!VAR HAGENDOORN

Beating the Market with Mathematics

In 1990 the Econometric Society of the Erasmus University Rotterdam invited the maverick Belgian entrepreneur Jean-Pierre van Rossem to speak at its annual symposium. It was probably one of the best-attended lectures in its history. I remember students elbowing to get in. In the previous year Jean-Pierre van Rossem had made news headlines with Moneytron, a quantitative investment fund allegedly based on an econometric model, which could systematically beat the stock market. It was, in all likelihood a scam, part Ponzi scheme, whereby the returns paid out to the initial investors are financed by the deposits of subsequent investors, part money laundering, with the econometric model as window dressing. A few months after his lecture the fund collapsed. In 1995, after a brief stint as a politician, he was convicted of fraud.

Van Rossem’s reason for publicly speaking about his model was that he also wanted academic recognition for his ideas, or so he claimed. Following the lecture his assistants sold copies of the manuscript in which he outlined the alleged theoretical foundations of his model. I managed to get hold of a copy. It’s as confused as George Soros’ Alchemy of Finance, but in-between the lines there are some interesting ideas.

He writes that any econometric model should be based on a consistent and coherent general theory and that many models in econometrics are purely ad hoc and either mix elements from different and possibly contradictory theories or lack a foundation in economic theory altogether.

Much of the manuscript is devoted to a critical appraisal of modern portfolio theory, the efficient market hypothesis, the Rational Expectations Hypothesis and various strands of economic theory (neo-classicism, neo-ricardianism, the Chicago school and different forms of Keynesianism). It then moves on to develop an alternative theory, based on four categories of economic agents, distinguished by the information they base their investment decisions on, e.g. technical analysis or fundamental analysis, whereby the price is expressed as the weighted sum of each agent’s price expectation.

While I was at university I once sat down with a friend to see how far we could get in replicating the model. But we didn’t have access to all of the data and at some point we got lost in all the equations.

At the time Van Rossem may have employed some econometricians who were actually working on the model, but I guess those who did were ashamed to come out, at least I don’t recall ever reading an interview with any former employees.

In recent years some quantitative hedge funds have proved that it IS possible to systematically beat the markets with mathematics and econometrics. Or at least, for the time being. The current reigning emperor of quantitative investing is Jim Simons, of Renaissance Technologies, a former mathematics professor with some influential papers to his name.

The January issue of Bloomberg Markets magazine has an extensive profile of and interview with Jim Simons. It’s an interesting article, but don’t expect any insights into the models used by Renaissance Technologies. Quantitative hedge funds closely guard their models and require their employees to sign non-disclosure and/or non-compete agreements. By the way, when I was working as a quantitative analyst it said something similar in my contract.

As Andrew Lo, a professor of finance at MIT, says in the article, since Renaissance Technologies is so successful a lot of people like to speculate about what they are doing. Well, perhaps they have a very sound risk management, to limit their downside. And perhaps they’ve just been lucky over the past few years. The biggest danger is ascribing what may come down to mere luck to your own talent. Perhaps at Renaissance Technologies they are aware of this risk and for that reason down to earth. Only time and the numbers will tell.

Links

Richard Teitelbaum, The Code Breaker, Bloomberg Markets, January 2008.