Simon Carr: Predictions are interesting only if they predict that which is unpredictable

While billions are spent on rune-reading and forecasting, nobody can reliably do better than chance

The dollar's down and property's up. Normally it wouldn't matter. It fact, it doesn't matter much anyway, but two years ago, having sold my house to fund a project, I avoided the sterling slump that experts said was imminent by going heavily into dollars. The move was also designed to defeat the property crash that has been predicted every year since we came out of the last crash.

The commentary columns are quite unfazed by developments, and experts explain the reversal with the same wise authority that they'd be explaining why the dollar had risen and property had collapsed, if that's what had chanced to happen. Why we continue to fall for their assertions is beyond me. How have we failed to learn? Maybe we don't want to learn. Maybe we long to believe that someone knows. Or even that these things are knowable.

There have been occasions when I've felt I've known. I remember a feeling 35 years ago so powerful, at such a visceral level, that it cried out to be backed. There was a fabulous sterling crash; in a sudden slump the pound lost one-third of its value. When I ventilated my feelings in a wine bar, saying I'd seen it coming, a banker looked at me coldly: "If you knew it was going to happen did you put any money on it?" It was one of those blank, inescapable questions that concealed vast contempt for people who claimed such knowledge. It put me on a track of backing my hunches. I developed an ability, I really believed for a while, to move markets. Everything I backed failed, flopped, rolled over on to its back to present its fat, white belly to any passing predator.

Nobody knows what's going up when, and nobody knows what's going down when. Even inside knowledge can point a fellow in the wrong direction. I was once given a speech-writing package for a national telecoms company. They were making half their workforce redundant. That would go punished in the market, I reasoned. People wouldn't like the sense of desperation. A company on the verge of collapse. Clearly, I should sell the shares short in the market. As insider trading wasn't even illegal at the time, I could have lost a fortune. As soon as the redundancies were announced, the share price rose and quadrupled in 18 months.

My incompetence in these matters had even further to go. While the company was going from $2.50 to $10, I bought and sold its shares three times and lost money on each transaction.

It is tempting to make a general rule out of one's own imbecility: "I know nothing, therefore nothing can be known." More legitimate, perhaps, to make a general rule out of our general imbecility. For predictions to be interesting they have to predict that which is unpredictable. That's why interesting predictions are only right at random.

Some predictions that fall into that category:

* Whether it will be a cold winter or not.

* Where the next hurricane will hit the US mainland.

* What will prompt the next world war.

* When the dollar will rise and property fall.

* Who will win the next election.

* When the economic cycle will end.

If anyone could reliably predict any of these things they would be rich. Unimaginably rich, in fact. They'd have all the money in the world. It would be like knowing the race card results before the day.

But while hundreds of millions, indeed hundreds of billions of dollars, are spent on tipping and forecasting and rune-reading, nobody can reliably do better than chance in their forecasts. The chimps who outperform stock analysts are famous; there is also a team of Swedes who often do better than City economists: they stand on the other side of the room from a broadsheet share-price page and throw darts into it.

This isn't to say that people can't get the right answer; it says that people can't regularly get the right answers. There will be individuals who outperform random chance, but it is impossible to identify who those individuals will be. And the fact that a person or a company has made predictions better than chance for nine years in a row doesn't mean the tenth will follow the trend. They are called winning streaks, after all.

Much conspires against the sceptic in these matters. Most people quite want to believe and there are others who really want to be believed. There's a lot of money and great reputations to be made for those who look like they know what they're talking about.

Ed Balls put a small but significant part of his reputation into the idea that the Government would balance their books "over the economic cycle". I used to ask him a question whenever we met (I thought it got funnier, as time went by): "Where are we in the economic cycle, Ed?" "I'll tell you to within two decimal points where we are," he once said.

Nobody jeered; even I stopped after a while. But the fact remains that the beginning and end of the cycle are fundamentally unpredictable; it isn't even clear in retrospect when they've started or finished. In this last cycle, the Treasury has put the end of the cycle back twice, and its beginning back once.

But appearing to know when the cycle starts and finishes is important to the air of omniscient government officers like to project. It dramatises that they have access to translunary information that we in the secular world can only dream about.

We are also deceived by the fact that economic forecasting has become more accurate. This is partly due to an interesting phenomenon. The naïve forecast outperforms the professional forecast. Over time, the prediction that the weather tomorrow will be like the weather today outperforms more ambitious predictions based on airflows and atmospheric pressures.

In economic terms, we've been through a long period of steady growth, partly as a result of so many independent central banks being established. Thus, next year's results really have been more like the previous year's performance. The naïve forecast has been fortified by reality. And when you can control the volume given to your successes and failures you can play up your hits and play down the misses. It gives the business of prediction a much better name than it deserves, and keeps us subjugated in our fretful idiocy.

It's why we live in an anxious age. We have come to believe anything that is spun to us with enough vigour. Knowledge is power, we used to say, but belief is more powerful still.