Beyond Performance: Exploring trade-offs in the design of financial algorithms

We investigate the trade-offs involved in encouraging the adoption of stock-trading algorithms. In a three-week artificial stock market experiment, investors experience trading both independently and with the help of a financial algorithm. They then decide whether to adopt the algorithm. Across treatments, we vary the algorithm by its trading strategy and whether its decisions can be overridden. Our findings show that adoption rates are generally low, but investors are more likely to adopt an algorithm that trades actively and that they can override. An investor’s trading style does not consistently affect algorithm take-up. Instead, adoption primarily depends on the relative success of a trader’s performance when trading independently vs when the algorithm was trading on their behalf. Analysis of an exit questionnaire matches those observations with the reasons given by individuals for rejecting or adopting a financial algorithm.