Predicting Venture Success and Prediction Markets

This an interesting article in the New York Times, about a small startup that has got a lot of high powered funding to predict the success of startups. As the NYT notes in its article:

The idea of a start-up predictor has drawn skepticism. Some venture capitalists say that the idea of using formulas or historical data from past deals to predict how other start-ups will do in the future has been tried many times in vain.

Paul S. Kedrosky, a venture capitalist and the author of the Infectious Greed blog, said that his industry was indeed inefficient at picking winners; typically, 90 percent of venture investments are not home runs. But he does not particularly trust a company that professes to be able to do better than venture capitalists.

“If their tool did such a good job, they’d raise a fund themselves and beat the tar out of us,” Mr. Kedrosky said. “It’s hard to imagine what they’re mathematical combination of factors is.”

On that point, the founders of YouNoodle.com are not forthcoming. They say their algorithm uses sophisticated modeling pertaining to how social capital and networks can affect an organization’s performance.

It would appear to be a play on the magic of the de jure darling, social networks etc (the "who you know" aspect) in driving success. I would be fascinated to know how they will measure Social Capital, its a very interesting area and we also are of the opinion that its a major determinant of success, but its metrics are currently very new.

Good luck to them anyway, we've done quite a bit of work in predicting market success for various technologies and startups over the last few years (at better rates than 10%, Mr Kedrosky ), but found that it is decidedly non-linear (ie is prone to tipping points, black swan effects, "unkonwn unknowns" and other elements of complex system behaviour. After all our experience, I think the best approach is probably what is emerging as a fairly well proven approach - Prediction Markets (we've written on these before).

Last word to the NYT:

So the question arises: Has YouNoodle used the predictor to determine if it will itself succeed?

“So far, we haven’t run ourselves through it,” Mr. Goodson said, adding that the results could prove baffling. “If it says we’ll fail, and it’s right, that’s something of a paradox.”