How to get Fair and Competitive Wars, and Avoid Engineered Clans

Quote:

---

Originally Posted by CasimirEffect

This assumes that any max TH9 which moves to TH10 is lopsided within 2 months after adding 1 spell, a few wall, and troop upgrades? I find this very difficult to believe.

Regardless, we have seen the same trend before and after our group of max 9s made the decision to upgrade. Thus I posted a lengthy explanation.

Basic responses that assume I am uninformed and do not understand the game are expected, but do not add value to the conversation :) In some cases, it is better to assume others to be informed so that meaningful conversation can occur... with the amount of nonsense that is posted on forums daily, I imagine this is difficult for everyone, but thanks anyway: yes, we have pulled accounts that are not square and fit into a square hole.

The issue is how matchmaking handles (near) max TH10/11 accounts in larger mixed TH9+ clan wars. Our issue is that we are on the wrong side of weight distribution when enemy TH8s are added to the bottom to replace enemy 9s.

---

I don't assume that you are uninformed at all, however, I would like to share my experience with you. My clan was very heavy on .5's, most of our 9's and 10's were .5's in fact, many of them with completely max offense and very little defense at the new th..think 71k th10 with 30/30+ heroes and max offense. After the October patch we were hit with a seemingly never-ending litany of engineers. We started throwing down infernos, xbows, and leveling up our defenses. It took a couple of months, but since we did that we've seen a HUGE change in our matchmaking. As I mentioned earlier in the thread, we literally went a month with consistently-fair wars. Then, one of our old th9's, who hadn't gotten the memo yet, came back, and we hit 3 wars in a row that were all unfair to one side or the other. Literally just that one change made all the difference for an otherwise extremely balanced clan. The 9.5 has sat out the past few wars and our matches are back to being even again. It's certainly possible that this is a coincidence, but there is enough evidence in just that one clan to draw a tentative conclusion from it. Add that in to my experience with a wide range of other bases in my old clan + in forum goat herd, and I feel confident in saying that any .5 other than a very new one can cause you to draw a serious mismatch in war. Sometimes good mismatches if all you desire is war victories, but you'll be far more likely to also draw serious mismatches in your opponent's favor, as well.

Quote:

---

Originally Posted by CasimirEffect

Chang405, thanks for the constructive response. I'll take this back to the team to consider, as we have been discussing whether or not to drop infernos. The concern that our members have is that adding more weight could draw in even more polarized matchups. Is this not what your team has seen since dropping infernos?

It's still odd that 9.5s would pull in additional 11s when they already pull in full 10s. I don't understand how that is mathematically possible. To my main point: it appears that TH6-8 at the bottom allow teams to semi-consistently gain additional 11s up top. Which really has an impact on war outcome. With 3-5 11s up top we don't care that our 9.5s are pulling in additional 10s as that seems to be the case, but pulling in full 11s with a 40/40/20+ and eagle and full 10s is rough.

TankSinatra, Off topic again and consistent with banter these posts are littered with. A few more ellipsis for addition emphasis perhaps? Asserting yourself as an "expert" by straying off topic and citing "feelings" is juvenile at best, but that's the nature of building arguments with faceless forum contributors. You may want to consider politics if you haven't already. As for me, I'll stick to the sciences. ;)

---

To reiterate, my clan has found matchups to be a lot more consistently-fair since dropping infernos (and xbows though they're not as important).

Your 9.5's aren't drawing 11's, what they're doing is making the war mm algorithm think that you have an unbalanced clan, so it's widening the search parameters to include other unbalanced clans. If your 9.5's drop infernos then your war clan will start to look more balanced to the mm and you significantly reduce your odds of drawing clans with 11's sitting on top and 6's on bottom.

Quote:

---

Originally Posted by CasimirEffect

Our experience:
Pre October Update Record: 197-33-2 (84.9%)
Post October Update Record: 31-14-0 (68.9%)


Many details that I won't go into here that are important but not essential, but many of our losses are to clans with additional TH11 offense(s) and/or defense(s).


Some of our members have been disappointed with results and pairings. Many have altered upgrade path based on observation in an attempt to safeguard the clan in future wars. A group of 7 max TH9s have started the move to TH10. In one extreme case one of our max hero TH9s skipped directly to TH11 and began upgrading his warden to mimic one of the opponents upgrade paths which had 42/41/20 max walls, max troops, full TH9 w/ no TH10 defenses. He is no longer with us after being informed he would be benched. I'm sure we are not the only clan to witness this shift in logic brought on by frustration, in fact this is likely the root of the debate: losing to clans a team should not lose to (based on comparative technically competency) or pair with (based on offensive and/or defensive assets) provides incentive for competitive players to seek an edge to combat flaws in the algorithm by copying exploits observed.

---

My clan is quite similar to yours in that our winning % overall is around 85%, but we also dropped down to about 70% post-october with the .5 nerf. We were already hurting a bit before that, however, as so many in the clan were already taking the leap to th10. The opponents in th10-heavy clans are just generally significantly stronger than opponents in th9 heavy clans. We had a bit of a learning curve with our newer 10's, so perhaps some of our more-recent reversion to the mean (20 out of 23 wins) on our winning % is due to an increases skill level of our 10's. However, even if it has had little effect on our winning %, I do think that dropping the infernos and getting consistently-fair and competitive matchups has done wonders for clan morale. If we lose now, it's not because we were out-engineered, it's because we ran into a lvl 13 clan with a bunch of stud attackers and solid bases.

Quote:

---

Originally Posted by ryryshouse

Thats cool. We each are such a small sample that its a blip in the overall scheme. Can you post some rosters or too much? I have posted a bunch..cannot find time for all them
( esp when it is for naught) last war we saw more max th11 on top plus the engineered ♥♥♥♥ at the bottom. was nice.

What we did for awhile was use defensive weights and plot both clans graphs. ( Y axis was weight, X axis was map position )

What we saw/see is lines of each clans weight in the same order, normally the lines are even then stair step down as you go down in TH. Often the lines crossed at some point. so the lines have different slopes. obviously, the sharper the angle is normally the roster with the upper hand. But this was not always the case, sometimes an entire roster would be under the line of the other roster. ( real mismatch occurs in this scenario )

Maybe i will look for a couple, or generate some new ones. It is interesting, and really illustrates the problem visually. Over time no pattern immerged, so we stopped generating this data.

---

Come on man, have you read the OP and followed those guidelines? Even a single big .5 can pull you out of whack, or something as small as letting the search go on too long. I had to restart search 3 or 4 times yesterday, even changed roster from 20 down to 15 to try to get a tighter spread of weights. Ended up with a 15v15 with 2 brand new 11's and 1 mid 8, weights ranged from 90k down to 50k, and that search hit in 15 minutes with an almost exact match th vs th across the board.

Quote:

---

Originally Posted by MossackFonseca

Come on man, have you read the OP and followed those guidelines? Even a single big .5 can pull you out of whack, or something as small as letting the search go on too long. I had to restart search 3 or 4 times yesterday, even changed roster from 20 down to 15 to try to get a tighter spread of weights. Ended up with a 15v15 with 2 brand new 11's and 1 mid 8, weights ranged from 90k down to 50k, and that search hit in 15 minutes with an almost exact match th vs th across the board.

---

I have posted multiple rosters. Did you see any .5s? Nope.

Quote:

---

Originally Posted by MossackFonseca

My clan is quite similar to yours in that our winning % overall is around 85%, but we also dropped down to about 70% post-october with the .5 nerf. We were already hurting a bit before that, however, as so many in the clan were already taking the leap to th10. The opponents in th10-heavy clans are just generally significantly stronger than opponents in th9 heavy clans. We had a bit of a learning curve with our newer 10's, so perhaps some of our more-recent reversion to the mean (20 out of 23 wins) on our winning % is due to an increases skill level of our 10's. However, even if it has had little effect on our winning %, I do think that dropping the infernos and getting consistently-fair and competitive matchups has done wonders for clan morale. If we lose now, it's not because we were out-engineered, it's because we ran into a lvl 13 clan with a bunch of stud attackers and solid bases.

---

Thank you for feedback and comments regarding learning curve are not unwarranted. This is true and we have experienced it in a few wars. The suggestion you are making is also listed as a possible solution to the TH11 matchmaking issue as I previously state. I'm not surprised that your group is back to winning wars given this. Congrats.

However, our clan had a fair roster of mostly max 9s with a few 11s up top prior to October update as a result of retention alone and we faced clans that significantly outgunned us multiple times due to TH 11 MM issues stated in over 1000 of my own words so far. So a group of our max 9s moved to 9.5 together in early-December and we still see the same issue. The next step is to drop infernos which our group is still debating. The fact that our once max TH9s which comprised a predominately fair roster feel the need to speed upgrade to fix match up issues that have occurred since the October update is a separate problem from that of drawing in extra 11s (on one side or the other) that I would is likely a business decision and working effectively (reason for my Big Brother comment from earlier). This thought bothers me more than anything related to balancing. Your clan already had this unbalanced accounts in your lineup with mostly max offense as you said so dropping a few defenses is only a few hours of work, imagine how much tougher this is when your clan mates are starting this journey from TH9. That's all I'll say on this topic as I let it fade from a public forum. Personally, I have very strong opinions regarding this and have acted on them.

One point left out is that perhaps, SC considers a roster which omits TH10s to be lopsided or engineered by design. In our case this is clearly not true, but it would explain why we mostly win or lose as soon as prep day begins since October... (This also serves to incentivize a new group of players with strong investment into the game to jump into that top 10% of the revenue stream, just saying)

Quote:

---

Originally Posted by Skrags

I just wonder if the 'recent war performance' checkbox, as originally stipulated in the March 2016 release notes, has been expanded somewhat.


Food for thought!
Or maybe, I'm just trying to justify why my own clan has lost 2 wars in a row, after a fairly credible post March 2016 war win ratio (similar to yours!)

---

Fair point and its difficult to say whether this feature has been expanded with no announcements from SC. I would say this fits under the definition of 'greater transparency'. The fact that we have to speculate at other variables being tweaked without notice in a game as mature as this is a hot topic indeed.


I probably need to add some details to avoid being another member of the community complaining about losses since that is not my intention:


We don't mind losing to skilled teams with or without a slight advantage on either side as variance is inevitable. In fact, since October 6 of our losses have come to very good clans and either our clans or the opposing clan reached out after war to network and setup future friendly wars out of mutual respect. Then there are at least 4 losses where we where blown out by 2 additional TH11s and a handful of 9.5s where 1 or two phase mass attacks cleared all of our 9s despite enemy 9s having <= 20% clear rate, in one extreme case enemy 9s did not clear a single heavy 9 on our map yet we lost due to the effect 2 eagle artillery and 4 max attacks have on 20v wars. We were clearly the more technically proficient clan, but 2 additional full 11s in the lineup broke our back. We came very close in a few instances, but the discrepancy in required skill level to compete in these instances is a roster balancing issue specific to TH11.


Notice that I have not accounted for all our losses above, there are a few wars that we deserved due as a result of subpar performances and/or to clans which are better than us. It happens, these clans exist, and learning from these losses should be what we all play for :)


It would be wrong to focus solely on losses given your incinuation. Our clan is interested in competition. While we are frustrated with close losses we have endured due to extra 11s at the top, we are equally frustrated with the blowouts other clans have experienced when we are granted the advantage of additional TH11s. Since October, many of our losses are close due to effort put forth by our team and robust strategy elements, but many of the wins in our log are blowouts.


Since October update, we have seen almost a digital result from MM to start prep day with few exceptions: (1) Run a perfect war up top and we have an outside chance to steal a win (0) Plan to attack safe rather than pushing our limits because our opponent has already lost a war that has not begun. Neither of these scenerios promote "competition" as SC has clearly states as a core design principle.


Above is the same story in different words with more detail. I'm happy to keep going if it catches SCs attention: we need a fix for imbalance TH11 offense and defense causes to current MM algorithm and the issue is apparent in multiple forms as this thread has indicated. Whether it is classified as account enginnering, clan roster engineering, or a mathematical oversight in the MM algorithm it makes no difference. I am confident that proper statistical analysis paired with good game design practices will find this to be an issue and a solution is fairly simple to include in existing MM architecture which is said to control for lopsided accounts and overall roster mismatches.

Quote:

---

Originally Posted by ryryshouse

I have posted multiple rosters. Did you see any .5s? Nope.

---

I'll repost some info from the OP:

1. Keep a tight spread of th's - The tighter the better. I generally recommend no more than 3 total th levels, preferably with a relatively even mix of them all, though 4 th levels with all evenly built bases is usually pretty safe, too.

2. Balanced bases - mild .5 upgrading by starting with a few key offensive troops, or getting your queen to 5 or 10 is fine, but once you start getting real offense at the new th then you need to start working on your defense. Failure to do so causes issues with your balanced bases checkbox in the mm algorithm.

3. Don't dump low th's at the bottom of your roster. This puts you into the full on "unbalanced war roster" category. Ie, the mm algorithm views you as roster engineered and includes you with the other engineered clans.

4. Recent war record really matters: as you win more and more, you get a progressively more-punitive war weight penalty. Eventually, in every clan that I've ever been in, this gets to the point that you start matching up with engineered clans, or just regular clans that have a clear advantage on paper over you. I've been extremely active in 3 different clans, one of which has gotten as high as a 50 win streak, and believe me when I say that you'll eventually run into an unwinnable war matchup unless you heavily engineer. I, personally, war to have fun and compete, and I think that most people in this game are of a like mindset. But if it's just all about winning to you, then this is the wrong thread.

5. Restart your war search every 30 minutes: the longer your search takes, the more relaxed the mm algorithm gets on the parameters for your opponent. I have never seen an unwinnable war that hit in under 30 minutes, and the vast majority that I get that fast are extremely even and fair..ie, fun. - I must mention here that a close friend of mine was personally told by Jonas that restarting war searches has a "minimal" impact on your war matchup. However, "minimal" is quite subjective, and sometimes it doesn't take much relaxation of parameters to allow a well-engineered clan to slip into your pool of prospective opponents.

You don't need to come here and have a tent-revival about it and stand up and wave your arms saying hallelujah, but I suspect that your answer lies in one of those things listed above.

Quote:

---

Originally Posted by CasimirEffect

Thank you for feedback and comments regarding learning curve are not unwarranted. This is true and we have experienced it in a few wars. The suggestion you are making is also listed as a possible solution to the TH11 matchmaking issue as I previously state. I'm not surprised that your group is back to winning wars given this. Congrats.

However, our clan had a fair roster of mostly max 9s with a few 11s up top prior to October update as a result of retention alone and we faced clans that significantly outgunned us multiple times due to TH 11 MM issues stated in over 1000 of my own words so far. So a group of our max 9s moved to 9.5 together in early-December and we still see the same issue. The next step is to drop infernos which our group is still debating. The fact that our once max TH9s which comprised a predominately fair roster feel the need to speed upgrade to fix match up issues that have occurred since the October update is a separate problem from that of drawing in extra 11s (on one side or the other) that I would is likely a business decision and working effectively (reason for my Big Brother comment from earlier). This thought bothers me more than anything related to balancing. Your clan already had this unbalanced accounts in your lineup with mostly max offense as you said so dropping a few defenses is only a few hours of work, imagine how much tougher this is when your clan mates are starting this journey from TH9. That's all I'll say on this topic as I let it fade from a public forum. Personally, I have very strong opinions regarding this and have acted on them.

One point left out is that perhaps, SC considers a roster which omits TH10s to be lopsided or engineered by design. In our case this is clearly not true, but it would explain why we mostly win or lose as soon as prep day begins since October... (This also serves to incentivize a new group of players with strong investment into the game to jump into that top 10% of the revenue stream, just saying)

---

I don't know that they consider it "lopsided" per se. However, it seems reasonable that a couple of ~ 100k bases with a huge gap down to a whole bunch of 60-70k bases would be a very tough matchup for the mm to make. One thing that I haven't seen much written about lately in engineering discussions is extreme outliers on top. Often, we will see extreme outliers on bottom in an attempt to cheese the mm and/or hide high weight bases on top. However, this exact same thing can happen in reverse, with a very consistent, solid core in the middle but a couple of pretty extreme outliers on top. People don't usually do that nowadays as it's likely to draw some pretty wild opponents, many of whom will have a pretty big advantage over you.

Phew. So I'd say that the guy who rushed to th11 might not have been on such a bad track for you guys, as long as he slapped down infernos right away and was a previously-max 30/30 th9. Um, and he's a good attacker of course, the jump from th9 to even th10 is huge, I'd hate to contemplate going straight to th11. If I were you I'd probably do one of 2 things:

1. have the 11's focus on upgrading/maxing out heroes for the next couple of months and just warring with the lower th's, with maybe a max of 1 of the 11's in any given war. or

2. leave everybody in as it is, encourage the 9.5's to become full 10's, and do a crapload of FC practicing to learn how take out inferno bases consistently across your lineup.

I know a few people who make it look routine, and are a regular threat to 3* a good th10, but I know a whole lot more people who are very inconsistent and far more likely to get 0* or 1* than they are to 3* a good th10.

Quote:

---

Originally Posted by CasimirEffect

Above is the same story in different words with more detail. I'm happy to keep going if it catches SCs attention: we need a fix for imbalance TH11 offense and defense causes to current MM algorithm and the issue is apparent in multiple forms as this thread has indicated. Whether it is classified as account enginnering, clan roster engineering, or a mathematical oversight in the MM algorithm it makes no difference. I am confident that proper statistical analysis paired with good game design practices will find this to be an issue and a solution is fairly simple to include in existing MM architecture which is said to control for lopsided accounts and overall roster mismatches.

---

I think that I hit this one well-enough above, but this post has me thinking more about it. Look at it like this:

clan roster:
2 th11 war weight 100k each
12 th9-9.5 weight 60-72k
6 th8's weight 48-55k

How is that much different in the mm's eyes from:

12 th9-9.5, weight 60-72k
6 th8's weight 48-55k
2 th5's weight 25k

As I said above, those are very different clan rosters, but each of them has a couple of outliers that are confusing the mm. People who dump 25k bases on the bottom of roster are usually deliberately confusing the mm, and people with big th11's on top are often NOT deliberately confusing the mm, but the result will still be a difficult to find match for both. And when the match is difficult to find, and the mm starts relaxing search parameters...then you get a lot of engineered opponents.

Quote:

---

Originally Posted by MossackFonseca

If I were you I'd probably do one of 2 things:

1. have the 11's focus on upgrading/maxing out heroes for the next couple of months and just warring with the lower th's, with maybe a max of 1 of the 11's in any given war. or

2. leave everybody in as it is, encourage the 9.5's to become full 10's, and do a crapload of FC practicing to learn how take out inferno bases consistently across your lineup.

I know a few people who make it look routine, and are a regular threat to 3* a good th10, but I know a whole lot more people who are very inconsistent and far more likely to get 0* or 1* than they are to 3* a good th10.

---

1. All of our 11s in war are all mostly max (atleast 40/40/20 with appropriate troops and all TH11 structures upgraded). Sitting 11s is an easy fix which I have also stated as a solution. One of our sister clans cut the number of TH11 participants from 4 to 1 due to unforeseen events and a split in clan last month. The result since making the change is a record of 12-1... they struggled with the same issues I have outlined prior to the change, but they have 2-4 TH10s in the lineup with 11s.

Fixing a roster and exiling (near) max TH11 accounts from war lineup is one way to improve war record, but only serves to shield the clan temporarily from a design oversight and likely does more harm than good when these folks get bored and leave the clan benching them. As a TH11 with only walls left to upgrade, I can confirm that Clash of Clans without war is boring.

2. Under discussion over the last 2 weeks within clan. Our 9.5s have been semi-consistently 3 starring low 10s and are capable of 2 stars on 11s allowing our 11s to hit the top 10s. Our issue has not been competency in most cases. If our 9.5s pull in extra 10 that's fine and once it occurs often enough they will agree to drop infernos, but they are pairing with additional full 10s and 11s.

Neither tip addresses how additional TH11 offense and defense can end up at the top or bottom of the chain on either side. Hopefully, we agree that this needs to be fixed.

Quote:

---

Originally Posted by MossackFonseca

I think that I hit this one well-enough above, but this post has me thinking more about it. Look at it like this:

clan roster:
2 th11 war weight 100k each
12 th9-9.5 weight 60-72k
6 th8's weight 48-55k

How is that much different in the mm's eyes from:

12 th9-9.5, weight 60-72k
6 th8's weight 48-55k
2 th5's weight 25k

---

Understood. However, if MM cannot distinguish between TH11 offensive and defensive structures and treat the cases you outline similarly then current iteration of MM is broken and needs a separate case to handle swingy matchups created by TH11 specific structures and troops. Your example provides a perfect situation which might come up given current MM system but should never occur in a system designed to meet the statements provided by SC regarding MM.

You are talking around the main focus of everything I have written. It is easy to pick out a specific instance and develop a counter argument, so I will do the same. There are many instances for a variety of clan compositions where wars end up being unwinnable because of how TH11s stack up thanks to MM. The most obvious instance exists in large clans that are full of max TH10 and TH11 accounts. These clans will pair on occasion with clans which run TH11 top to bottom. The clan with TH10s ALWAYS ALWAYS ALWAYS loses. Is this something we should accept from a product we pay for? Absolutely not!

My point (as an actual engineer irl) is this: If you develop a product and a portion of it breaks as a result of some modification intended to improve the product then you fix the piece that is broken. This happens all the time with software applications due to oversights by the developer and/or unintended use by the user. This is why an individual (actual) engineer or team of engineers owns the application - so that they can fix it when it breaks. Specifically as it relates to clash of clans, the introduction of TH11 has broken MM for over a year so SC has a responsibility to fix it.