The Performance of Stocks Earning Extreme Single-Day Returns: Evidence from Sweden

Sammanfattning: In 2011, Bali et al. presented evidence that stocks with extreme one and multi day-returns significantly underperform stocks with less extreme returns in the following month. They attributed this to investors exhibiting a preference for stocks with lottery-like payoffs. Motivated by this, we examine the relation between extreme positive daily returns and returns in the subsequent month in the Swedish stock market. Using a sample containing stocks listed on the largest Swedish stock exchange, NASDAQ OMX, from January 1984 to December 2018, we find significant evidence of a negative relation between positive extreme returns and subsequent returns. The negative relation is robust to controls for size, book-to-market, liquidity, momentum and short-term reversal as well as idiosyncratic volatility, skewness and lottery-characteristics on portfolio-level. Multivariate regression analysis suggests that past extreme returns only show a significantly negative relation with future returns when we control for other firm characteristics. The negative cross-sectional relation is more significant and more economically relevant during recessions and insignificant during expansionary periods and varies considerably within our sample period. Excluding microcaps makes the relation insignificant for value-weighted portfolios. We conclude that the underperformance of stocks with extreme daily returns in the Swedish market may be driven by a combination of very short-term reversal, cumulative prospect theory and an increased inclination to gamble during economically bad times. Limits to arbitrage may prevent arbitrageurs from exploiting and correcting potential mispricings.