Page 1 of 2 12 LastLast
Results 1 to 10 of 11

Thread: Is the win-trading ban being enforced?

  1. #1

    Join Date
    Sep 2014
    Location
    Digging another tunnel in the data mine
    Posts
    6,660

    Is the win-trading ban being enforced?

    Back in January SC announced a fairly forceful ban on win-trading in CWL in response to some fairly obvious shenanigans that were starting to cause a lot of controversy in CWL: https://forum.supercell.com/showthre...on-Win-Trading

    Quote Originally Posted by Darian[Supercell] View Post
    Win-Trading/Collusion
    When Clans compete against one another in the Clan War Leagues, agreeing to “fix” the results through negotiation, removing defenses, Star trading, or any other means other than playing legitimately is viewed as win-trading. In our competitive ladder system, the Clan War Leagues are intended to test a Clan’s skill in War. Therefore, we are putting extra scrutiny to ensure that win-trading is not tolerated and actions are taken to maintain the fairness of Clash of Clans.

    When a Clan has been identified as participating in win-trading, the offending Clan will receive a 2-season temporary ban from all Wars (Clan Wars and War Leagues). Any Clan receiving this temporary ban will be unable to join the Clan War Leagues and will be unable to participate in normal Clan Wars until the restriction has lifted.
    One form of collusion in upper leagues that's very easy to spot - and what many people were complaining back in January - is the war where both clans agree to give each other max stars, or close to it. At a similar time to this SC also changed the rewards structure so that stars alone didn't earn stars, and so that tanking didn't help - which ended much of the early controversy around these shenanigans.

    One thing I've noticed posting my interim results each month is that there's always been a trickle of these perfect(ish) wars. They get black-flagged in my results list as they are very easy to detect automatically. So easy that I can pull up the list of all 31 wars with scores of 44-44 or better in Master/Champ leagues since Feb in a minute or so. I've been meaning to have a look at what happened to these clans for a while - after all the entire of CWL is public (you can see every attack on ClashOfStats, for example). I finally had a look, and the results are a bit disappointing...
    (I'm filtering to 15v leagues in Master, so there are no high 30v scores getting counted)

    Taking as a start the August season. There was just one colluded war, in M1 league. With a 44-44 score. In case anyone thinks this is natural score, here's the complete dump of day results from that league group:

    War tag ClanA ClanB Day StarsA StarsB DestructionA DestructionB
    #2PVQ8PCLU #88QLGVQJ #PQRPL0R 1 28 22 1200 898
    #2PVQ8PGUP #29GJV0LLU #QR9PRR9 1 29 27 982 970
    #2PVQ8PJ8G #AAA #BBB 1 44 44 1488 1488
    #2PVQ8PQG0 #VGJ0JRJ #CLP9YQJ8 1 9 39 440 1305
    #2PVR8VC8U #29GJV0LLU #CLP9YQJ8 2 28 26 1086 960
    #2PVR8VRUG #QR9PRR9 #BBB 2 24 32 891 1188
    #2PVR8VUQ8 #VGJ0JRJ #PQRPL0R 2 6 43 393 1445
    #2PVR909JP #AAA #88QLGVQJ 2 25 32 964 1269
    #2PVU2GGGU #QR9PRR9 #CLP9YQJ8 3 19 24 898 921
    #2PVU2GQPG #VGJ0JRJ #88QLGVQJ 3 10 43 366 1450
    #2PVU2GRV8 #PQRPL0R #BBB 3 33 25 1200 978
    #2PVU2GUQJ #29GJV0LLU #AAA 3 23 31 934 1205
    #2Y02QQ0L0 #29GJV0LLU #PQRPL0R 4 26 28 917 1144
    #2Y02QQG9J #QR9PRR9 #AAA 4 24 32 952 1112
    #2Y02QQJUP #VGJ0JRJ #BBB 4 13 40 496 1371
    #2Y02QQPPU #88QLGVQJ #CLP9YQJ8 4 39 18 1434 694
    #2Y0PR88CJ #VGJ0JRJ #QR9PRR9 5 12 36 433 1284
    #2Y0PR8LJG #CLP9YQJ8 #BBB 5 22 32 786 1338
    #2Y0PR8P80 #29GJV0LLU #88QLGVQJ 5 33 31 1173 1254
    #2Y0PR8YLP #AAA #PQRPL0R 5 29 25 1084 1084
    #2Y0QVCGUU #QR9PRR9 #PQRPL0R 6 20 31 900 1135
    #2Y0QVCJ98 #AAA #CLP9YQJ8 6 35 16 1363 487
    #2Y0QVCQGG #29GJV0LLU #VGJ0JRJ 6 38 10 1328 420
    #2Y0QVUQVJ #88QLGVQJ #BBB 6 30 31 1091 1115
    #2Y0C88C28 #QR9PRR9 #88QLGVQJ 7 20 38 757 1385
    #2Y0C88RJU #VGJ0JRJ #AAA 7 9 40 517 1404
    #2Y0C88UYL #PQRPL0R #CLP9YQJ8 7 34 18 1300 514
    #2Y0C88VRJ #29GJV0LLU #BBB 7 22 33 793 1160

    I've retagged the 2 involved clans to #AAA and #BBB to protect the guilty. Their results are highlighted in orange and red. There was also a clan tanking with low bases in that group. They, and the scores against them, are highlighted in green - these are all high scores. That's quite legal - a clan is allowed to give up and drop, if they really want.
    Apart from the greens there aren't any scores anywhere near 40 anywhere in the group. And the scores for these 2 clans look to range from about 25-32. Yet they mysteriously got 44* each (in fact tying on %). Suspicious, no?

    Well, apparently, the octopus doesn't agree. Looking at just the 2 clans, and the 30 players who actually attacked in that war, and were therefore clearly involved:
    - both clans were playing in the following season.
    - of the 30 players, 26 of them attacked in the next season. 28 of them attacked in one of the next 2 CWL seasons.
    We can't see player bans, but if a player attacked in CWL they clearly weren't banned. Some players who didn't attack would be people who quit the game, or just skipped a month. Or played in a low league (I only track crystal and up)

    Maybe the octopus was asleep in August? What about July then? There were 5 such wars in July - mostly 45-44 results, where presumably they agreed who would get the win bonus. Lets look at the one of these from Champs 2 (one was in Champs 1 even!), which was the only 45-45 result. Here's the entire league group results again:

    War Tag ClanA ClanB Day StarsA StarsB DestructionA DestructionB
    #2YQUYYC99 #9YCG8J2Y #9Q2P99GQ 1 30 24 1040 927
    #2YQUYYRUV #20LP8JUUG #28R0QVG2 1 26 25 910 873
    #2YQUYYUQQ #QPQYCGJ #PQQ29UP 1 31 27 1134 1137
    #2YQUYYVCC #AAA #BBB 1 45 45 1500 1500
    #2YG0L0GPR #AAA #28R0QVG2 2 31 28 1152 873
    #2YG0L0Q0Y #QPQYCGJ #9Q2P99GQ 2 37 22 1342 825
    #2YG0L0RGV #9YCG8J2Y #BBB 2 28 29 1009 1180
    #2YG0L0YJ2 #20LP8JUUG #PQQ29UP 2 28 20 923 806
    #2YG9PVC82 #9YCG8J2Y #AAA 3 30 30 1070 1094
    #2YG9PVULY #PQQ29UP #9Q2P99GQ 3 35 26 1284 959
    #2YG9PVVJR #BBB #20LP8JUUG 3 32 29 1224 915
    #2YG9Y0GUR #QPQYCGJ #28R0QVG2 3 30 30 1137 871
    #2YGLJY2RQ #BBB #QPQYCGJ 4 30 30 1144 1081
    #2YGLJY8VC #9YCG8J2Y #20LP8JUUG 4 23 28 956 1028
    #2YGLJYPP2 #PQQ29UP #28R0QVG2 4 32 26 1170 893
    #2YGLJYYGY #AAA #9Q2P99GQ 4 38 26 1342 866
    #2YGRCG2UY #9YCG8J2Y #PQQ29UP 5 29 28 1023 1080
    #2YGRCGQ2Q #AAA #QPQYCGJ 5 27 27 1021 989
    #2YGRCGYC9 #BBB #28R0QVG2 5 33 19 1219 591
    #2YGRCQV9C #20LP8JUUG #9Q2P99GQ 5 35 16 1258 605
    #2YGV8Q2GY #BBB #PQQ29UP 6 32 26 1205 1003
    #2YGV8Q8UR #9Q2P99GQ #28R0QVG2 6 16 24 580 795
    #2YGV8QP8V #9YCG8J2Y #QPQYCGJ 6 30 33 1125 1189
    #2YGV8QYQ9 #AAA #20LP8JUUG 6 33 31 1143 1004
    #2YR8PC9YQ #QPQYCGJ #20LP8JUUG 7 32 29 1150 1035
    #2YR8PCL02 #AAA #PQQ29UP 7 30 27 1186 1052
    #2YR8PCPRC #BBB #9Q2P99GQ 7 40 12 1429 512
    #2YR8PCQPY #9YCG8J2Y #28R0QVG2 7 34 2 1208 136

    I've retagged them again (although if anyone is that interested the war was correctly black-flagged as clear cheating in my interim results lists, so this information has been publicly visible for months).

    At the end of this week there were a few high scores as a couple of clans gave up. But apart from that, nothing close to 45. Not from these 2 clans or any other.

    And the next season... both clans were in CWL again, and all 30 of the players involved attacked in August. The pair of clans in Champ 1 league who got 45-44 were playing in August along with 28 of the 30 players. In fact of the 10 clans involved in the 5 near-perfect results, all but 2 were playing next month and 143 of the 150 players were playing in one of the following 2 seasons.

    OK, so going all the way back to February just after this announcement. There were 3 such 44-44 or better wars in Ch3 and 1 in Ch2. This gets a little more interesting:
    - Neither of the Ch2 clans is listed in March. But both played in April
    - Of the 30 players from Ch2, only 8 played in March, but all 30 played in April.
    - Of the 6 Ch3 clans, all played in March.
    - Of the 90 Ch3 players, 82 played in March, amd 84 played in March or April

    So it looks like maybe the octopus got a bit interested in the Ch2 result - as those clans and most players were out. But any penalty only lasted one month, and didn't seem to affect all of the players. It seems to have ignored Ch3 entirely.

    Of the 31 wars with 44-44 or higher, I could only see 2 where it looked like the clans were out next month. In fact there's a couple of clans that appear twice in these wars - one of the Feb clans was involved in another in July (and a July clan did it again in Nov). Only 5 players featured in both of those Feb&July wars - but just looking at the first couple of them, they seem to have been active in every CWL season before and after.

    So a question - has anyone heard of the win-trading ban hitting players or clans? Maybe I've messed up my analysis? Maybe a 45-45 result doesn't count as win-trading?!? But I can't see how it could possibly not be covered by:
    agreeing to “fix” the results through negotiation, removing defenses, Star trading, or any other means other than playing legitimately is viewed as win-trading
    (There are, of course, other sorts of win trading. But double-perfect wars are just an easy one that can be pulled out of the database with a trivial query)
    Various data posts: Who plays what? CWL data 1 (stars), 2 (3* rates), 3 (rosters), 4 (start times and other) Data at end of old Legends And (non-data) how max bases are ordered in war
    --------------------
    Forum cup organiser. The forum cup discord server is: https://discord.gg/KEUxV2e

  2. #2
    Forum Legend Warios's Avatar
    Join Date
    Feb 2014
    Posts
    8,506
    We have run into this situation once, and I did report it. My follow up on the 2 clans the next season showed no change in league gained by nefarious means, and both clans participated in the next league season. Don't think SC puts much effort into leagues below Champ 1, and the clans in the spotlight.
    Level: 256 | TH12 | GG 2B
    Highest Trophy Count:5376 | Donations: 1.26 million | Hero Levels: 170
    Leader of Rogue Elite
    #8200L0UJ
    Retired League Clan and Players recruiting experienced attackers to play in a relaxed and fun environment.
    ​Our Discord link https://discord.gg/pwz4W2h






  3. #3
    Pro Member higgsbosons's Avatar
    Join Date
    Nov 2016
    Location
    under a rock
    Posts
    621
    It's almost as sad as expectable. During the first ingame qualifiers you could watch players being banned from top 10 clans on clashofstats, this procedure took so long that the bans where probably done manually. A banned player can still be found in the member history of the clan, has no profile image anymore and if you try to load the player page you'll get an error.

    Detailed scanning was often performed after the CWL season has been finished. I doubt the clans in champ 1 and below were systematically scanned and checked, probably because it requires detailled manual checks before the ban.

    I haven't heard of a bigger ban wave regarding collusion,
    but i've heard a few times that suspicious wars were reported and no action was taken,
    which is also indicated by your analysis.

  4. #4
    Forum Veteran Thegreatpuma's Avatar
    Join Date
    Mar 2016
    Location
    Druidia
    Posts
    1,736
    Meh. This won't gain much traction here. Why not post this in Darians original post so it gets some views in General?

  5. #5
    Quote Originally Posted by Thegreatpuma View Post
    Meh. This won't gain much traction here. Why not post this in Darians original post so it gets some views in General?
    Probably gain more traction here than to get buried in Darians original post.

  6. #6
    Senior Member
    Join Date
    Jun 2014
    Posts
    362
    I’m not sure if it’s too much of a PITA with your data, but i think an interesting one to test this is whether a clan has win traded 2 (or more) times & then what’s happened to them. may need a wider sample, hence PITA.

    A single event of trading behaviour may be trading or some fluke, but 2 for the same clan may be their threshold for action (subject to the issue not becoming so widespread that tougher action is needed).

    I’m not disputing your conclusions in these cases btw, but the costs (complaints, eyeballing cases), false positives & actual cheaters stopped needs to be balanced. The other interesting stat may be traded wars as a % of total wars in these leagues over the sample period - if low fraction of a %, that would also point towards the issue being ‘within appetite’ unless peeps got greedy & did it multiple times, which then may trigger action.

  7. #7
    Forum Superstar joshsgrandad's Avatar
    Join Date
    Jul 2017
    Location
    NE England, UK.
    Posts
    3,030
    Quote Originally Posted by sparxmarx View Post
    I’m not sure if it’s too much of a PITA with your data, but i think an interesting one to test this is whether a clan has win traded 2 (or more) times & then what’s happened to them. may need a wider sample, hence PITA.

    A single event of trading behaviour may be trading or some fluke, but 2 for the same clan may be their threshold for action (subject to the issue not becoming so widespread that tougher action is needed).

    I’m not disputing your conclusions in these cases btw, but the costs (complaints, eyeballing cases), false positives & actual cheaters stopped needs to be balanced. The other interesting stat may be traded wars as a % of total wars in these leagues over the sample period - if low fraction of a %, that would also point towards the issue being ‘within appetite’ unless peeps got greedy & did it multiple times, which then may trigger action.
    Disagree.. Its cwl, at the high end.. So we are talking 1 million dollar prize money, and ESL, for which CC will probably need to show the integrity of the game is foremost in their minds.. And the way to do this is by taking action, when obvious collusion appears to be present.. They were quick to act in the early stages of cwl, but they appear to have taken their eye off the ball, for whatever reason.

    Cheating is cheating.. Even if its a small percentage of overall matches, for the matches in question, it is 100%

  8. #8
    Senior Member
    Join Date
    Jun 2014
    Posts
    362
    Quote Originally Posted by joshsgrandad View Post
    Disagree.. Its cwl, at the high end.. So we are talking 1 million dollar prize money, and ESL, for which CC will probably need to show the integrity of the game is foremost in their minds.. And the way to do this is by taking action, when obvious collusion appears to be present.. They were quick to act in the early stages of cwl, but they appear to have taken their eye off the ball, for whatever reason.

    Cheating is cheating.. Even if its a small percentage of overall matches, for the matches in question, it is 100%
    At the very top end, I’d totally agree with you (& did see one clan mentioned was Champ 1). Beyond that... you say it yourself, obvious collusion APPEARS to be present. It can feel tough making a call on a single data point if there’s folk saying ‘are you sure? is it just a fluke?’ - far easier where you’ve not just got smoke, but can see some flames too.

    It’s probability vs appetite for the number of innocent folk you’re willing to punish to catch the guilty ultimately. I may be jaded, as i work in fraud & regardless of the fact that it’s both illegal and has societal cost, it’s still probability when evidence is circumstantial (i.e. score only) & simply a cost of doing business - there are cut offs below which it gets through.

    I totally respect your position, just trying to understand the data through my jaded lenses!

  9. #9
    Forum Superstar joshsgrandad's Avatar
    Join Date
    Jul 2017
    Location
    NE England, UK.
    Posts
    3,030
    Quote Originally Posted by sparxmarx View Post
    At the very top end, I’d totally agree with you (& did see one clan mentioned was Champ 1). Beyond that... you say it yourself, obvious collusion APPEARS to be present. It can feel tough making a call on a single data point if there’s folk saying ‘are you sure? is it just a fluke?’ - far easier where you’ve not just got smoke, but can see some flames too.

    It’s probability vs appetite for the number of innocent folk you’re willing to punish to catch the guilty ultimately. I may be jaded, as i work in fraud & regardless of the fact that it’s both illegal and has societal cost, it’s still probability when evidence is circumstantial (i.e. score only) & simply a cost of doing business - there are cut offs below which it gets through.

    I totally respect your position, just trying to understand the data through my jaded lenses!
    i entirely see your POV.. Is just a pity they started big.. then nothing.. Of course, its possible they took a look then decided no evidence/or innocent, since they dont release any info..

    Doesnt affect me, same as much of the grievance.. simply like to see those it does affect have a fair shot.. As with all grievances posted on forum.

  10. #10

    Join Date
    Sep 2014
    Location
    Digging another tunnel in the data mine
    Posts
    6,660
    Quote Originally Posted by Thegreatpuma View Post
    Meh. This won't gain much traction here. Why not post this in Darians original post so it gets some views in General?
    I got told off because I'd put it in general. I don't suppose necroing the old thread would have gone down much better.

    Quote Originally Posted by sparxmarx View Post
    I’m not sure if it’s too much of a PITA with your data, but i think an interesting one to test this is whether a clan has win traded 2 (or more) times & then what’s happened to them. may need a wider sample, hence PITA.
    I did mention there's a couple of clans that appeared twice. I think I only checked the first (Feb+July) and nothing had happened to them. The other was recent (I said Nov, but must have meant Oct) and wouldn't have any data on what happened after the second of those yet.

    I’m not disputing your conclusions in these cases btw, but the costs (complaints, eyeballing cases), false positives & actual cheaters stopped needs to be balanced. The other interesting stat may be traded wars as a % of total wars in these leagues over the sample period
    It's low - 31 out of the entire of Champ/Master leagues over some months. However, it's something that can be found instantly - it only took me a minute to pull off the list of 31. And it would give an easy way for SC to be seen to be doing something.

    Quote Originally Posted by sparxmarx View Post
    At the very top end, I’d totally agree with you (& did see one clan mentioned was Champ 1). Beyond that... you say it yourself, obvious collusion APPEARS to be present. It can feel tough making a call on a single data point if there’s folk saying ‘are you sure? is it just a fluke?’ - far easier where you’ve not just got smoke, but can see some flames too.
    I don't buy that there's any chance these wars are flukes. If you look at the league dumps I put in my post you can see the typical data. 44s and 45s are WAY out of range. The overall 3* rate is around 10% in most of these leagues. For 2 clans to do 30 consecutive triples against each other is... unlikely.

    Here's the breakdown, in Champs leagues, of what a clan scores if they concede 44 or 45:
    Stars Freq Freq %
    0 577 21.53%
    1 276 10.30%
    2 239 8.92%
    3 206 7.69%
    4 159 5.93%
    5 156 5.82%
    6 137 5.11%
    7 101 3.77%
    8 101 3.77%
    9 90 3.36%
    10 71 2.65%
    11 71 2.65%
    12 59 2.20%
    13 67 2.50%
    14 45 1.68%
    15 43 1.60%
    16 37 1.38%
    17 28 1.04%
    18 35 1.31%
    19 33 1.23%
    20 28 1.04%
    21 13 0.49%
    22 19 0.71%
    23 12 0.45%
    24 8 0.30%
    25 9 0.34%
    26 5 0.19%
    27 5 0.19%
    28 3 0.11%
    29 2 0.07%
    30 1 0.04%
    31 0 0.00%
    32 1 0.04%
    33 0 0.00%
    34 0 0.00%
    35 0 0.00%
    36 0 0.00%
    37 0 0.00%
    38 0 0.00%
    39 1 0.04%
    40 2 0.07%
    41 3 0.11%
    42 3 0.11%
    43 4 0.15%
    44 11 0.41%
    45 19 0.71%

    Half of the time they score 0-4. Basically all of these are tanking clans. Some tanking clans do attack for stars. Outside these wars, good clans are scoring 30-35 typically. The top clans maybe 35-39. But there are almost no 45-X results in this range - most cells are literally zero. If a clan concedes 45 they just don't get a proper score themselves. There is quite a spike at the very top, though, where they get a more or less perfect score in return.

    And just looking at some of the very few results with sort of "normal" scores in response to a 44/45. The 2 29s are:
    Group IC in Ch2 June. They clearly got given that - the opponent suddenly entered non-max bases. In fact there appear to be various shenanigans going on there as some of the other results are flagged in my results list.
    Group F in Ch3 June, day 4. One clan obviously put out FWA bases every round, but attacked (which is unusual), getting typical scores. So the 45 wasn't a genuine score.

    The single 32 in the table above was a 32-44 from Group Q in Champ 1 in August. This is interesting as it's probably a genuine result; ie the 44 was earned. That clan however was the #1 clan in the world that month - and by some margin.

    So... I can only see one 44 or 45 result that looks genuine in the entire of champs leagues. And that was the outright CWL winner that month. So the odds of 2 regular clans suddenly pulling a 45-45 against each other is "whelk in a supernova" odds.
    Various data posts: Who plays what? CWL data 1 (stars), 2 (3* rates), 3 (rosters), 4 (start times and other) Data at end of old Legends And (non-data) how max bases are ordered in war
    --------------------
    Forum cup organiser. The forum cup discord server is: https://discord.gg/KEUxV2e

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •