Tried this yesterday and the enemies got stronger. Only good point was that somehow my low point value team and my team with only two troops won more than my strong defense team 
Will try it again today just to make sure!
2 Likes
Matchmaking is terribly inconsistent. My level 380 account is, at the moment, getting all minimum scores in the 5800+ range (up to 8k+) on the leftmost option in Casual PvP no matter how much I refresh. My level 1227 account, during the same time period, saw multiple sub 5000 scores and a couple sub 1000s within 20 refreshes, including one that had a troop so weak that it died from Cursed trait firing. Both have only played a single ranked battle this week. The high level has a defense score of over 10000, the low level has a defense team score sub 4k. The high level invades with max score teams, the low level with ~7500 score teams, tops (some in the 6k range).
At one point, defense team score seemed to affect the frequency at which you saw low score battles (low score team = more frequent low score battles faced, but Iâve always seen maxed or near maxed teams on accounts that have no business even seeing them at all) but that no longer appears to be the case. Also, team scores donât even use same formula when displayed for defense as they do when you are making the team. The introduction of this seems to coincide with when matchmaking went completely off the rails after the Unity port and no longer works in a way that I can predict, so it would be hard for me to thing they are not related at all, even though matchmaking should be server side.
Iâve long suspected matchmaking doesnât even work as intended, but there are just too many factors to prove it.
9 Likes
This is a lot of interesting information, thanks so much! Matchmaking is so obviously broken that I donât see how anyone could even deny it. This isnât something that should just become near impossible overnight, especially for a beginner.
Also, thanks for the laughs about the troop that died from Cursed. I certainly needed it 
7 Likes
Whatâs worse is that the issue seems randomly spread around the player base. And that adds to the problem of figuring out what is actually going wrong. 
Good luck @Argentin.
5 Likes
I am over twice the Argentinâs level and what I post here is not even the easiest set I got this week.
I still use 4xElspeth team on defense (I set that last week and didnât bother to change). I use refresh button quite often (I donât fight goblins and such, because even weak team can get into a loop with a bit of bad luck). At least for a few weeks now I do 3 trophies matches in Ranked PvP up to tier 1 and then do only casual 1 trophy matches. In casual I use Firebird/Rowanne/Fire Bomb/ Fire Bomb so it may contribute to game being fooled I suffer heavy loses each battle, Iâm not sure about that. I canât think of anything else that can contribute to the differences besides a random bug.
1 Like
Donât be obtuse, and stop abusing your Regular status to change the category of the thread as though the topic was wrong. You realise the post history shows who made the changes, right? Lay off or stay the hell away from this thread.
My level is 4 times that of @Argentinâs and it makes no sense that she faces teams with an average of 7,000 team score at minimum when I can find two-digit level opponents or opponents with <5,000 team score in 2-3 refreshes. If you still fail at understanding why this is a bug, I suggest looking into remedial ESL classes in your neck of the woods.
5 Likes
@Argentin I just noticed you had a very high % of defense wins. 62 wins to 26 loses - I donât think I ever had such a ratio. Perhaps the issue lies here. Try setting 4x Fire Bomb defense. There is no real penalty for losing defense battles if youâre not going for the leaderboards (perhaps your pride may suffer a bit 
). You lose a few points but it may be worth a shot to lower defense win ratio to check if itâs going to affect the matchmaker.
2 Likes
I set two extremely weak troops to defense for an afternoon and won more times in a row than I would win in ranked myself lol. Plus @Fuligin has confirmed that he has an absurdly high win rate with his defense team and still gets weaker teams than me in Casual. Maybe Iâm forever cursed for playing ten games of Ranked and leveling to almost 500 in my first week Thanks so much for the advice though! Iâll continue to play with things and see if it helps.
3 Likes
hmm⌠The issue is you should probably not get such win ratio - it just looks too high. Even with my best, evil and griefing teams I were never able to get such scores. Not even close. The best I could do was about 1:1 ratio. Can you check the levels of players who lose with your teams on those defense battles? Is it like you get the weakest enemies on defense and absurdly strong on offense?
I found this extremely unusual too and after I went back to a normal team I mostly lose defense matches (not sure when it will ever even out though). How can I tell which matches were defense?
The biggest thing for me is that this is obviously not caused by anything I have done and cannot be fixed by anything I do, if thatâs the case.
1 Like
You can check levels of those players you faced. Iâm not sure if that matters much but if the players you defeated are high level (like over 1000 lvl) there is possibility that the game registers something wrong and is âgriefingâ you because of that.
1 Like
The matchmaking is a black box, and it seems only vaguely related to your skill. I thought a lot about it a few months ago because I went through some difficulty spikes. Hereâs two quick paragraphs, then what I think it does.
The one thing matchmaking gets right that many people donât is it seems to believe Troops are a better measure of a playerâs power than player level. I can field a 4x level 1 Moa team and it will be exactly as strong at player level 1 as it will be at player level 1500. But if I swap it to an all-level 1, no-traited Gorgotha/Glitterfang/The Dragon Soul/The Worldbreaker, it gets a LOT more powerful, doesnât it? And if I level/ascend/trait those troops, it gets even stronger. None of this is tied to player level at all. (Thereâs a fun separate discussion to be had about what it DOES indicate but letâs table that.) What matters is the only way player level factors into your power is if your team uses a hero.
Thus, I think team power is the strongest factor, but itâs still just one of many. And itâs still flawed. Forest Troll / Nyx / Kraken / Queen Mab rates in the high 9,000s. So does 4x Fire Bomb. One of these teams is seen as so difficult itâs called a âgriefer teamâ. The other is a joke. Team Power is some function of base rarity, ascensions, levels, and traits. But since some fully-maxed cards are much weaker than others, itâs imperfect.
So I think matchmaking weighs several factors.
- It feels like revenge/rival matches get slotted in outside of calculations. If I see a very âtoo strongâ or âtoo weakâ team itâs commonly one of these.
- Next-most-important seems to be if youâre on a streak. If youâre on a winning streak, it becomes more likely you see strong matches. Losing streaks get pity weak matches.
- Next, team power level. More specifically, I think it keeps a rolling average of your team power level. So if you switch to a weaker/stronger team, it takes a while for that to affect matchmaking.
- Player level has SOME impact, but itâs unclear what. I think very low-level players (maybe < 200?) face a limited pool to help ease them into the game, but canât verify.
- The pool is volatile, and made up of Defense teams from players who have been active within some time window we do not understand.
- The pool could be maintained across multiple servers and NOT synchronized.
I think (1) and (2) have the biggest impact. I have a feeling my numbers are out of order and it works something like this:
- At the start, every possible defense team in the pool specified by (5) and potentially (6) is considered.
- (4) is applied, in the sense that if the player is beneath the ânewbieâ threshold a large chunk of the total PvP pool is eliminated as too strong. This is probably a blanket player level restriction, even though in theory a player could sit at level 1 and buy resources until theyâre fielding a maxed meta PvP team.
- (3) is considered. Ideally, the game wants to pick a âlowerâ power level for 1-trophy, âcloseâ for 2-trophy, and âgreaterâ for 3-trophy.
- (2) is considered. If you are on a winning streak, maybe only the 3-trophy pool is considered. My guess is it tries to find other teams that have been on long streaks.
- (4) is applied again. I have a feeling it tries to find matches near your player level for 1 and 2 trophies, and above your level for 3.
- Finally, (1) is applied. A rival/revenge match might replace one of the already-selected matches, which can look very different for the difficulty.
So what happens when you get a long string of âtoo hardâ matches? It could be many things.
- (5) can work against you. If the people who play within the time window are exceptionally weak or strong, this will âtaintâ the pool and make it harder for matchmaking to do what it wants.
- (2) can break the ânaturalâ flow and insert harder/easier matches when you think it shouldnât.
- (6) would mean two players could see an entirely different pool at the same time and have no chance of seeing the same matches.
- Interactions between (4) and (3) are confusing. If we believe in (5), it could be that there are no opponents who can satisfy the rules of both of these rules. In that case matchmaking may choose randomly.
But this is all speculation and potentially wasted time. We do not know the matchmaking algorithm. We can only look at what it does and try to reverse engineer it. But I think itâs clear we all think itâs a little fishy.
2 Likes
I think youâre right but that means Argentin obviously either faces some bug or there is another factor that contributes to matchmaker going wild. I mean I am over 1100 level, I have a streak of 21 defense wins, my defense team consists right now of 4, fully traited Fire Bombs with Team Score of 9,918 points. My main offense team is one made of dragons with a team score of 10,133 (the other one with Rowanne scores 9,683 points). So it doesnât make sense that I face such teams
in casual pvp when Argentin gets much harder ones.
1 Like
I donât feel like defense streaks are considered for the matchmaking. I didnât even include them in my post, though I didnât clarify that âstreakâ means your offensive streak.
Iâm level 1090. My PvP teams are ranked about 9500 and 9300.
I just went through 10 or 15 refreshes. The lowest player level I saw in Casual was 523. The lowest team ranking I saw was in the 7500s. Right now, in Ranked PvP, I have a Revenge match vs. a level 375 player with an 8,154 point team in the 1-trophy slot.
I would love for my Casual matches to play out like yours, but Iâm seeing behavior more like your friendâs: there is no difference between my Ranked or Casual selections, and itâs very rare that my 1-trophy match is a complete pushover.
I guess before this turns into a full derail:
The matchmaking algorithm is a complete secret unless we disassemble the game. Because of this, we canât tell if what it is doing is a bug or intended. Even if it does dramatically different things for different players, that COULD BE by design. So itâs very hard to pin down if this is a bug.
3 Likes
1110 right now, but I was 1109 a minute ago. Thatâs why I said over 1100 
. It just changes really fast when doing explore with a fire bomb team (we won GW on sunday). Anyway I think that using Fire Bombs on offense may be what contributes to me getting easier matches. Like I suffer âheavyâ loses during my fights because I loose 2-3 troops every battle. I donât claim that Iâm right though, itâs just the only possible difference I can think of. I guess we just have to wait for a dev to show up
1 Like
Huh. OK. Something you said is something I didnât consider.
This is definitely true in Guild Wars. That algorithm is âsort ofâ documented in that a dev once outlined the things that are factors in the score, and what things count as âmore pointsâ. One of the factors it considers is how many troops you lose, and you get more points if you donât lose any.
What if all of our PvP matches are scored with something like the GW criteria, but that score isnât displayed? What if thatâs a factor in PvP matchmaking? Itâs a better explanation of the âstreaksâ than what I had.
That said, Iâm not sure I believe Explore matches affect PvP matchmaking. For all I know, they do, though.
2 Likes
I use the same Fire Bomb team in Casual PvP as in Explore. I can farm gold that way very efficiently.
1 Like
Iâm getting lvl 500-1,000 players and mostly teams scores around 7,000-8,000 in Defense. Will have to see how often my farmer and his friends win since I only set them a few hours back. So far they lost to a lvl 500 with a team score of 9,500, won against a lvl 800 with a team of 8,700.
3 Likes
Note that it is likely the âmaxed player finding sub 5k score teamsâ thing that is more likely to be seen as bugged. Based on the original post, the leftmost teams are âmuch weakerâ (by the payouts, Iâd estimate their strongest team comp at close to 9000 score, which all of the leftmost options are significantly below). Any potential fix is likely going to not satisfy anyone.
3 Likes
The Fire Bomb team is what made me realize how difficult my Casual PVP is. I canât use it at all in PVP like people describe. In this case though wouldnât me losing many matches/troops with my stronger team still do the same though? Because I run Gorgotha/Tesla/Skrymir/Frost Archer and usually lose 1-2 fighting in Casual as it is. Then again maybe itâll balance itself out after the reset next week (but I doubt it seriously).
2 Likes