The winner effect in humans


Jason Collins


June 11, 2015

I am using some material from John Coates’s excellent The Hour Between Dog and Wolf for a presentation I am giving next week, and decided it was worth sharing here:

During moments of risk-taking, competition and triumph, of exuberance, there is one steroid in particular that makes its presence felt and guides our actions – testosterone. At Rockefeller University I came across a model of testosterone-fuelled behaviour that offered a tantalising explanation of trader behaviour during market bubbles, a model taken from animal behaviour called ‘the winner effect’.

In this model, two males enter a fight for turf or a contest for a mate and, in anticipation of the competition, experience a surge in testosterone, a chemical bracer that increases their blood’s capacity to carry oxygen and, in time, their lean-muscle mass. Testosterone also affects the brain, where it increases the animal’s confidence and appetite for risk. After the battle has been decided the winner emerges with even higher levels of testosterone, the loser with lower levels. The winner, if he proceeds to a next round of competition, does so with already elevated testosterone, and this androgenic priming gives him an edge, helping him win yet again. Scientists have replicated these experiments with athletes, and believe the testosterone feedback loop may explain winning and losing streaks in sports. However, at some point in this winning streak the elevated steroids begin to have the opposite effect on success and survival. Animals experiencing this upward spiral of testosterone and victory have been found after a while to start more fights and to spend more time out in the open, and as a result they suffer an increased mortality. As testosterone levels rise, confidence and risk-taking segue into overconfidence and reckless behaviour.

Could this upward surge of testosterone, cockiness and risky behaviour also occur in the financial markets? This model seemed to describe perfectly how traders behaved as the bull market of the nineties morphed into the tech bubble. When traders, most of whom are young males, make money, their testosterone levels rise, increasing their confidence and appetite for risk, until the extended winning streak of a bull market causes them to become every bit as delusional, overconfident and risk-seeking as those animals venturing into the open, oblivious to all danger. The winner effect seemed to me a plausible explanation for the chemical hit traders receive, one that exaggerates a bull market and turns it into a bubble. The role of testosterone could also explain why women seemed relatively unaffected by the bubble, for they have about 10 to 20 per cent of the testosterone levels of men.

Coates tested this on a London trading floor:

I set up an experiment on the trading floor of a mid-sized firm in the City of London. The floor employed 250 traders, all but three of whom were men. They were all engaged in high-frequency trading, … meaning they bought and sold securities, sometimes in sizes ranging up to $1 or $2 billion, but held their bets only for a matter of hours or minutes, sometimes mere seconds. They therefore occupied the same market niche as the black boxes.

These traders were therefore up against some of the world’s most sophisticated and well-capitalised competitors. They lacked the large capital base and informational advantages of the flow traders at the big banks, and the deep pools of capital and inhuman processing speeds of the black boxes. Yet they were astonishingly successful: David against Goliath, John Connor against the Terminator. In fact they were some of the best traders I have ever seen: highly disciplined, consistent, and profitable.

I sampled testosterone from these traders and recorded P&L over a two-week period. What we found was that their testosterone levels were significantly higher on days when they made an above-average profit. More intriguing, though, was what we found when we looked at testosterone levels in the morning, because these predicted how much money the traders would make in the afternoon. When the traders’ morning testosterone levels were high, they went on to make a lot more money in the afternoon than they did on days when their morning testosterone levels were low. Moreover, the difference in P&L between high- and low-testosterone days was large, amounting in statistical terms to one full standard deviation, a difference that if annualised could amount for some of the traders to over £500,000 in pay.