The prospect of a perfect ending: Loss aversion and the round-number bias

•We develop a theory of round-number bias via prospect theory and analog heuristics.•We test the prediction that bias is greatest in choices involving loss (cf. gain).•The theory is supported via analysis of both laboratory and real-world data.•Findings reveal the importance of the editing phase in prospect theory.•Findings have broad implications for inefficiencies in financial markets.

Studies across a range of domains have shown that individuals tend to focus on round numbers as cognitive reference points; a so-called left-digit effect. We explain this effect by combining analog numerical heuristics with prospect theory in order to develop an analog value function that predicts the key characteristics of the left-digit effect. Most importantly, this value function predicts an unreported phenomenon, namely; that the left-digit effect will be more pronounced in situations involving losses (cf. gains). We confirm this prediction in both a laboratory experiment regarding hypothetical investments and analysis of buy–sell imbalances in over 15 million trades by investors in a financial market. We conclude that our analog value function is a promising explanation for the left-digit effect. Furthermore, we suggest that interventions aimed at reducing costly buy–sell imbalances in financial markets should focus on the decisions made by investors when they are facing loss.