Where humans see patterns, machines see opportunities

Published in The Sunday Times on 4th October 2026.

Is rock-paper-scissors dying out as an arbiter of chance?

A friend of mine recently suggested he didn’t know the rules when we needed to split 2 rooms amongst 3 of us. Curiously enough, he won his version of arbitration – names in a hat - which he denied were all his. 

Rock-paper-scissors is simpler – and objectively fairer. It’s a random process in which chance dictates that your odds of winning are even – akin to a coin toss. But it turns out, people do not play randomly - they follow hidden patterns.

A large rock-paper-scissors tournament at Zhejiang University in China, recruited students and divided them into groups of six. Each competitor played 300 rounds of rock-paper-scissors against other members of their group. As an incentive, the winners were paid - in proportion to their number of victories. 

Humans are predictable 

What they found was that winners tend to stick with their winning action, whilst losers, tended to switch -and they did so in order of the name of the game - moving from rock, to paper, to scissors. 

This "win-stay, lose-shift" strategy is known in game theory as a conditional response. Whatever about the biology underpinning it, it produced patterns that machines were able to exploit. 

In experiments, an AI strategy was able to pick up on this bias and exploit it - defeating over 95% of human opponents.

This is extraordinary. In a game where chance should leave everyone roughly level, humans turned out to be so predictable that machines completely outmanoeuvred them. 

Fooled by randomness

Nassim Taleb wrote about this over twenty years ago in his seminal text, Fooled by randomness. Taleb argues humans are wired to mis-handle randomness, which makes us exploitable. We are “pattern-seeking” often finding meaning in random sequences.  

As machines take over larger and larger swathes of financial markets, this should provide pause for thought.  

Humans used to decide what stocks to buy in a fund – but stock pickers are vanishing by the day -– passive funds now hold over half of US fund assets. Of the active firms that remain, many of the world’s largest and most successful are increasingly systematic. And trading in financial markets has long since been automated.  

Taleb's core insight is not just that randomness exists. It is that we are very uncomfortable with it. We manufacture explanations, patterns and forecasts that make an uncertain world feel predictable. But it is here where I think we maintain an edge on the machines. 

As machines get better at detecting “non-randomness” in human decision-making, markets become more efficient at the micro level. But paradoxically, humans’ persistent biases  - driven by fear/greed and herd behaviour -  still create macro-level mispricing’s. Something Paul Samuelson recognised a long time ago. 

Focus on the long term is trite advice

One takeaway from all of this, perhaps an obvious one, is: don't play their game. Focus on long-term investing, diversification and avoiding the behavioural traps that make humans so predictable. This is trite advice – which I’ve spent years encouraging clients to adopt, with mixed success. 

The deeper lesson is that financial markets are not one homogenous thing. At the micro level, they increasingly resemble an engineering problem. Wherever there are repeatable patterns, measurable signals and well-defined rules – like rock paper scissors -  machines are becoming relentlessly better than humans.  

Public stock markets still offer a compelling proposition to investors. Stock markets are a positive sum game in an absolute sense. Economies grow, businesses create wealth, and shareholders participate in that growth over time. It’s only in a relative sense they are zero-sum - one investors gain is another’s loss. As machines get smarter, we should expect to find ourselves increasingly on the losing side of that trade.  

At the macro level there are opportunities 

At the macro level, markets remain a social system. They are driven by human hopes, fears, narratives and changing beliefs about the future. That's a much messier challenge. Economies evolve, politics is fickle, bubbles form, and investor sentiment continues to oscillate between fear, hope and greed.  

Outside of public stock markets, areas where relationships, judgement, negotiation and imperfect information matter, such as private markets, real assets and complex illiquid structures, remain more resistant to automation. It is also why wealth management is likely to prove one of the most resilient corners of the investment industry – an admittedly convenient conclusion from somebody in my line of work! 

We may be an easy mark for the machines when the challenge is identifying patterns in data. Investing has never been just a pattern-recognition problem. Markets are ultimately a study of human behaviour. And for all our flaws, outside of well-defined domains like rock-paper-scissors, humans remain wonderfully, frustratingly and profitably unpredictable.  

Gary Connolly is Investment Director at Davy. He can be contacted at gary.connolly@davy.ie or on X at @gconno1.