Trading Momentum: How Shares and Investors Behave During Momentum Periods

Abstract

In this paper we argue that momentum profits are driven by both past performance and the relative proximity to an available reference point, the 52-week high. We construct momentum-style portfolios that are driven strictly by past returns which we call ‘run’ based measures, and compare these to portfolios driven by nearness to the 52-week high, defined as ‘range’ based measures. While both run and range based construction techniques generate positive momentum profits, they appear to be driven by separate behavioural biases. Portfolios constructed on the basis of range are driven by anchoring, while those constructed based purely on past returns are driven by representativeness and the hot hand fallacy. Portfolios selected by combining run-based methods with range-based methods outperform those selected on a single metric alone.We then undertake an analysis of order imbalance in individual and institutional investors in the six months following high and low range and run variables. The results show that individuals have significantly lower OI in all but one measure supporting that prediction that individual biases cause investors to be contrarian, thus leading to observed momentum. We lastly explore limit order use by investor class and find that all groups are sensitive to the 52 week high and significantly increase their use of limit orders. While this is not the case for past returns thus suggesting limit order use is strongly related to anchoring bias.