Abstract EN:

When a factor is missing from a given asset pricing model, the resulting squared alpha should be positively related to the residual variance. The alpha obtained in parallel with idiosyncratic volatility estimation reflects the short-term reversal. The asymmetric allocation between positive and negative stock return generated by sign of squaring explains the puzzling relation between idiosyncratic volatility and expected return. Further, employing extreme return leads to the proportion of positive returns is much higher relative to negative ones, explaining the return anomalies related to daily extreme returns. In the second essays we confirm recent results for US data and find evidences of higher performance of residual return reversal strategy than those of conventional reversal strategy. Moreover, the residual return reversal in the EU sample are robust to market microstructure biases. However, the results are heterogeneous across countries, are robust in France and Germany and seem to be fragile in smaller countries. In the third essays, we show that a simple subtract the average of past 12-month return from previous month return allows us to alleviate the momentum return. Consequently, the reversals are significantly stronger. We also find that states of market have significant impact on reversal profit indirectly through momentum effect. In down market, when momentum effect appears weak, the profit of reversal strategy is significantly higher than in up market, when momentum effect is strong.