Authors: Elena N. Asparouhova, Peter Bossaerts, Kristian Rotaru, Tingxuan Wang, Nitin Yadav, Wenhao Yang
Abstract: We present results from an experiment where participants have access to a set of robots (automated trading algorithms), which they may deploy, launch, halt and replace at will, while still trading manually. We hypothesize that mispricing would be reduced. Yet, we observe equally large and frequent mispricing and, in early trading, significantly higher effective bid-ask spreads, and flash crashes/price surges. Nevertheless, participants who engage in both robot and manual trading perform better. Inspection of the types of robots deployed reveals a potentially disturbing source of bias in traditional field studies of algorithmic trading.