Nate Silver's diminishing returns

Nate Silver's successful prediction of the election outcome should be celebrated as a triumph of science over waffle-monkeys and reality-deniers. But I fear there's a danger here.

His success should have two consequences: others should emulate his methods; and those who this time bet against his forecasts on the basis of wishful thinking will be unlikely to repeat their error.
This means that it will be harder to make money by betting on the next election, as markets will be pricing in more scientific method and less flim-flam.

How, then, will a Nate Silver wannabee make money with such poor odds? He can only do so by betting with borrowed money - in effect leveraging up to improve his returns.

But this carries two dangers. One is exposure to tail risk, as even the best methods occasionally go wrong: as Nate said, 8% chances do come up once in a while.

The other is performativity risk. If voters believe that a particular candidate is likely to win, they might not turn out. Because the leading candidate - by definition - is the one who attracts widespread but lukewarm support, he might be more vulnerable to this, which would put the election result back into doubt.

On both counts, our Nate wannabee risks losing a lot. He is, in Taleb's phrase, hoovering up pennies in front of a steam-roller.

What I'm describing here is no mere possibility. It's a pretty close analogy to what happened with mortgage derivatives; people geared up, thinking they were onto a safe way of making money but in fact were magnifying risk rather than reducing it.

This is an example of what Andrew Lo calls adaptive markets. A smart guy (Nate Silver) sees a way of making money, but as people emulate him, profit opportunities turn into losses. Just as there are population cycles in biology, so there are profit cycles in finance: the waxing, waning and waxing again of returns on small cap stocks since the 1980s is just one example of this.

And it's not just in financial markets that this is true. Billy Beane and his hero Arsene Wenger (pbuh and a painful death to his enemies) have seen their success dwindle as lesser men emulated them (though perhaps this also shows that big money beats brains).

The message here is that intelligence, science and ability might make you money - but perhaps only briefly, and trying to repeat your tricks can lead to losses.

Comments

[Anonymous]

Or, as Maggie Thatcher once said, "You can't buck the market" (Not in the longer term anyway).

[Brian]

Nate Silver wrote a few months ago that it was no great feat to make an accurate prediction on the eve of an election as the quantity and quality of polls were quite high and there is no time left for surprises. He pointed out that the bar for being good at predictions was very low when he entered the fray.
The real challenge is to make accurate predictions several months out and to capture the changing nature of the race. His belief is that the story of the day that gets pundits excited has little impact and that races are a lot more stable than the news would have us believe. Therefore in future there may be money to be made in betting against the many "game changers" in a race.
Of course, as you point out, if the models prove consistently accurate then they will be priced in to the betting markets.

[Steven Clarke]

It's like Asimov's Psychohistory: the predictions only hold when people are ignorant of them and so don't alter their behaviour because of them.

[ian]

Did Taleb really coin that phrase? I can recall it as fairly common currency in the financial world in the early '00s and didn't link it with NNT.

[FromArseToElbow]

Two observations:

1) Having gone out of fashion around about 2008, mathematical models are now back in fashion in 2012. DSGE is still a dirty word (OK, acronym) while Bayesian is the new hip.

2) Wenger's success has not dwindled because lesser men have emulated him. It's entirely a matter of money (see here). In fact, Wenger has continued to be highly successful, relative to his resources and the competition.

[FromArseToElbow]

Here being: http://fromarsetoelbow.blogspot.co.uk/2012/11/how-to-frame-argument-with-ai-wei-wei.html

[Tom Addison]

I'm with FromArseToElbow, it was money rather than lesser minds that pushed Wenger out of title-contending positions, there was no way he could squeeze enough from what he had to keep up with Abramovich's billions.

http://www.amazon.co.uk/Pay-As-You-Play-Success/dp/0955925339/ref=sr_1_1?ie=UTF8&qid=1352496499&sr=8-1

[Amouse]

Frightening that anyone writing about finance would think "pennies (or nickels) in front of a steamroller" is Taleb's phrase

Phrase is from before 1997

[Tim Newman]

"Billy Beane and his hero Arsene Wenger..."

What a shame there wasn't a line in Moneyball where Brad Pitt said this!

[Andrew]

Chris, one would think that you had just discovered the concept of arbitrage.

I feel your conclusion is incorrect. If you use intelligence, science and ability (well really just intelligence then) then you won't continue to try to exploit a market mispricing that no longer exists. You'll find another one.

[chris]

Andrew: 2 points:
1. the point of evolutionary finance is that mispricings appear, disappear and re-appear. This is contrary to the standard view, in which arbitrage causes mispricings to disappear permanently and cause the market to become efficient.
2. What makes you think the smart money will find another nice mispricing? It could be that the mispricings are far from the individual's sphere of expertise. One reason why many hedge funds fail is that experts in one area of the market often can't swiftly become experts in another, but simply lose their edge.