Idea that hit me today while driving — there is a lot of timing bias in the behavior of an individual stock due to the fact 1) humans are on a daily cycle and 2) opening prices gap from yesterdays close / close positioning. There is also the fact of after market hours news to move prices.
Thing that I am looking for — model a stock performance as a random variable that is *normally distributed* <- we find that modeling the daily return of $AAPL or $MSFT is not normally distributed (because of things like October 1987 <- that is an event that is so many standard deviations off the curve that it olny had a 10^-79 probability event, but it happened anyways). Hypothesis: We know daily price movements are NOT normally distributed, but perhaps the price movements from, say 11am to 1pm ARE normally distributed.
Check the correlation of $XXX from daily performance to 11am-1pm performance. Are they correlated for something like $AAPL? What is the 1 year return of $AAPL using only 11am-1pm vs full day performance? Need to test this and report the findings here later.
