Is trading success all in the numbers?

I guess I'm not the only one wondering if private trading has been rendered all but impossible by the current market volatility. What is the point of me carefully building a portfolio when the entire market can be blind-sided by a single media report and placed on an entirely new trajectory?

Lately, my long time friend and former research colleague Dr Alessandro Usseglio Viretta has been applying his considerable mathematical talents to this very problem. He is using a form of machine learning known as Support Vector Machines (SVM) to perform a complex analysis of the market in real time.

SVM works by classifying stocks into simple categories, for example ‘buy’ versus ‘avoid’, on the basis of reams of market data. But while the predictions of the algorithm are straightforward, the stock picking rules created by SVM are feverishly complex. Every piece of market data chosen for inclusion in a SVM analysis is carefully factored into the final trading strategy. “It really matters what you choose to put in the model,” explains Alessandro.

Alessandro has developed a platform able to perform SVM on timescales of minutes and even seconds. This is crucial for participation in markets such as foreign exchange, not to mention volatile equity markets. “That is where high frequency trading comes in,” comments Alessandro. And note that the New York Stock Exchange has recently introduced a fee structure that is more accommodating to high frequency traders.

With an intellectual rigor befitting his early graduate physics work at CERN, and professional verve reflecting his recent collaborations with quant financial services provider Oliver Wyman, it is not surprising that traders are coming forward to test their strategies using Alessandro’s computational tools.

But in the end, will it work? “I’m doing the math, but the important thing to realize is that I am collaborating with experienced traders,” explains Alessandro. “Together, we’re in with a good chance”.