Technical analysis has been so pummeled by evidence supporting the efficient nature of market behavior that most researchers gave it up for dead years ago.' By now it seems a foregone conclusion that any sort of general trading rule based on past prices is bound to fail. That is, any general trading rule is quite unlikely to pick up sufficient dependencies to overcome the extra transactions costs of a switching strategy compared with a buyand-hold strategy. Trading rules which deal with special situations, however, may still be found to be useful. For example, Smidt suggests that there may be price dependencies associated with demand for liquidity, reaction lags, and speculative bubbles.2 Following up Smidt's hypothesis, several studies suggest profitable trading strategies associated with block trading.3 On the other hand there have been many attempts to find usable price dependencies associated with splits, dividends, and earnings reports, with mixed results.4 This paper is in the tradition of those studies which have searched for "special situation" price dependencies.