one million words's picture
By: one million words, Pete Jahn
Mar 01 2008 1:04am
4
Login to post comments
3523 views


League Play and League Fraud
 
Recently, I noticed a post on the MTGO discussion boards about possible fraud in a Coldsnap league. The League had 24 players with 5-0 records, while having only 70 players. The poster called shenanigans, and a few days later, 10 players were DQed. So what happened? If you play in a league, are you at risk.
 
First, let’s talk about leagues, and how they work at present.
 
Leagues are sealed deck tournaments. You get the equivalent of five booster packs: either five boosters packs (Tenth Edition, MED and Coldsnap) or a tournament pack and two boosters. You build your deck from that pool of cards. Then you play matches against any of the other players in the league. Matches can take place any time during the week the league runs, and you will be paired against anyone requesting a game (provided you haven’t already played that person.)
 
In the past, leagues used to run for a month, with players adding an additional booster every month. That is no longer true: to make sure that the leagues will be over when MTGO v2.0 shuts down for the transition to v3.0, leagues are just one week long.
 
In league play, your first five matches count for league points. Any additional matches count as tiebreakers. This means that the first five matches you play are critical. The ones after that are gravy.
 
Leagues are capped at 256 players, but only Lorwyn leagues typically get that many players. Most of the old set leagues get far fewer players. In recent weeks, Lorwyn leagues have been consistently hitting 256 players, while MED leagues have hovered around 100, Mirage and Future Sight leagues have been in the 75 player range, and Coldsnap leagues have had between 80 and 120 players.
 
Leagues stretch the prize payouts far down the line. Here’s the current league payout schedule:
 
First Place: 27 packs
Second Place: 21 packs
Third & Fourth place: 15 packs
Fifth through Eighth place: 8 packs
Ninth through Sixteenth: 6 packs
Seventeenth through Thirty-second: 3 packs
Thirty-third through Sixty-fourth: 2 packs
Sixty-fifth through 128th: 1 pack
Dream Cache
 
Right now, league play is a bargain, provided you need the commons and uncommons. Other than Lorwyn, the leagues are going to pay everyone, even if they play no matches at all, at least a pack. In a recent Mirage league, everyone who played at least one match got two packs.
 
In leagues with three sets out, the payout used to be last set released, then middle, then first – for example, a Mirage league will pay one Weatherlight booster to 65-128, Visions plus Weatherlight to 33-64, and a Mirage draft set to anyone with a record that puts them in 17-32.   MED leagues pay entirely in MED packs, while Coldsnap and Tenth leagues pay in those packs, of course.  
 
The Pros and Cons of League Play
 
Leagues – especially one week leagues – are a lot like more flexible sealed PEs. The leagues offer more flexibility, however. If you do not have the time for 5-6 rounds, plus an immediate top eight draft, leagues can let you spread the matches over an entire week. Secondly, leagues let you rebuild your deck anytime, and play a revised build in game one of later matches. That lets you do some experimenting – although that experimenting can cost you some points. 
 
You can also play any number of league games – you can basically keep playing as long as you can find an opponent.
 
On the down side, leagues do not include free drafts for the Top eight players. That means fewer cards – and a free Top eight draft is a free draft. More importantly, in the non-Lorwyn leagues, actually finding an opponent can be tough. I have often had a match request up for over an hour without getting a hit. It’s like a bad day fishing, but with the disadvantage that you are not outside somewhere, and you can’t fall out of the boat.
 
The other disadvantage is that you might lose out to some people rigging the leagues. Maybe. Let’s look at that for a bit.
 
What Supposedly Happened
 
The best theory I heard was that a group of foreigners in a sweatshop were playing multiple accounts, and seeking to corrupt the economy of MTGO. I don’t buy that one.
 
On the other hand, it does look like a group of players got together and created a number of accounts, and entered them all in Coldsnap league #1137265. These players chose some accounts to be the losers, and had each other account pair matches against these accounts, and win. That quickly got a number of their accounts to the magical 5-0 level that put them into contention for first, or at least T8. The players then played the winning accounts continuously to rack up tiebreakers, and ensure that they won lots of packs. 
 
Could this work? Well, not in Lorwyn, but it probably could in some of the less played leagues. Players would have to get on at the less played times – like 6am eastern time. The players would have the loser account request a match. The loser account would then wait a bit, to make sure that no other players had match requests up. If it was not paired in a match in three minutes or so, then the winner account would also request a match. At that point, loser and winner should be paired, and the loser account could throw the match to the winner. Then the player running the loser account could log into another account and do it again.
 
The next question is whether that happened. Let’s look at some data. 
 
Lorwyn leagues: 
 # 1141118: 256 players, 14 at 5-0 - that’s 5.5%
 # 1140107: 256 players, 15 at 5-0 - that’s 5.8%
 # 1139036: 256 players, 15 at 5-0      (ditto)
 # 1137890: 256 players, 15 at 5-0      (ditto)
 # 1137262: 256 players, 18 at 5-0 - that’s 7.0%
                      Average: 6.0%
 
For Lorwyn 1 Week Leagues, that seems pretty consistent. However, the larger number of players & decks means that powerful decks will see equally powerful competition more often. Let’s look at some other leagues as well.
 
Time Spiral block leagues:
 # 1141097: 142 players, 8 at 5-0 - that’s 5.6%
 # 1137264: 116 players, 6 at 5-0 - that’s 5.2%
 # 1130289: 155 players, 9 at 5-0 - that’s 5.8%
 # 1126155: 124 players, 10 at 5-0  - that’s 8.1% (an outlier?)
 # 1121800: 115 players, 6 at 5-0 - that’s 5.2%
                      Average: 6.0%
 
MED leagues:
 # 1141093: 99 players, 4 at 5-0 - that’s 4.0%
 # 1137267: 87 players, 5 at 5-0 - that’s 5.7%
 # 1130279: 106 players, 8 at 5-0 - that’s 7.5%
 # 1126141: 107 players, 8 at 5-0 - that’s 7.5%
 # 1121801: 104 players, 9 at 5-0 - that’s 8.6%
                      Average: 6.7%
 
Tenth Edition leagues
 # 1141092: 213 players, 7 at 5-0 - that’s 3.2%
 # 1137263: 202 players, 7 at 5-0 - that’s 3.5%
 # 1129970: 223 players, 12 at 5-0 - that’s 5.4%
 # 1125807: 199 players, 15 at 5-0 - that’s 7.5%
 # 1121566: 225 players, 15 at 5-0 - that’s 6.6%
                      Average: 5.2%
 
Mirage leagues
 # 1141091:   66 players, 5 at 5-0 - that’s 7.6%
 # 1126901:   71 players, 5 at 5-0 - that’s 7.0%
 # 1122078:   71 players, 4 at 5-0 - that’s 5.6%
                      Average: 6.7%
 
 
The numbers vary a bit, as people get lucky or unlucky, but the results seem to be pretty consistent. In any one week league, approximately six percent of the players will go 5-0 in their for-points games. Some leagues are swingier than others, but the overall numbers are pretty well grouped.
 
Let’s look at Coldsnap.
 
Coldsnap leagues:
 # 1141096: 88 players, 7 at 5-0 - that’s 8.0%
 # 1137265: 82 players, 13 at 5-0 - that’s 15.8%
 # 1130308: 118 players, 11 at 5-0 - that’s 9.3%
 # 1126173: 98 players, 23 at 5-0 - that’s 23.5%
 # 1121802: 117 players, 10 at 5-0 - that’s 8.5%
 # 1116399:   79 players, 4 at 5-0 - that’s 5.1%
                      Average: 11.7%
 
Houston, we have a problem. 
 
The first Coldsnap one-week league showed typical results – 5.1% of the players went 5-0. The second one-week league bumped up a bit, but it was not outside all bounds of reason. The third league was ridiculous. Almost one in four players went undefeated? No way.
 
Shenanigans.
 
It is possible that Coldsnap has more broken decks in the format than exist in other formats – but that still does not explain the high percentage of 5-0s. They still have to play someone – you cannot just assume that all their opponents are morons. 
 
Let’s assume that Coldsnap has one-third broken decks, and two thirds trash decks. The broken decks always beat the cream puffs. In that case, one third of the players should be playing broken decks – and the other two thirds should be dropping before playing a match, since they will know they cannot win. If you assume that every cream puff deck plays two matches, then drops, in a 100 player league 165 for points matches will be played by broken decks, and 132 matches played by cream puffs. The 23 undefeated decks need to play 115 of those matches – without hitting another broken deck. Remember – we postulated 33 broken decks in the league, and to get these results 23 of them have to avoid ever playing against another broken deck.
 
Even assuming that the cream puffs play five matches, the other decks will be playing tiebreakers. Many of the undefeated decks played dozens of matches. The odds of all those great decks missing each other in the for-points rounds is minuscule. 
 
Somebody cheated. They arranged matches.
 
Let’s break down the matches played a bit further. Let’s look at how many matches each group of players played.   Record first, then total matches. I am using points. In League play, a win counts two points, a loss one. Thus someone with four points may have played and lost four matches, or played and won just two.
 
10 points: 435 matches
9 points: 75 matches
8 points: 169 matches
7 points: 170 matches
6 points:   96 matches
5 points: 36 matches
4 points: 11 matches
3 points: 8 matches
2 points: 3 matches
1 point: 3 matches
0 points: 3 matches
 
Total: about 900 matches – with just under half of all those matches played by people in the Top eightg
 
I see a few other anomalies. One play has a 7,4 record (seven for-points matches, plus 4 tiebreaker points), but played 36 matches. The player just below him is also at 7,4, but that player played just seven matches. Other players also show a ton of games played, but almost no tiebreaker points. This does not necessarily mean that they were throwing games, but it looks suspicious. I have seen a few other players with similar records – players who just want to keep playing their decks even though they have bad pools, but it is rare. 
 
The Economics
 
So, does it make financial sense for a group of people to rig a league in this way. Let’s look at Coldsnap League #1137264. Here are the final standings, if the players in italic had not been DQed. I also don’t know when in the week those players were DQed – I think it was before the league ended. Therefore, they could have increased their tiebreakers and climbed in the standings, but I’ll skip that speculation and use the final numbers. 
 
First: 10, 46 - 27 packs
Second: 10, 36  -  21 packs
Third: 10, 32 - 15 packs
Fourth: 10, 20 – 15 packs
Fifth: 10, 13 - 8 packs
Sixth: 10, 8 - 8 packs
Seventh: 10, 4 -  8 packs
Eighth: 10, 2 -  8 Packs
Ninth: 10, 2  - 6 packs
Tenth: 10,1 - 6 packs
Eleventh: 10* - 6 packs
Twelfth: 10 – 6 packs
Thirteenth: 10 – 6 packs
...
about Thirty-fourth: 9 - 2 packs
...
about  Fiftieth: 7 - 2 packs
...
about Sixty-fifth: 6 – 1 pack
 
Assuming the ten folks that were DQed were the entirety of the scam, even without playing more tiebreaker matches to bump up prizes, the team would have won 60 packs. Entry into the leagues costs 5 packs, plus 2 TIX, per player. The total cost for all ten players to join the league would be 50 Coldsnap packs plus 20 TIX. 
 
At first glance, that looks like a lot of work to buy 10 Coldsnap packs for 20 TIX. However, that does not include the value of the cards they received in their pools. 
 
I did a bit of number crunching on values of Coldsnap cards. I extracted the current (as of February 27, 2008) sale price of all Coldsnap rares and uncommons from the MTGOTraders,com website. The average sale price of a Coldsnap rare is $1.55. The average sale price of an uncommon is $0.41. Obviously, that average is bumped up by a few costly cards (like Ohran Viper and Scrying Sheets), but even if you exclude the high priced cards, the average values are still $1.20 and $0.30.  Moreover, since the scammers are opening 50 packs, and Coldsnap has just 50 rares, odds are that they will open one Viper and one Sheets.
 
In short, the retail value of the cards they opened is 50 times $1.55 plus 150 times $0.41, for a total of about 150 TIX, by the time you throw in commons and foils. Selling those cards in the marketplace and to bots might generate 75-100 TIX – call it a conservation 90.
 
Coldsnap boosters run about 4.5 TIX each. Using 4.5 TIX per booster as a buy price and 4.0 TIX as a sell, here’s the breakdown:
 
Cost to enter the league: 245 TIX
Gross profit: 330 TIX
Net Profit: 85 TIX   Since TIX sell for about $0.90 or less, say $75 total profit.
 
The profits could have been a higher if the banned folks had played more league matches, but that may or may not have been possible.
 
Although ten accounts were involved, I doubt that ten different people were involved. More likely, it was a small group, say 2-4, each with multiple accounts. The banned accounts played a total of 122 league games. Seven of the banned accounts had 5-0 records, and one more had a 4-1 record, meaning that those accounts probably played at least 35 games against the loser accounts – meaning that 70 of the 122 matches probably took very little time. The remaining 50 matches probably took, on average 30 minutes each. That means they played for 25 hours – or for an hourly wage of about $3.00 an hour. 
 
Or about half of minimum wage. 
 
That does not seem worth risking your account over, but some people are really stupid.
 
Which Leagues are Likely Targets?
 
Coldsnap is probably the only league in which someone could do this sort of thing. The payout is all in boosters that could be used to enter the next league, and a number of rares, uncommons and even commons are worth something. The leagues are also small, so the crooks are likely to find dead times to engineer pairings between their accounts. 
 
Tenth Edition is the next option. This league also pays out in boosters that can be used to enter the next league, and keep the scam going. However, Tenth Edition leagues are more likely to have players looking for games at all times, which means that the players are more likely to randomly end up in a match against a real opponent, and run the risk of losing a match.
 
MED leagues are slightly safer, because you need three MED boosters and 2 Tenth Edition boosters to enter the league, but the league pays out only in MED boosters. This makes chaining tournaments harder – at least in terms of the effort required. At least, I hope that is enough to keep players away – I am still playing in MED leagues, since playing in a league is better than just busting my last MED packs.
 
Other sets, like Time Spiral, Mirage and Lorwyn, are less likely to see this problem. All require tournament packs and other product that cannot simply be won one week and plowed back into the leagues the next. Lorwyn, especially, is busy enough that scammers cannot expect to arrange matches. Real players are always waiting for matches in those leagues.
 
Will It Happen Again?
 
I doubt it. Right now, Wizards is aware of the problem, and I expect that they will keep a close eye on the smaller leagues. Wizards will be especially interested in extremely short games – since that is how the scheme was exposed initially (a league player liked to watch games, and kept seeing 2-0 games complete in seconds.) If players “play” 2 minute games, Wizards will spot that.   If they try to play the games out, but throw them, they will end up spending a ton more time in set-up. At that point, the scheme will be unprofitable – even for people with a lot of time on their hands.
 
Which does not mean that you should not be paying attention.
 
PRJ
 
“one million words” on MTGO
 
 

8 Comments

by Anonymous (Unregistered) 71.169.183.107 (not verified) at Sun, 03/02/2008 - 20:10
Anonymous (Unregistered) 71.169.183.107's picture

Can someone write an article on non-cheating ways to win stuff. Like, give hints at how us bums can expand our collection?

Is it unethical to try to get all my friends to join a league with me, so we have more chances to play games?

by MirrorMage at Sat, 03/01/2008 - 21:18
MirrorMage's picture

Ack! (Sorry for the blank post)

Nice article Pete. I think you're right on the money on this one. When you break things down, it's just not worth it to cheat. Even if you created some elaborate untraceable strategy, you would still need a variety of things to happen (No one waiting for a match, etc.) in order to earn some marginal profit. It was worth it because it was spread over a few individual. 

It was a small group of 2-4 people. 10 people were DQed from the league, but in reality I suspect 13-14 should have been DQed. I suspect the reason they were not was because CS was able to connect the 10 accounts to each other, while the other three accounts were a little more difficult.

I doubt league collusion will happen anytime in the near future. In the last MED league I was in, 2 people were DQed. I suspect it was on account that they were two accounts registered to the same individual- so they are taking a much tougher stance on the issue. 

by LulThyme (Unregistered) 194.249.98.61 (not verified) at Sat, 03/01/2008 - 09:35
LulThyme (Unregistered) 194.249.98.61's picture

"That means they played for 25 hours – or for an hourly wage of about $3.00 an hour. 

Or about half of minimum wage. "
 
I find this quote is also missing a point.
Something like 99% of players on MTGO are losing money to be able to play.

Why?
Because it's a game, and most of its players find it fun.
 
Why assume that things are different for cheaters?
 
By your logic above, most players on MTGO are actually paying to work...
 
 

by HydraLord at Sat, 03/01/2008 - 11:07
HydraLord's picture

Agreed. We don't know how many previous "smart" cheaters there may have been, but this is a clear example of dumb ones.

I think the method in the article is unlikely to give false positives. You can get a confidence interval for the rate of going 5-0 given random pairings. 10% might be a strong outlier, but still believable. I doubt that 18 or 23% would be. But yeah...all would-be cheaters will have to be smarter now. Shouldn't people clever/dedicated enough to run this kind of shenanigans being playing, I don't know, blackjack?

Thanks for the precise analysis. by insomniax at Sat, 03/01/2008 - 05:14
insomniax's picture

As a Math Grad (2002) i can say that this article was so fun to read. Thanks for the  precise analysis on the topic :)

by theauthenticsimpsonian (Unregistered) 24.2.188.53 (not verified) at Sat, 03/01/2008 - 06:39
theauthenticsimpsonian (Unregistered) 24.2.188.53's picture

I remember reading about something like this happening a long time ago in 7th edition leagues in the WotC Boards archive. They did something similar to what the people in the Coldsnap leagues did, but eventually they got caught. So something like this probably will happen again, but not for a while.

by LulThyme (Unregistered) 194.249.98.61 (not verified) at Sat, 03/01/2008 - 09:30
LulThyme (Unregistered) 194.249.98.61's picture

I think this article slightly missed the point.

The main problem with this whole approach is that it is completly useless in any specific case.

Take, for example, the leagues you bolded and called shenanigans on. From a statistical point of view, there seems to be a problem. What if you look in the league and all the counting matches of 5-0s were played against a wide variety of opponents?  Wizards can't ban accounts here. In fact, this situation is way less suspicious than a league with an average number of 5-0s but where all these played their counting matches against the same 5 players.

Smart cheaters could wait near the end of leagues with a low number of 5-0s for example, and cheat in those and avoid being detected by your methods. I cant think of lots other workaround.

 

In conclusion: I think counting 5-0s in a league can too easily have too many false positive and false negatives to be a useful method.

 

Looking for pattern in opponents in counting matches of 5-0s is much stronger.

 

One has to know the point. by wannareset1990 at Tue, 07/11/2017 - 09:38
wannareset1990's picture
5

One has to know the point. That would be hard then. - Phillip Elden