Читать книгу Uncommon Sense - Pape Scott - Страница 12

I saw a documentary a few years ago called The Next Nostradamus. It featured political scientist Professor Bruce Bueno de Mesquita, from New York University. He'd developed a predictive computer program based on game theory. While Bueno de Mesquita didn't come up with the idea of game theory, his name has become closely associated with it.

Game theory is the study of strategic decision making. It aims to predict outcomes by first identifying who the big movers and shakers are in any deal. Then it identifies the outcome each player is looking for. That's relatively easy. Put yourself in their shoes: humans are driven by self-interest, so what benefits each is the outcome they're driving for. Next, assess how much influence each player brings to the table. The biggest mover and shaker has the greatest influence on the outcome. Then throw all the variables into a computer and let it spit out the answer.

But Bueno de Mesquita reckons his technique isn't applicable to all situations, the stock market being one of them. He applies it principally to the political arena. Is Iran going to produce an atomic bomb? Is the President going to be re-elected?

So if it can't be applied to stock markets, why have I mentioned it? The reason is that Bueno de Mesquita is an intelligent guy who has devoted his career to making a science out of prediction. He believes in prediction. And even he is ruling out the stock market as something that's predictable.

In February 2009 he delivered a presentation on game theory in Long Beach, California, as part of the popular TED series. He told the audience that the use of game theory didn't facilitate stock market prediction. But he immediately followed with a tongue-in-cheek stock market prediction of his own: ‘Okay, it's not going up any time soon.' He was spot on with his ‘inability to predict' comment: 12 months later the Dow was up by 21 per cent.

But in the same TED presentation Bueno de Mesquita proposed that the power base of Mahmoud Ahmadinejad, the President of Iran, was being eroded and that he would soon cease to be a force in Iranian politics. He was correct on the first point but wrong on the second. Ahmadinejad was re-elected to a second term just four months after the TED presentation. But there were cries around the world that the election was rigged – the only way Ahmadinejad could regain office. Now you could argue that the model couldn't be expected to predict foul play. But on the other side of the coin the model is based on consideration of the political influence each player brings to the game. Clearly Ahmadinejad possessed more influence than Bueno de Mesquita had dialled into his model.

While game theory presents nuances upon subtleties upon uncertainties upon judgements, Bueno de Mesquita reckons he gets a 90 per cent strike rate when using it in the appropriate arena.

One very interesting aspect of game theory is the number of potential interrelationships between the players in the game. Bueno de Mesquita explains it this way: if you're considering a problem that involves five participants, there are 120 different ways in which those five can interact. This figure is calculated by using the following factorial operation:

Factorial 5! = 5 × 4 × 3 × 2 × 1 = 120

Adding extra factors or participants sees this number expand dramatically. For example, consider 10 players. The potential number of direct interrelationships is 3.63 million. One hundred players (defined by factorial 100) is a staggering 9.33 × 10¹⁵⁷. That's nine (and a bit) with 157 zeroes on the end!

You can start to see why game theory can't be applied to financial markets: the number of variables in the stock market is mind-boggling. There are unknown factors that impact unknown factors that impact unknown factors that impact the unknowable outcome – multiple layers of unquantifiable complexity that are impossible to imagine, let alone measure.

As if that's not enough complexity, yet another layer needs to be added. The system is dynamic. The factors and the relationships between them are forever changing. There are no constants.

Bueno de Mesquita is the first to list the limitations of game theory. He doesn't hold it out as a tool for long-term prediction under any circumstances. The players must actually be in the act of playing the game when the call is first made. I know it's a fine point, but this means he isn't really making absolute prophecies; rather he's predicting outcomes. But, despite its limitations, game theory is one of the more credible ‘prediction tools' around.

So is Bueno de Mesquita's model similar to those used by economic forecasters – the ones politicians, economists and stock brokers use to formulate their bold claims regarding our economic future? Let's investigate.