So what has happened historically after a situation in which the difference between stocks trading above their three-day averages moves 50+% in a single day? By definition, that captures a move in which we start with most stocks below their short-term moving averages and finish with most of them above.
Going back to mid-2006, when I first began archiving these data, this has occurred 39 times in a VIX environment < 20. Over the next five trading sessions, SPX has been up 24 times, down 15 times for an average loss of -.26%. Winners have been more common than losers, but the average size of losers has exceeded that of winners handily.
No edge in that particular pattern.
Which can be information.
But suppose I *need* to find an edge. I keep running historical queries until something "significant" pops out. Perhaps I even fall prey to confirmation bias and keep running the studies until something supports my preexisting market view. If we run 20 studies, we've got a decent shot at finding the one in 20 that meets a significance criterion at the .05 level!
Confirmation bias is an understandable challenge in a field where confidence is needed to put on trades. Running studies until something looks good--or cherry picking the charts to look at--is an understandable bias when traders are seeking confidence.
Sometimes, however, the answer is that there is no meaningful directional edge in a given market. It's like drawing a poor hand at the poker table. The pros know when to play and when to muck their hand. There is an important difference between the motivation to trade and the motivation to make money.
Further Reading: Avoiding Confirmation Bias
Posts
