richardhutnik on 11 January 2010
| ioi said: Potentially if we can build it up enough... |
You might want to run something like intrade does: http://www.intrade.com/
Here people "wager" on predictions, and like they have betting lines in Vegas, where the line ends up is usually more accurate a predictor than other methodologies. I believe if there is a sufficient sample size, with the right mix, you can do something better than what we have been seeing like the recent thread on here pointed out. Combine this, with other methodologies, and you have a chance to end up with something that would work real well. Key would be to filter and weight things properly.
