Prediction Markets Vs. the Pundits in the French Election

Michael Giberson

On prediction market group blog Midas Oracle (and also posted at the NewsFuture blog), Emile Servan-Schreiber offers an assessment of “prediction markets vs. the pundits” after the French presidential campaign debate between Ségolène Royal and Nicolas Sarkozy:

After the debate, the pundits were all agreed that Royal had scored some points, and even die-hard Sarkozy fans openly worried that Royal had bested their champion. The next morning, newspapers and radio stations still conveyed the impression that Royal’s performance had probably helped her. However, the trading pattern one could observe on NewsFutures and other prediction markets told a different story altogether. The price of Sarkozy’s contract actually rose a little during the debate and just after… The next morning, this gain held, even as political pundits on the radio stations were still praising Royal’s performance. The disconnect continued until the afternoon, when the results of a poll taken just after the debate showed that it was Sarkozy that had come out on top, confirming the market’s impression.

So, once again, prediction markets performed quite well in an electoral context. This, of course, won’t come as a big surprise to anyone familiar with the field.

An interesting interpretation, but I think Servan-Schreiber is unfair to the pundits. His views seem to conflate opinions about who won the debate with conclusions about who would win the election. From my scan of news stories offering post-debate opinions, there is indeed a view that Royal may have bested Sarkozy in the debate, but I didn’t see a single comment not attributed to the Royal campaign that suggested she did so well she would overcome Sarkozy’s lead in the polls and win the election.

Unless a prediction market carried a contract on the debate winner, or the post-debate pundits were opining on who would now win the election, then the “prediction markets vs. pundits” comparison is a bit off.

Of course it is still true, as Servan-Schreiber said, that “prediction markets performed quite well in an electoral context.”

While Sarkozy and Royal contract prices at NewsFutures were in a near dead heat in January, beginning in February – several months before the election – the “Sarkozy wins” contract price began a relentless rise and post-February the Royal contract never got close to suggesting she was likely to win.

UPDATE: Over on the Midas Oracle blog I have a few additional brief comments trying to parse the prediction markets vs. pundits vs. polls issue.