Getty Images Back in the old days, an ethically challenged trading desk might find it profitable to step in just prior to large client orders, coordinate a plan of attack over phone, and then swarm in to bet in the opposite direction of the big trade. It’s called front-running and it is both bad and illegal. But in the 21st century, that kind of grubby, low-tech grift is no more. Instead, as New York’s Department of Financial Services alleges in a $135 million fine against Credit Suisse, it’s now possible to fully automate the whole process with a sleek and sexy algorithm: From at least April 2010 to June 2013, Credit Suisse employed an algorithm designed to trade ahead of clients’ limit and stop-loss orders. Credit Suisse programmers designed the algorithm to predict the probability that a client’s limit or stop-loss order would be triggered. Credit Suisse programmers and traders had ongoing and significant discussions about how the algorithm could be used to “front run” customer orders. Credit Suisse predicted that this particular algorithm would generate approximately $2 million in profits for the calendar year 2013. Now if you read that section of the consent order carefully, you might come away with the impression that Credit Suisse’s foreign exchange desk wasn’t trying to front-run their customers’ per se, but merely implement an algo that made front-running possible, depending on the accuracy of its predictions about client orders that hadn’t actually been triggered yet. One could imagine this kind of functionality being deployed innocently and in the best interests of the bank’s customers. To prove the opposite, a prosecutor would need some pretty blatant documentary evidence displaying conscious intent to do crime. Thankfully, Credit Suisse was all too willing to oblige: For example, in May 2012, an eFX trader unabashedly bragged to the Head of eFX: “it is the name of the game if we front-run orders. Sometimes you win sometimes you lose. It’s up to us to define the front run rules.” In July 2012, this eFX trader wrote to another eFX colleague, “ think with a smart frontrun we can make money. We will lose most of the LO [limit order] clients if we start executing on bucket rates. The algo should be smart enough to optimize this.” The lack of subtlety here carried over into bland and seemingly quotidian business discussions. These are kinds of conversations that are usually carried out on big whiteboards, not the message apps whose shady transcripts we’re so used to reading in cases of forex misdeeds. Improving the effectiveness of this algorithm was an important business objective of eFX senior management in 2012. In one e-mail from February 2012, for example, the Head of eFX identified as a priority: “Improve order front running (e.g., limit front running in % of order amount).” Apparently to encourage traders to employ the front running strategy, the Head of eFX wrote, in a January 2012 e-mail, that Credit Suisse should improve its “PnL tracking on the front running order.” The lead algorithmic researcher in eFX management echoed that imperative, noting that the Quantitative group sought to “Attribute P&L to front running of limit orders.” Reading this, it almost seems like the managers involved had no idea that front-running was a bit of a no-no. It’s definitely not the sort of thing you want to explicitly call out in department-wide emails. Elsewhere we see bankers not at Credit Suisse acting with at least a modicum of self-awareness. A 2012 chat between a Credit Suisse trader and her counterparts at at least one other global bank shows the group conspiring to drive down the price of the Euro/Yen trade. Once the price approached the limit, the Credit Suisse trader typed, “ok pulling my bids,” confirming that their only purpose was price manipulation. This behavior manages to shock even the benumbed souls of the forex traders in her chat: [Non-Credit Suisse trader] quickly pleaded with her not to withdraw them, concerned that it “smlls unethical haha” and “if custy ever found out we both dead.” Another chat room participant, noting the explicit nature of the discussion, stated (with evident sarcasm): “what the hell are u two planning . . . this chat is a compliance dream.”
If that was an individual doing such with an algorithm instead of a large corporation doing such with algorithms...he/she would have been arrested and then sentenced to to 2 - 5 years in prison along with being fined. Also, all of his/her assets would have been frozen.