Why Humans Are Better At Investment Than Computers....

In my Money Week column this week, I've been looking at why humans are better at investment than humans. Here's a taster.


Two years on from the start of the credit crunch, it is easy to observe that not a great deal has really changed. The banks have gone back to paying big bonuses, the traders and dealers are as speculative as ever, and the hedge funds are still raking in fortunes.
Still, one corner of the capital markets has been hammered hard – the so-called-called ‘quant funds’.
A few years back, the intellectually super-charged hedge funds that used mind-bogglingly complex formulas to trade assets and make huge profits for their owners were the hottest sector on both the City and Wall Street.
But now the combined assets of the quant funds are down from around $1.2 trillion at the industry’s peak to around $470 billion now, a drop of more than 60%, according to data from the research firm eVestment Alliance. Around a quarter of the quant funds have closed in the last two years, according to figures from Lipper Tass.
In part that tells us that investing styles go in and out of fashion. Sometimes people want gurus, sometimes charts, other times they want geo-political trends, and so on.
But it also tells us something more interesting.
Despite the best efforts of thousands of incredibly bright people, and despite the billions of dollars at stake, no one has ever really managed to mechanize the markets. They remain stubbornly human.
The quants, as they became known in the markets, were obsessed with taking the human element out of their trading strategists. The scoured university campuses, taking astro-physicists and mathematicians blinking out of the library, and paying them hundreds of thousands to have a shave, put on a suit (or at least some Boss chinos) and sit all day in an office on Mayfair’s hedge fund alley.
Once installed, they came up with programmes that could trade on minute price discrepancies between different markets, and make a fortune on the results. Or the scoured the record books for past price relationships, and when they found them, built trading computers that could exploit them.
Sometimes they came up with interesting results. The price of oil, for example, expressed in gold has remained virtually static for generations: as soon as it deviated from that norm, there was a trading opportunity.
But, although they had some big successes in the bull market, the quants were undone by the crash. None of the complex mathematical models they built predicted the credit crunch. All the expensive computer programmes were about as useful as a bucket and spade in Birmingham. The quants reputation was largely destroyed. The reason the value of the funds fell so fast was because they preformed so poorly, and because disillusioned investors withdrew their money.
In fact, the markets remain impossible to mechanize. If you could build a programme that predicted the markets, you would make, quite literally, billions. Yet no one ever manages it. Indeed, the harder they try, the worse the results usually are. In the 1990s, another hedge fund, Longer Term Capital Management, which had more Nobel prize winners on board than Man City have expensive footballers, collapsed with vast losses and bought down the markets with it.
There are three reasons why the market remains so defiantly human – and so resistant to smart mathematical models.
First, the markets are chaotic. Computers are very good at capturing fairly straight-forward relationships. They are very bad at modeling complex ones. Chaos theory is one explanation for that. For example, a butterfly flaps its wings in China, and it causes a thunderstorm in Britain. There are so many complex inputs making up the way weather works, you can’t hope to capture them all. It’s the same with the markets. A mortgage defaults in Florida, and a month later The Royal Bank of Scotland is bust. They are inherently chaotic, and so incredibly hard to predict.
Secondly, computer programmes are very bad at capturing human reactions and emotions – and the more people are involved the worse they get. That’s why they are good at chess, but not very good at bridge or poker: both card games are essentially about judging what your opponent will do, whilst chess is mainly about crunching a lot of numbers. The markets are much more like a card game than a board game. Investment decisions, and hence the direction of the markets, are driven as much by emotion as anything else. Sentiment is strong some months, and weak in others, even if not very much really seems to have changed in the meantime. It is tough for any computer programme to understand that, let alone model it, and start building it into its predictions.
Finally, behavior changes. The way that investors behave and the markets respond, evolves all the time. It isn’t static, or predictable, like the way the moon revolves around the earth. It will be different today from yesterday, and different again tomorrow. The quant funds were building predictive models based on the way market behaved in the past. But, whilst interesting, they didn’t really discover anything very useful. Just because a price has moved in a certain way historically does not mean it will move the same way in the future.
The lesson is a simple one. The markets will remain an arena for great traders, with an instinctive feel for where assets prices are going. They can’t be predicted with any kind of precision by computer programmes, no matter how much brain-power has gone into creating them.
And all those astro-physicists who for a few years could drive around in Porsches as they got paid million by a hedge fund can go back to the dusty poverty of the university library.
Still, there is one comforting thought. You might find it fiendishly difficult to predict what any market will do in the next years. But the smartest brains from the best universities couldn’t do it either – the markets caught them out, the same way they usually do the rest of us.