How to Catch a Proprietary Trader

How can regulators successfully distinguish proprietary trading from market making in a straightforward, easily understood way? Our answer would be to apply a general framework to evaluate trading using metrics that would indicate the possibility of proprietary trading.

The new era of financial reform has promised increased transparency and a reduction in systemic risk. With the Securities and Exchange Commission and Commodities Futures Trading Commission moving through the research and rule-making phases of Dodd-Frank, there has been substantial talk of cost, efficiency, and practicality of the process. We are all for clarity and consistency - where it makes sense. However, there are instances where establishing an intractable boundary is impractical. In our opinion, such is the case with identifying proprietary trading.

Regulators and practitioners alike think it would be easy to differentiate market making from proprietary trading by examining individual trades. But we don't think it will be so simple. Both the trading community and regulators are trying to establish criteria that can identify trades as either proprietary or customer trades. Such clarification would arm regulators with what they need to penalize offenders of the Volker Rule. But we think a precise set of criteria to determine one type of trading from another is not entirely possible: one person's proprietary trade would be another person's "pre-hedge" in a client book of business.

So how can regulators successfully distinguish proprietary trading from market making in a straightforward, easily understood way? Our answer would be to apply a general framework to evaluate trading using metrics that would indicate the possibility of proprietary trading. In-depth follow-up using other measures and unbiased, expert evaluation of trading patterns could then be undertaken where necessary to make a determination in areas where there is suspicious trading that may or may not be proprietary. We think this is the preferred method since each instance would receive the individualized focus warranted to either confirm or deny proprietary trading.

The number of instances may not be substantial in the long run, as any false positive can be used to refine the flagging of suspicious trading. Is this appropriate? We believe it is because the proprietary intent behind an individual trade cannot be determined on a trade-by-trade basis. Proprietary intent should be determined by the manner in which a book is handled over an extended period and how a particular book compares with the industry norm, in terms of risk and return.

So how should regulators go about distinguishing proprietary trading from market making? One would expect a book of agency business to be hedged in such a way as to minimize market and basis risk. Return on the book is generated largely by volume of trading at the bid/ask spread and changes in inventory value. One would expect a book with embedded proprietary business to make more pronounced market risk plays to amplify returns.

We would expect a book managed in a proprietary manner to display greater market risk, on an adjusted basis, than a pure agency book. We think normalized comparisons can be made across firms of various sizes by looking at the ratio of the total market risk relative to the amount of customer-generated market risk incurred over a specified period.

To better explain this approach, if a particular book of business at one market-making firm was shown to have decisively greater market risk on an ongoing basis than comparable books across the industry, this may be an indication that the additional risk is proprietary. Such an indicator should serve as a flag for regulators to examine the book, trades, trading patterns, and returns of the firm to determine if a case can be made that the book is being managed in a proprietary manner and in violation of the Volker Rule.

How would risk be viewed on a normalized basis across firms? We think this should be done with existing metrics in the market place and some additional reporting mandates. We would begin by taking a basic value-at-risk (VaR) measurement across comparable business units at similar financial institutions. We would then evaluate the customer-driven business and risk traded in each book on a product-by-product basis.

This could be measured in a variety of ways, with the simplest example being the sum of the absolute values of the risk associated with the individual trades in each book. This would provide a basis to normalize differences in the size of firms' books and allow for an appropriate comparison.

We believe that financial risk associated with running a customer facing and properly hedged book for a particular product would be constant on a risk-adjusted basis. A book's VaR for a particular product should reflect, and be a function of, the volume of customer business in that product. A book that shows abnormally greater value-at-risk, relative to the amount of customer risk, would be suspected of being run in a proprietary manner. However, the added risk would have to be proven long-term and not just a daily anomaly.