Risky Finance 23 August 2007

The British cryptanalysts at Bletchley Park and at its postwar successor GCHQ would often face a dilemma. Their secret code-breaking techniques were sometimes fabulously successful, allowing decisions of critical military or policy importance to be made almost in real time. But being too obvious about this success would alert the enemy that his signals were being broken, drying up the flow of profitable information – or worse still, prompt a counter-flow of disinformation.

When Jim Simons, the topologist-turned-billionaire behind the hedge fund Renaissance Technologies, hired a dozen or so signal processing experts from British spying centre GCHQ a few years back, it cemented the mystique of quant trading. Utterly disconnected from the fundamentals of the markets, Renaissance functioned as a giant listening station, secretly gleaning predictive signals from billions of trades every day.

Such methods, used to similarly lucrative effect in funds launched by Goldman Sachs, work by attempting to identify hidden messages in the torrent of market data. It’s a needle-in-a-haystack process, requiring enormous computer power and the latest statistical machine learning techniques to pick up these secret messages and make money from trading on them. Unlike strategies based on derivatives pricing that involve holding some kind of illiquid asset in the belief that it is mispriced, proponents of quant trading only trade the most liquid markets, albeit in very high volume.

In the markets, there is no real equivalent to the German High Command or the KGB that the British codebreakers were up against. But the penalties for market code-breakers that are too obviously successful are the same: the signals dry up, become unreadable or misleading.

As Simons ruefully acknowledged in a recent letter to investors, a new coded phrase appeared in the markets that his computers didn’t understand. The phrase was ‘massive deleveraging’. Some would argue that Renaissance and the Goldman funds may have grown so big that they couldn’t get out of the way when this juggernaut appeared. But Simons blamed other quant funds that insolently encroached on his space, driving his portfolio down when they were forced to sell to pay margin calls on other positions, such as sub-prime asset-backed securities (ABS).

Where did these other quant funds come from? Many were developed in-house by broker-dealers as order-execution systems. Using the vast storehouse of client transaction records, algorithms for best execution in different market conditions could be inferred statistically, and a simple rewiring would transform the algorithm into a hedge fund strategy. It could be that the industrialisation of these strategies made life harder for Simons.

But there is one sure-fire way to hedge against risks that Simons faced, and that is to provide trading services to others. In the same way that a broker-dealer can rewire its block trading system into a quant hedge fund strategy, the algorithms behind a successful quant trading hedge fund can also be rewired to become an order execution system. Interestingly, Renaissance’s rival Citadel appears to have done exactly this, selling equity broking services to clients. When markets become turbulent, the fees earned from such services increase.

One doesn’t know if Jim Simons ever considered such an idea, but surely his former GCHQ employees would have told him the following anecdote. At the end of the Second World War, the British government found itself in possession of thousands of encryption machines captured from the German army. It made a tidy sum selling these machines to the armed forces of neutral countries. What it didn’t tell these countries was that it knew how to decrypt the codes that the machines produced. There’s got to be a lesson there for quant funds today.

This article originally appeared on www.breakingviews.com.  If you wish to take out a free trial to breakingviews, click here.

Copyright © breakingviews 2009