One hope of Metaculus is not just to allow users to accrue a track record, but to assess, given the large set of statistics assembled, which predictors are doing much better than chance, i.e. are far enough out into the possible probability distribution that their predictions could not have been just luck.

Otherwise, it is difficult to sort out who has simply been lucky in the past.

As a case in point, consider this article about Sandy Jadeja, a technical analyst who has purportedly predicted four previous market turndowns, to the day.

How many people have been making such predictions? How many (if any) did he predict that did not come to pass? We don't know. What we do know is what he has predicted. From the article:

He now warns that the following dates spell trouble for the Dow Jones in the US that could spread to other markets.

Between August 26 and August 30, 2016.

September 26, 2016.

October 20, 2016.

Will he get any of these right?

Question resolves as positive if the DJIA has a one-day drop (market opening to close) of more than 5% on any of the 5 days mentioned above (the 27th and 28th are a weekend.)