From 1980 to 2001 I've sorted out positive and negative days, and calculated the average move the next 1, 5, 10, 100 and 1000 trading days

Move	Number of trading days after move
	1	5	10	100	1000
General Motors
Up	.087 %	.134 %	.349 %	4.868 %	40.680 %
Down	.028 %	.366 %	.622 %	5.619 %	43.007 %
Coca-Cola
Up	.149 %	.377 %	.730 %	8.298 %	123.939 %
Down	.042 %	.506 %	.945 %	8.749 %	121.261 %
IBM
Up	.063 %	.226 %	.602 %	5.434 %	80.824 %
Down	.061 %	.386 %	.646 %	6.794 %	83.170 %

In GM and Coca-Cola we see a huge difference the next day. Buying on a positive day will give - on average - a better move the next day. However, this is not the case for IBM.

For a longer time frame, buying on a negative day shows up to be the most profitable.

As I mentioned in the S&P 500 test, these numbers may not be the case today because of the computer revolution and ever-changing cycles. However, let's say the effect still exists. Then it's clear that you should buy these stocks on a negative day. If you buy on a positive day, the estimated long time return is a so much lower that you would better wait until a down day. On the other hand, If you buy on a down-day the next day move is still better than the interest rate the next day, so there's no need to wait.

When it comes to selling, the average next day move is better than the interest rate on money (actually the lowest next day move, GM; 1.00028^365 - 1 = 10.8% annually). Therefore selling is - at least according to this test - something that should not be timed, but postponed to as late as possible. On the other hand, if you have a good alternative placement of the money a comparison of the expectancies must be considered, and selling on a down-day is generally better in the really short run.