an idea of the normal variance there too., as far as level, impacts of purchases play in advancement etc.
it’s certainly there.
I can assure you it’s minor and does not extend far out from level in the average case… not talking endless purchases, including endless items. Only player pulls.

Mostly separated by what I would call down time… One feeding non valued players while the other feeds much better hero’s…
Both suffer drawbacks. One in the way of lower hero’s. One in the way of, waste through flavor of the week syndrome.

A gap still exists, but it’s pretty minor through 40.
Beyond that I really can’t say… I would imagine that gap increases as you get higher and the syndrome finds a balance… yet the non purchaser still could lack grade A targets.
But you can’t assume buyer got the best out there either. So again the gap is only in the guarantee of a target vs down time.

There are well in excess of a million active players. You have a vanishingly small sample size with which to figure out the variance. How many players’ full rosters have you seen? 100? 500? 1000? That would still be less than 0.1% of active players. You can’t possibly know what the variance looks like within a level.

The algorithm isn’t matching your lvl 12 players against the opposing high level players.

And you keep trying to put words in my mouth by saying I’ve claimed low level players would have a “deeper bench.” Please don’t.

The closest I’ve come is to say that a weak lvl 50 who used to cannibalize his bench and a moderately strong lvl 35 who focused on war depth could easily have similar strength benches.

1 Like

I am sorry, but when I see people misusing statistics to defend the quirks of this game I get annoyed. All the justifications I see are basically, “you haven’t seen enough to make your opinion relevant”. That is completely wrong. Basically, people defending the game’s “randomness” say that overall everything evens out. For example, on pulling random color heroes, someone may pull 10 reds in a row and complain on the board, but defenders will say, yeah but others are pulling more blue. That is not randomness, if that is truly the case that there is a cap on colors, I.e. overall 20% will be red, and once that threshold is reached, other colors are favored, that is a problem.

Likewise, you are using extremes to make your point, but reject it when others do so. The point is that on average a higher leveled player will have a deeper bench. But you pull out the extremes that you could have a match up with a lower level player has a better bench strength. Sure that could happen and they may meet on an alliance war, but the probability is lower than average.

Here’s the attack history for both Sir Dan and myself from the war. As you can see, our first 3 flags were 1-hit kills…All in high-value players. For Sir Dans last 3 flags, he mopped up partial players. All 3 of these opponents where about 1/2 of their total worth. My last 3 flags were rather miserable as I was trying to take out a particularly nasty tank.

While getting the kill points on a mostly dead team does drive up points, the majority of both our points came from 1-shots in the 1st round. Sir Dan did finish off teams with his last 3 flags - as did several of our opponents on the list - but most of his points came from his 1st 3 flags. As you can see from my scores, I only gained 48 points in the 2nd 1/2. Again, the majority of my points are from killing high-point teams.

Not sure which one you’re referring to…I posted screenshots of each of the top 7 player. Let me know and I’ll post the one you want.

Again, not sure what you’re asking as there was no blowout. The only time I mentioned a blowout was surmising that, since we kept up with these guys who outranked us in player levels, what would happen if we were matched with an alliance with the same player levels. Based on our performance in this war it could easily be a blowout.

1 Like

You’re missing the point is not matching a 12 vs a 35… However the experience points to obtain from a 12 to a 35 is comparable to going from our 36 second match to a 56. Likely is far more. As those levels are into the 600k range vs the 350k per level to increase…
So I’m not sure why you see them as unfair… As you’re the one saying levels are levels and little more.

But the point stands… if you wouldn’t do it in the one, why is it okay with you to match it in the other…
it’s the same sort of variance…

And the attack by the teammate was as I said it would be. Saying they scored more in the first 2 attacks means nothing as they used their strongest teams…

Flag for flag average… they gained more from the cleanup and lesser attacks.
That was not a guess. I knew what you’d find. It’s why they scored more. They actually had targets… insist next time they only attack up by 300 then 600 then 900 and let me know how it goes.
That’s our match

There are several in this thread who have expressed the same opinion as @Garanwyn. If you reread the thread since youve started posting your hypothesis you will find several who have stated the same opinion.

In addition, there have been examples of how to manipulate your system (creating new alts and becoming completely P2W) as well as the flaws in the logic (players who already had a high level before wars were introduced).

There have been several… @Garanwyn is not the only one.

1 Like

Cookie Settings