Testing VPIN on Big Data – Response to 'Reflecting on the VPIN Dispute'

In their note titled “Reflecting on the VPIN Dispute,” Andersen and Bondarenko (AB) questioned the validity of using the false positive rate to measure the effectiveness of VPIN. However, their objection was based on an imperfect definition of the false positive rate, not taking into account the fact that we have addressed the key shortcoming of that definition. AB argued that VPIN was ineffective because it would produce too many events, but our report showed that the average number of events is quite modest on a large variety of data. They also raised a number of non-technical issues about our work. In the interest of maintaining the integrity of the research record, we would like to address their points