Institutional Behavior At The Close: Does It Affect The Next Open?

When institutions enter large numbers of buy or sell orders on an MOC (market on close) basis, does this affect the next open?

I went back to 2004 (N = 744 trading days) and divided the sample in half based upon the net buying vs. selling in institutional MOC orders. When there was relatively strong buying at the close (N = 372), the next open in the S&P 500 Index (SPY) was up by an average of .01% (168 up, 204 down). Conversely, when there was relatively weak buying at the close, the next morning's open in SPY was up by an average of .06% (225 up, 147 down). Interestingly, when there was outright net selling at the close (N = 87), the next morning's open was up by an average of .11% (55 up, 32 down).

If you've ever had the feeling that you bail out of positions at the wrong time, at least you now know that you're in good company. Even at the market close, at a very short time frame, we see evidence of reversal effects.

So here's the puzzle: How do we know what institutions are doing at the market close? I'll verify the answer if anyone figures it out and posts it as a blog comment. It's actually pretty simple.