A Stock Price Predictive Model Based on Changes in Ratios of Short Interest to Trading Volume

The findings, based on price movements alone, are that trading on the assumption that a large (small) proportion of increases in stocks' short ratios is bullish (bearish) produced significantly better than expected results at all Alphas tested above .02. This test-by-predictiveness strongly supports the validity of the assumptions.A test of the components of the short ratio produced no evidence that the success of the ratio as a stock market predictor can be attributed to either of its components singly, i.e., to either changes in short interest or changes in trading volume.A second test of the assumptions incorporated in the SR Expectations Model consisted of a comparison of model results against buy-and-hold results. In all cases except at Alpha .01 model results exceeded buy-and-hold results–and greatly so at Alphas above .02, thus strongly supporting the validity of the assumptions.The test against the buy-and-hold “control†standard was then extended to incorporate dividend and commission considerations. These considerations sharply reduced the model's performance. Therefore, an alternate strategy was tested which markedly reduced commissions, offset the opportunity cost of dividends missed when the portfolio was not long stocks, and avoided the explicit dividend drains caused by short positions. This strategy consisted of substituting Treasury Bill holdings for short positions in the basic model. This policy, consisting of switches between Treasury Bills and SSP “stocks†in accordance with Alpha .05 signals produced a terminal portfolio value greater than the buy-and-hold policy even after the introduction of a 30 percent income-tax consideration. Moreover, this higher return was generated with less risk than that inherent in the buy-and-hold policy.With respect to the optimum filter, it was found that Alpha .11 was best for price predictive purposes, with Alpha .05 a close second, but that once commission considerations are allowed for, Alpha .05 was the optimum filter.In conclusion, the hypothesis of this study is that when speculative expectations become extremely one-sided, a high probability exists that stock prices will reverse towards the unanticipated direction. This view is consistent with the theory that the stock market is generally efficient, but not perfectly so. A test of a model using changes in short ratios as a measure of shifts in investor expectations is consistent with the hypothesis: returns generated by the model significantly outperform random expectations. The test indicates a systematic tendency for investors to over-discount events when an overwhelming majority share the same optimism (pessimism) about future stock prices.