Undercover Economist

Trading places – with a rat

Financial price data are converted into music, the music is played to a rat, then the rat guesses whether the price will fall or rise

Eighty-five years ago, a young psychologist called BF Skinner developed what is technically called an operant conditioning chamber but is more famously known as a Skinner box, designed to contain and train laboratory animals. The simplest version rewards a rat for pressing a lever. More complex devices can play sounds, display lights and even deliver electric shocks, although Skinner himself preferred to use rewards rather than punishments.

The Skinner box acquired an unpleasant reputation: rumours circulated that Skinner raised one of his own daughters in one, that she lost her mind, that she sued him and that she killed herself before reaching middle age. None of this is true. One explanation for the rumours is that Skinner did design an air-conditioned crib for his daughter and described it with fancy technical terms such as “apparatus”. But perhaps the false ideas circulated because Skinner’s ideas of modifying behaviour with rewards in a carefully controlled environment seemed somehow manipulative and threatening. These days, of course, we call behaviour modification “nudging” and it is perfectly respectable.

I thought of all this when I discovered RatTraders.com, a website offering, in its own words, “a professional service to the financial industry; rats are being trained to become superior traders in the financial markets.”

RatTraders displays photographs and short films of rats in Skinner boxes. As the website explains, “RATTRADERS rats can be trained exclusively for any financial market segment. They outperform most human traders and represent a much more economic solution for your trading desk.”

The brains behind the project is Michael Marcovici, who has a double life: he’s both a conceptual artist and an offbeat investment guru. So is RatTraders a work of art or a business proposition? The basic conceit is that financial price data are converted into 20 seconds of piano music, the music is played to a rat, and then the rat guesses whether to bet that the price will then fall or rise. If the rat is successful, it receives food; if not, a mild shock. After weeks of training, the site says, the best rats outperform most humans.
Rat graph illustration by Harry Haysom for Undercover Economist©Harry Haysom

In an online video, Marcovici deadpans to a “reporter” that “rats are much better at this because you can train a rat on a very specific thing. It will not be distracted.” The interviewer asks how much it would cost to buy one of these elite rats. “It’s hard to put a price tag on a rat. A rat will not help. Rat trading is a system,” responds Marcovici. But he adds that even the full service would be “just a fragment of what a human trader will cost”.

Just when the viewer’s credibility is strained to breaking point, Marcovici announces that investment banks have already installed trading floors full of rats in boxes. His ultimate business vision, he says, is that rats will also be trained to do marketing and general management. The comic timing is impressive. And it’s a piece of art with a message that does not need underlining.

Marcovici says that all this began as a joke. But after he had bought some rats and built a rather beautiful Skinner box, his girlfriend urged him not to get rid of the rats but to see if they actually could learn to trade gold, oil and foreign exchange. Most of them couldn’t but after five months a few of the rats were pretty good – better than tossing a coin, anyway, which is a tougher benchmark than you might think.

So should we start employing rats as traders? The idea is not totally absurd. Ben Vermaercke of the University of Leuven recently published a paper with four colleagues titled “More complex brains are not always better”. They showed that rats were better than humans at distinguishing certain kinds of striped patterns from others – the humans, it was hypothesised, were thinking too hard rather than relying on instinctive pattern recognition.

Perhaps a properly controlled scientific experiment would also find that rats make excellent foreign exchange traders. I doubt it. If rats really can make money after being played a sonified snippet of recent price history, that suggests a basic inefficiency in financial markets. If price movements were predictable in such an elementary way then profit-seeking traders, human or murine, should exploit the pattern and thereby change it. (If there are patterns in market prices, they are not simple.)

RatTraders is five years old but the project has recently enjoyed a fresh burst of media attention. Marcovici tells me that several small hedge funds promptly got in touch with him to find out more. Did he really expect serious interest in financially savvy rats?

“Yes, yes. Honestly I did,” he says. “I’ve been involved with the financial industry for some time and people try anything.”

Fair enough. Greed is a powerful motivator for humans and rats alike. But it is not always a great investment strategy. Some people wanted to believe the worst about BF Skinner. If there seems to be money to be made, others will gladly believe the best about trader rats.

Also published at ft.com.