After over a month tracking the predictions of some of America’s top pundits, today we are launching PunditWatch. TechCrunch covered the launch.

Bear Sterns is doing just fine shouted Jim Cramer just before they got bought by JP Morgan for $2. Google could hit $2,000 claimed Henry Blodget of Silicon Alley Insider just as their stock price peaked and started its long slide to under $500. Hillary Clinton is going to lose in New Hampshire claimed just about every commentator before her dramatic come back.

These are just some of the examples the recent examples of ‘experts’ failing miserably to predict future events. But we still listen to them. Are these isolated examples or does the seer/sucker theory hold true (”No matter how much evidence exists that seers do not exist, seers will find suckers”)?

The Pundits

Well now we will find out. PunditWatch’s mission is to track the predictions of public commentators to see how they perform. Initially we are going to select 10 commentators from the fields of Politics, Entertainment, Sport and Technology. The pundits are:

• Politics: Chris Matthews (MSNBC), Pat Buchanan (MSNBC, NBC), David Brooks (PBS, NYT)
• Entertainment: Perez Hilton, Nikki Finke (Deadline Hollywood Daily), DListed
• Sport: Peter King
• Technology: TechCrunch, VentureBeat, Silicon Alley Insider

There were plenty more we wanted to add but after exhaustive investigation we found that they don’t ever in fact make any real testable predictions. We do however plan to add a second wave of pundits to this list so do comment with your suggestions.

How it all works

We have created a Hubdub account for each of the pundits (for example the TechCrunch user id is techcrunch_pw).

1. Every Monday morning we will post up a round of all of their previous week’s predictions, a look at their performance and also a call for readers to be on the look out for them spotting new predictions.
2. During that week each time a pundit makes a prediction we will place H$10 on it from their Hubdub account.
3. If the question doesn’t exist we will create it with an estimate of the correct probabilities. We will then invite some of the top Hubdub forecasters to make predictions in order to set the prices. Then when the prices have settled down we will place a prediction for the pundit from their account.
4. We will update the PunditWatch blog with a post to say that the prediction has been placed.

Pundits can make anything from zero to three predictions per week. If they make more than three then we will ignore the later ones. After analysing many of these pundit’s predictions it is clear we are going to have to give some interpretation to their predictions to make them testable (hey, if they don’t like that then they can just start making real concrete predictions!). We will have to feel these out as we go along. One point is that if a pundit reports rumors then we will treat these as a prediction. If they don’t think the rumors are likely to be true then why report them?

Click here to check PunditWatch out.

Popularity: 9% [?]