Flawed Match Making
QuikPik
Member Posts: 817 ★★★★
About 6 seasons ago, Kabam changed their match making algorithm from strictly war rating based to something else. The old system worked well until alliances started tanking in the off season in order to get easier wars at the beginning of the next season.
I've taken the time to look at the alliance ratings for the top 50 in each tier. I know this isn't perfect as a small handful of alliances may have sold champs. But generally speaking it's a good gauge to the relative strength of an alliance. What I have found is shocking.
You're match making system is clearly broken. The top 3 in master looks OK but then it starts to fall apart from there. There are (2) 23m alliances in Master while you have a 64m and 46m in Plat 1. Certainly the 23m alliances in master are not better at war than alliances double their strength in Plat 1.
I think your primary match making criteria is prestige with war rating second. This is why you see so many different strength alliances within the same tier. There is no way that 18/19 million alliance in Plat 1 ever gets matched up against 46 and 64 million alliance ranked lower than them. In a fair system, you would not see such a wide range of alliance ratings within the same tier. Look at Gold 2, there's a 43m at #7 and a 12m alliance at #12. In what scenario is an alliance 1/4 the size only slightly worse in war ranking than a much stronger alliance?
In all tiers outside of top 3 master, you have much smaller alliances in Gold 2 on up. I'm not saying these alliances don't deserve to be there as they've won wars to get to their war rating. However, they are not getting matched with alliances at the same war rating and prestige. What you have is the best 4k prestige alliances at war gets a spot in Plat something. And so on it goes through the various prestige ranges.
It's great that you have increased the rewards for war season rankings but you really need to look at how you match alliances. You have also sort of fixed the tanking problem during the off season. At this point the integrity of rankings for seasons is completely gone. Why should larger alliances even care about war when much smaller alliances finish in top tiers and they don't ever get a chance to face them?
Please Kabam give me a reason to believe in wars again...
I've taken the time to look at the alliance ratings for the top 50 in each tier. I know this isn't perfect as a small handful of alliances may have sold champs. But generally speaking it's a good gauge to the relative strength of an alliance. What I have found is shocking.
You're match making system is clearly broken. The top 3 in master looks OK but then it starts to fall apart from there. There are (2) 23m alliances in Master while you have a 64m and 46m in Plat 1. Certainly the 23m alliances in master are not better at war than alliances double their strength in Plat 1.
I think your primary match making criteria is prestige with war rating second. This is why you see so many different strength alliances within the same tier. There is no way that 18/19 million alliance in Plat 1 ever gets matched up against 46 and 64 million alliance ranked lower than them. In a fair system, you would not see such a wide range of alliance ratings within the same tier. Look at Gold 2, there's a 43m at #7 and a 12m alliance at #12. In what scenario is an alliance 1/4 the size only slightly worse in war ranking than a much stronger alliance?
In all tiers outside of top 3 master, you have much smaller alliances in Gold 2 on up. I'm not saying these alliances don't deserve to be there as they've won wars to get to their war rating. However, they are not getting matched with alliances at the same war rating and prestige. What you have is the best 4k prestige alliances at war gets a spot in Plat something. And so on it goes through the various prestige ranges.
It's great that you have increased the rewards for war season rankings but you really need to look at how you match alliances. You have also sort of fixed the tanking problem during the off season. At this point the integrity of rankings for seasons is completely gone. Why should larger alliances even care about war when much smaller alliances finish in top tiers and they don't ever get a chance to face them?
Please Kabam give me a reason to believe in wars again...
30
Comments
This isn't even a one-off problem, there are lots of smaller alliances in the platinum tiers.
That's why alliance rating isn't a remotely good proxy for war strength in this kind of context. Although as you say it *usually* correlates to war strength, that doesn't consider the fact that there are always going to be outliers in both directions, and those outliers are going to show up in the season brackets as obvious exceptions. A lot of rating points are in champs completely irrelevant to alliance war: they aren't used as defenders or attackers. These guys clearly focused on their highest strength champs to the exclusion of all else, which explains their relatively low alliance (and player) ratings. Furthermore the alliance in question has a war rating of 3115. That's a legitimate Platinum 1/Master level war rating you can't get without winning against a lot of strong competition.
There are certainly issues with the match making system, but pointing out that some alliances "don't belong" in their brackets because of alliance rating doesn't support that contention. You can't reach 3100 war rating by somehow matching against inferior competition and just beating them up. In fact, that's *why* alliances don't match against alliances of similar alliance rating. If this alliance can reach 3100 war rating, it would simply obliterate most alliances of similar alliance rating.
If anything, the presence of those alliances are an example of match making working *correctly*. Because those low rating alliances clearly seem to be able to beat alliances of much higher rating, and thus they should be matched against alliances of similar war rating without regard to alliance rating. And nobody complains about this. The complaints, when they occur, happen at the opposite end, where very high rating alliances confront much lower ones and the lower ones complain they are getting outclassed. Those alliances above aren't getting outclassed. They are competing effectively.
2 things though, related to OP.
#1 as DNA suggests this list doesn’t actually show their PRESTIGE (which, if Kabam “does” take strength into account and not just War Rating, it is probably moreso some type of Prestige factor and not just flat-out overall Alliance Rating).
And #2, OP doesn’t actually detail what the Matchups actually were during course of the Season that the list pertained to, so can not accurately state that certain ally's are only being matched up against certain ranges of other ally's, to prove whether that is the case or not.
While you are correct that some lower Hero Rating teams are alts and can have skilled players... but there is a degree of alliances that are actively switching between 2 or so alliances in order to consistently end in a bracket that they might take a loss. This is pitting alliances in lower brackets against an alliance that has much higher degree of skill.
Scores are locked in off season for some of the higher tiers (5 I believe), but something needs to be done to stop shell alliance manipulation. It might not be a popular mechanic, but I think that if an alliance has a sudden increase in Prestige and/or large number of alliance members change, they should be placed in a incubation period and cannot qualify for season rewards for 1 season.
There might be alliances that are adjusting to make a run for a higher run in an upcoming season that might get hurt by this, but if they are merging or recruiting they need time to adjust and practice. Mergers and recruiting will need to suffer, but it will help them when they are allowed to gain and they are fighting more evened matched teams in higher brackets. They wouldn’t be barred from participating in AW Season, but barred from placing in the Season Rank rewards.
The shell game is to start in a say Gold 2 and end up in Platinum 3-4 (which can be done). Because if they start in the tier they are placed in due to their locked War Rating at Tier 5... They would have to face teams that they would normally have to spend against to win. The issue is that Kabam’s system allows this exploit. By penalizing a sudden and larger increase in Prestige and/or number of players (say anything greater than 50% of the alliance is new - the actual % could be higher as this is just a point of reference), it would make it less of an incentive for a whole alliance to switch alliances in the off season due to the missed rewards.
Even my proposed solution isn’t perfect, but at least it can start a little conversation as how to deal with this issue. Shell alliances have simply replaced tanking in the off season, tanking was too much of a problem and this is the same thing (just a loophole with the current system).
I know I might get a lot of disagrees and shots fired, but this type of manipulation is why we cannot have nice things. Between alliances rotating members for units (gifting events), buying AQ donations, and hiring mercy... this is just another issue that makes the game less fun due to dishonest players always looking to exploit the system.
It is unfortunate that human nature dictates that if there is a loophole, people will exploit it. Not addressing it will only cause more people to do it, as it will become assumed that it is accepted and become normalized.
A perfect example is couponing, stores would offer double up value on $0.50 coupons or less and allow multiple coupons. However, Extreme Couponing led to huge exploits, sales and these types of offers could lead to a customer getting paid to actual buy the items or getting a $3 item for $0.50. Seriously, I use to extreme coupon and I know how easy it could be done. The issue is that more and more people started doing it and it was becoming an issue with profits and the customer experience of bare shelves for others who didn’t take advantage.
It is actually from the same mindset and human trait... if something is allowed, it must be okay (despite the fact if it is ethical or not). So, the masses have to suffer stricter rules and regulation because a select handful want to exploit systems selfishly. If nothing is done, it will only grow and be looked at as acceptable behavior.
I just hope we can be open and honest as a community in solving the issue. If these were truly skilled teams, they wouldn’t need a shell alliance or engage in tanking (like they did previously). They are simply trying to consistently end in a tier they could never perform in unless they do these tactics. I think a 1 season rewards penalty would completely stop the act.
The only question is how to systematically identify these events.
The problem the OP seems to be focused on is summarized as:
That's not the shell alliance problem. A high war rating and low alliance rating is circumstantial evidence you're looking at a shell if they are constantly losing. Before the rating freeze was implemented you'd see this during the off season when losing didn't matter. But an alliance with high war rating and low alliance rating that is *winning* during the on-season period is basically by definition not a shell.
To put the issue more simply and directly, most people think it is perfectly fair for two alliances with the same war rating to face each other, regardless of their alliance rating, provided the war rating fairly represents the strengths of the two alliances - in other words, both alliances fairly won and lost to receive that rating. The only time it potentially represents a problem is when that match up is somehow not fair, and it is always not fair in the direction of the higher alliance rating alliance outclassing the lower alliance rating alliance. Meaning: the lower alliance rating alliance gets smashed and loses. So the only wars we should be focused on, when it comes to unfair match ups, are wars in which a high alliance rating alliance faces a low alliance rating alliance and the lower one loses by a mile. When the lower alliance rating alliance wins, that can't be a sign of an unfair match up.
But if you dig deeper a lot of those small alliances are that...small. They sport 4* rosters and are placing in platinum tiers. There's no way an alliance that small won consistently enough against 20m+ alliances to get to their war rating. It's either...
1 - these small alliances swapped in to lose wars in a current shell. This doesn't seem to be the case as these alliances are full. Normally shells have a couple of alts in there for the sole purpose of starting wars to lose thereby lowering war rating.
2 - alliances are modding to win wars. This here I think is the rare case.
3 - alliances are only getting matched against similar alliances in the same prestige range.
I never said alliance rating is used in match making. I said it is more likely prestige based. I've looked at a lot of profiles of individuals within the alliances I pointed out. You have some players not level 60 (even as low as 50) and a lot of profiles that have less than 2-3 R5.
Or is that list from the current Season Rankings to-date ? Which would mean that they are actually beating the other ally's they have been matching against (for however/whatever reason they are being able to do so). **Still would be good to see a list of WHO they are actually playing against, and not just some ranking list without knowing who they are beating.
It’s blatantly clear the way we’re being matched is flawed as it’s night and day the quality you face being a lower rated alliance which is defiantly an advantage.
Doesn’t seem particularly fair but I guess most 25-30mil + alliances which are sitting in G3 don’t really care about rewards from AW as much anymore so there aren’t that many complaints
Obviously nobody can prove what they're using in the algorithm but I can say for sure it changes almost every season and sometimes during a season.
That's not the shell alliance problem. A high war rating and low alliance rating is circumstantial evidence you're looking at a shell if they are constantly losing. Before the rating freeze was implemented you'd see this during the off season when losing didn't matter. But an alliance with high war rating and low alliance rating that is *winning* during the on-season period is basically by definition not a shell.
To put the issue more simply and directly, most people think it is perfectly fair for two alliances with the same war rating to face each other, regardless of their alliance rating, provided the war rating fairly represents the strengths of the two alliances - in other words, both alliances fairly won and lost to receive that rating. The only time it potentially represents a problem is when that match up is somehow not fair, and it is always not fair in the direction of the higher alliance rating alliance outclassing the lower alliance rating alliance. Meaning: the lower alliance rating alliance gets smashed and loses. So the only wars we should be focused on, when it comes to unfair match ups, are wars in which a high alliance rating alliance faces a low alliance rating alliance and the lower one loses by a mile. When the lower alliance rating alliance wins, that can't be a sign of an unfair match up.
Actually, shell accounts might be the problem of the matchmaking system... what happens to the Plat 2-3 that gets left for an alliance in Gold 2? The Plat 2-3 has a high War Rating but would now have a low Hero Rating with alts remaining. The hypothesis is that the system tends to look for a similar Prestige or Hero Rating when matchmaking. So, the shell alliance with lower Prestige and Hero Rating are most likely facing each other and the the ones that are within normal rating/average tend to face each other.
However, in Gold 1-2, you have alliances that have relatively the same prestige, but not necessarily the skill to compete against the Platinum 2-3 players that have entered a shell alliance within the Gold 1-2 tier. It is my opinion the two are intertwined. Fixing the tanking is what is bringing us these shell accounts. This many low ranking alliances appearing in Platinum-Gold is more likely due shell accounts, and not the high level skill to complete at that level (Which would be an Ultra-Rare exception).
Kabam pitched to us that wars was a tournament where anyone can enter, no biases, most "skilled allies" win. Like any tournament ever created since the start of time! Seeing what I've seen now, this is not even close to the case. Kabam absolutely matches based on ally rating and prestige over war rating, 100%. And that is completely unfair because the best alliances do not finish where their skills say they should. War seasons was pitched to us as a "skill" mode. Kabam's matchmaking has moved this to a "what can make us more money" mode. And it's disgusting. I've put my money and time into a game mode where I thought was a fair tournament and once I read this post and did my research, I've come to realize I've been duped. This is not even close to a fair tournament. Such a shame this company.
My alliance has placed in Plat 2 and Plat 3 since season 2. The last couple of seasons, we've had a really hard time getting back to Plat 2. Every alliance we face just seems stacked and we honestly thought that people are just spending more than us. Reading this post and looking at the numbers, that's not even close to the case.
These following numbers are all facts. We are a 34 million rated ally, that's around 150 in the world based on just ally rating. Currently we are placed in Plat 3 number 152. That's place 252 in the world. We do map 7, we are a top 90 ally in AQ, we have guys with 6* rank 3s, we are a bunch of really good fighters. Now look at these numbers from who we have matched up against in war!
40 spots above us in war and 40 spots below us in war, so the 80 closest "skilled" alliances to us, there are FIVE with above a 30 mil ally rating. That means that 94% of the teams that Kabam says are closest to us in "skill" are below a 30 mil rating. So in a fair tournament (where ally rating shouldn't come into play), we should have some matchups with below 30 mil alliances right? WRONG! Check out these insane FACTS.
Here are the ally ratings of the last 10 allies we matched up with (in millions). Going 10 wars ago to most recent. 33, 39, 48, 32, 38, 32, 32, 36, 39, and 36. Remember above what I said? In a fair system where only skill matters (Like Kabam tells us), we should be matching a sub 30 mil alliance at a 94% rate. HAHAHAHAHA, more like a 0% rate! In fact, the war we just had with a 36 million alliance ... they were 600 spots lower than us! In a fair tournament, no ally no matter what skill lever should ever match with someone 600 spots away from them. The only reason we matched was ally rating!
This is absolute proof that war matching is done via ally rating and it's despicable. War is not a fair tournament, it's a way to get the most money out of every ally. And for Kabam to pitch war to us as a fair tournament where only skill matters is just lies. And it hurts me to the pit of my stomach to write all this. Because I loved wars, I honestly thought it was a fair tournament. It's not! All the allies close to each other in WAR RATING fight each other, making the wars close, creating the most amount of spend on each side.
Can this company just make ONE game mode fair? Just one, my gosh, it always has to be about what makes them the most money. I'm so sad that I spent money, time, and resources into the only game mode I thought was completely fair and based on "skill" and nothing else. I was lied to and I'm an idiot for not seeing this earlier. Thank you @QuikPik for pointing this out. I really think Kabam should answer to how much they lied to us about war matchmaking, it's really sad.
That doesn’t seem to be what is happening. This is most likely updates made a few seasons back. There was a lot of complaints of mismatches in AW... I believe Kabam did weight matches for AW to consider Alliance Rating, but I don’t think it was a primary factor. This might be what we are seeing here.
If the teams are always going against similar Alliance Rating, it is possible to get a high war rating and score high enough to place in platinum... yet they will never face a tough match.
I was under the impression that the first consideration for AW matchmaking was to look at the bracket they are in first and match with an alliance within that bracket. With that in mind, I was thinking that an alliance that uses a shell might try to place high on the first couple wars, and then flip. I was thinking the old, now shell would face others in their bracket and tank faster.... but that is not what is happening.
Sorry for the confusion I caused.
I am now seeing what Quik is pointing out.
The current iteration of war matching seems to be different, but in subtle ways. If you're in the middle of the field and don't have "wonky" ratings, you'll likely see those kinds of matches. But if your alliance has atypical war rating for alliance rating and vice versa, your matches can get weird. I believe this is due to the fact that the match algorithm operates in different ranges than it did before to implement the new mandatory match windows. I also think they are using "special sauce" at the very high end because the standard algorithms don't work when the alliances get too thin. In other words, if your alliance has a war rating of 1300, there are probably dozens of alliances with almost identical rating and maybe hundreds close by. But if your alliance has 2900 war rating there might only be a few alliances nearby and very few if any that are exact matches. So the criteria has to change to find matches at all (or the top two alliances would just keep matching against each other) and it is possible that especially way up there match making can get borked.
This "match against the same alliances over and over" potential is actually part of a larger problem. On the one hand, people generally want "fair" matches where the two alliances are similar in war and alliance rating (or prestige). But on the other hand, if you try to find "best" matches all of the time, sometimes the best match is identical to a previous match and you get a lot of rematches, and many people don't like repetitive rematches. But those two criteria act opposite each other, making it difficult to satisfy both.
I think a fixed tournament-style match system would be better as it resolves both problems (it might be the only way to do so in fact), but I don't see Kabam implementing such a thing any time soon.
After about season 8, Kabam changed their match making algorithm to prestige first then war rating second. This worked well for a season or two because no matter how far you tanked, you wouldn't get matched up against a much smaller alliance at the same war rating. This solved all the complaints about unfair matches because if you were an 8k prestige alliance and tanked all the way down 1500 war rating, you would not get matched up against a 6k prestige alliance at the same war rating. Instead you would get matched against another 8k-ish prestige alliance with a war rating closest to 1500.
After a few seasons, the better alliances at war in each prestige range started creeping up the rankings by winning 7 or more wars each season. The entire time only getting matched against alliances within a certain prestige range. This would explain why I have never seen a match against an alliance with more than a 500 prestige difference. Even though a lot of smaller alliances have passed us in war rating, we never get matched with them due to the new algorithm.
What you have now is bands of different prestige ranges within each season ranking. The top alliances in each prestige band will be platinum because they beat up on all the other alliances within the same range. They never play against an alliance with the same war rating if the prestige difference is greater than X.
That would explain why you see all these smaller alliances in the higher tiers in war. That 12m alliance with 4k prestige and 2800 war rating will never be matched against an 8k prestige alliance with the same war rating. Even though they should since alliances with similar war rankings should play against one another. Otherwise what Kabam is implying is that 12m alliance is better at war than all the 30m+ alliances battling it out in Gold.
It is like a NCAA Division II Football Team going undefeated in their schedule and their fans complaining because they are not in a National Title match vs a Division I team. The Division II team has a much weaker schedule against weaker teams and doesn’t deserve the National Title. In almost most cases, the Division I team will crush the Division II with rare exception with Bowl Games where it was a lower than middle of the road Division I vs a top Division II.
Being undefeated, but never seeing a hard matchup, is not evidence that the team should be in tier 5. You need to face other tier 5 teams and prove you belong there, regardless of the strength. If they do not have the roster or defense to compete against a team that is 10-30 difference in a tier, then they do not deserve to be there unless they have the skill to beat a much harder defense. Otherwise, the teams with bigger rosters and have a much harder schedule for AW and are not getting rewards equal to their time and dedication to the game. Their losses are often by a 1-2 death difference against much harder opponents. But these lower alliances can rank higher than them because they never face their true competition.
Update on our next matchup for war just now, you guessed it, a 37 Mil alliance that's 466 spots away from us! LOL, what a joke.
Matchmaking should be based entirely off war rating. It is not a fair system if it does anything else.
To say this is unfair is to say that even if two Chess players have the same ELO rating, we shouldn't be matching 20 year olds against 40 year olds because that's just obviously unfair. That completely ignores what the purpose of ELO rating even is.
The only thing that should matter, at least primarily, when it comes to creating fair matches in alliance war is war performance. Winning means you're better, losing means you're worse. At best alliance rating should be a secondary concern, which the game looks at only after it looks at war rating. I have no problem with saying that if an alliance has multiple potential matches of the same war rating the game should pick one with as close an alliance rating as possible. But if there exists a match partner of equal rating and all such matches have wildly different alliance rating, too bad: that's who you get matched to, because those are the alliances with the same war track record.
This assumes, of course, that you want "fair" matches in every war individually. The idea of "tournaments" has been brought up. Tournaments aren't typically designed to create "fair" matches. They are designed with completely different criteria. For example, in some tournaments everyone plays everyone else. There's no attempt to make the individual matches fair because the more important criteria is to make everyone have to face everyone else so the tournament as a whole is fair. The definition of "fair" is fundamentally different. In other tournaments where that's not possible the tournament can explicitly and by design give advantages to competitors that did more to enter the tournament. That's the whole idea behind the pole position in racing, home field advantage in the NFL playoffs, and the way many tournaments match the highest competitors against the lowest place ones: to give an advantage to the top seeds to encourage competitors to actually compete as hard as possible pre-tournament.
AW is *not* a tournament nor has it ever been pitched as one, in spite of what some others claim. Perhaps it should be, but it currently is not, and in its current form it is the individual match ups that matter not the season globally, because the game has no structure to compose season-wide matching.
There are twelve wars in a season currently, and that's enough wars to match 4096 competitors against each other in fixed tournament-style matching. There's ten times as many alliances, so if tournament-style match systems were used the season would have to be broken up into ten different brackets of 4000 alliances each. Alternatively, you could use this style matching only for the top 4000, and everyone else would continue to use the current system of individually calculated matches. I proposed such a system a couple times in the past.
The reason I say not in Off Season is that ratings are locked for tier 5 and up. Everyone below gets half points. Off season match ups wouldn’t be fruitful to an outmatched team except getting a beating. If anything, they should practice on alliance in similar matchups for practice and to help avoid spending in order to win. It is more of a curtesy, as their WR will not be impacted much. It would be bad enough that they face the higher teams within Season, where the WR is open dropping and normalizing itself within reason.
Look at it from a reasoning that should they truly be outmatched, losing within season can be extremely demoralizing. To allow that in off season could create alliance breaking stress and cause resentment. Just my thoughts, as I like to think on how these types of thing impact human feelings (as I worked in marketing for 12 years and you are always looking at the emotion that best correlates with product you are selling)... if there is a system issue that is preventing high AR/Prestige versus smaller ones, imagine the impact and outrage when these teams start to lose. It could cause people to quit and get frustrated due to a false sense of security... it seems more logic and better for the community to say these “whooping” should only happen within season when the most impact can be made to place teams within their correct tier and move up more organically with the understanding that only in AW might they face a high prestige/AR team that they may not win against.
Though I do see your position on the WR, so your concept is equally viable. I wouldn’t be opposed to your proposal.
Only other explanation I have is they matched us in war and then everyone left before placing and replaced with second accounts