Smarkets Blog

Since Hunter and I are from the financial industry, it’s been a major goal of ours to make acquaintances within the betting world. So, when a leading bookmaker based in Gibraltar invited us down for a meeting, we jumped at the chance.

As travel often goes, we encountered a few glitches with transportation, learning that the hardest part about starting Smarkets is getting out of Gibraltar.

Last night I gave a talk at the London EUG about how we use Erlang at Smarkets to build web applications. If all was successful, a video of the presentation should be available shortly. The slides of the presentation are available here as a PDF.

The main purpose of the presentation was to cover the application stack that we use including MochiWeb, Erlang Web Gateway Interface, SG Template Engine, and Smak, the beginnings of our open source framework. (The name? Polish for flavour).

Thanks to everyone who attended the talk!

Election betting odds in the week before election have indicated correctly in advance the results of Presidential fights in every campaign in the last thirty-six years, since Wall Street election wagers became common.

New York Times – October 30, 1916

Prediction markets were popular one hundred years ago on Wall Street. Former WSJ publisher L. Gordon Crovitz writes that “$165 million in today’s dollars was wagered in Wall Street on the 1916 election – twice the amount spent on the election campaign itself.”

Today, they’re regaining popularity. However, some are pointing to the failings of prediction markets, claiming that they get it wrong all too often, as Barry Ritholtz did in his post Why Prediction Markets Fail.

Markets always act on incomplete information, so it is effectively impossible for prediction markets to correctly indicate an outcome every time. If markets fail to predict a winner, that does not invalidate the idea. Furthermore, what many naysayers fail to point out is that the technical infrastructure is in its infancy. Even prediction market giants Intrade and Betfair, as Barry calls it, do not “parallel the population at large” because of poorly designed interfaces.  This encourages low widespread adoption leaving only highly motivated yet homogeneous traders in the marketplace.

So my prediction is this: prediction markets have been around for a long time and have shown to have empirical value. A March 2008 article in Scientific American reports that the Iowa Electronic Market “was closer to the outcome of an election 74 percent of the time.” There is no question that prediction markets are useful, fun and will play a bigger role as the technology improves.