Proving digital intent


Connecting the digital dots has never been easier, according to Fonetic CEO Juan Manuel Soto
Whether or not a couple of traders actually pull the trigger on some sort of mischief, Soto’s communications surveillance platform would have already captured and identified many aspects of the possibly illicit interaction, for what it could have been. “Those two traders don’t need to actually execute the transaction. We would be able to capture their conversation, for example, two guys discussing rate rigging. We would have behavioral indicators as well, out behavioral algorithms bring it all together,“ he says. “So normally, maybe these two traders don´t usually talk about certain things or not on a particular day. This may cause an alert, as we would capture this. Next, we might have one of the traders sitting at different station or logging into computers or other technology that’s not his. We would have that, too. All of these things are outside regular behaviors. Regular patterns. Together, all of these things
can be combined to help detect irregular behavior, and raise alerts.“ And, once alerts are raised, it’s SOP to include all the trader’s emails and texts, lMs, et cetera, plus a human to analyze the lot of it, to further investigate the trader’s behavior to see what’s really up.
“All of these things can be seen even if the traders didn’t execute. What you do have, are records, facts, of a trader’s intent to do it… You can detect two people, for example, trying to do something bad. The illegal act is already there,“ Soto says, “trade or no trade.“
We’ve been impressed with the Fonetic surveillance offering for this very reason: the precision of its behavioral and pattern recognition capabilities across different digital channels, from voice to IM’s. Soto noted that the company does quite a bit of one-off forensic work, deep trade reconstruction analysis and so on to uncover trader mischief. “We deal with facts, facts in communications. Facts to prove the intent of a crime, planned but possibly not even committed. Our system can raise alerts when a trader doesn’t do something that he might normally do, or if something is missing, or something is not in place.” He says they have done forensic investigations or reconstruction work for European regulators, including pattern searches across wide swaths of markets, searching vast call files or data files for conversations and interactions between certain people. We can almost picture Matthew Hunter filling out an agency budget request form.
Soto says business is pretty good in the European market right now as all the various MiFID II requirements kick in around mid-January. He expects more action in the US market to follow shortly thereafter. See
On this side of the pond, the debate is less about biometric pattern recognition for traders and more about whether or not regulators should have easy access to an AT firm’s algorithmic source code. Many now believe it to be the ultimate smoking gun report (the code that is) for any investigation into possible AT wrongdoing. Listening to Soto, we wonder if the algo source codes are really that necessary to nail high-tech perps nowadays. Maybe it’s time to simply refocus surveillance resources on quants and developers and perhaps even further up the decision tree, instead of simply focusing on the lowly trade desk. -the editor


The Risk Desk, Sept 28 2016, Vol15 Issue 18