James Rundle: A Different View of Data

The US Commodity Future Trading Commission’s (CFTC’s) inability to even perform basic analysis on the data from transaction reports, as admitted during Commissioner Scott O’Malia’s keynote address at the Swift Standards Forum in London last month, is alarming. It’s not good news for the market as a whole, which needs robust oversight to engender confidence, and it’s not good for individual market participants either, who struggled to put systems in place for reporting, only to find that the information they submit is essentially useless.

Elements of the CFTC clearly believe that they can use technology to overcome this difficulty, fix the glaring blind spot in their wing mirror, and go about actually taking control of market activity. And they can. The technology definitely exists to link futures and swaps markets, crunch the data, throw out analytics, and enable regulators to ask all kinds of weird and wonderful questions about where risk is. But while the technology exists, it’s doubtful whether the CFTC will be able to develop it without a significant increase in resources.

Indeed, some even question whether the CFTC necessarily needs to be able to install a Skynet-esque big data engine that will alert them, in real time, to all possible instances of market manipulation and risk vectors. Some say that a far simpler approach is manageable, and indeed, preferable.

Let’s break reporting down into its primary goals, which are to allow regulators to gain a view of market activity in order to assess weak points, specifically on an institutional level, providing them with a means to measure and manage systemic risk on a wider scale. While I was working on a feature regarding the problems that have occurred in both Europe and the US in terms of establishing reporting regimes with respect to derivatives transactions this month, most practitioners I spoke to pointed out that the CFTC could accomplish this by asking certain questions of swap data repositories (SDRs) alone.

The simplest question would be asking the SDRs to provide the CFTC with a ranked list of which institutions have the highest level of failures per day. Those in the top five consistently are then selected for a site visit and a deep dive into their holdings and books, because their house is clearly not in order. Likewise, as Sapient Global Markets' Cian Ó Braonáin argued, for concentration risk measurement, the most useful tool is clearly identifying products and the notional value on them, which should give regulators an indication of who is trading what and where. Once significant positions are identified, that too can lead to a visit to establish areas such as being over-exposed to a certain product or entity, or over-leveraged across markets.

What it comes down to is an argument over the very essence of surveillance on a regulatory scale.

Surveillance
What it comes down to is an argument over the very essence of surveillance on a regulatory scale. Is it better to be proactive, in which case, technology is the only way to institute real-time alerting, ad-hoc trend analysis, and all of the other areas that the CFTC is looking at? Or is it better to be reactive, where analysis of data is shared among various regulatory agencies, historical information is widely used, and people take priority over machine-driven forensics?

Both approaches have their benefits and drawbacks. The real-time surveillance aspect only works once the rules are properly configured, and it’s nearly impossible to account for every possible permutation of market abuse, systemic fraud, or for any other kind of risk for that matter.

Artificial intelligence would likely be a prerequisite, although on the post-trade, forensic side, it limits the ability of regulators to react quickly to events, and indeed, that loss of speed when the markets themselves move so fast is perhaps too wide to be realistically considered. But something should be done, because the system in its current form simply isn’t working.