The Low Frequency Trading Arms Race: Machines Versus Delays by Alexander Dickerson, Yoshio Nozawa, Cesare Robotti :: SSRN

We propose a novel framework to compute transaction costs of trading strategies using infrequently traded assets. The method explicitly accounts for the trade-off between bid-ask spreads and execution delays. The benefit of waiting for a better trading opportunity with lower bid-ask spreads is partly offset by the opportunity cost of delayed or missed execution. Applying this method to corporate bonds that trade infrequently, we show that even the latest machine learning-based trading strategies earn zero or negative bond CAPM alphas after transaction costs. Consequently, our results raise doubts about the realistic outperformance capabilities of active bond trading strategies relative to the bond market factor.