Short sales and trade classification algorithms

Abstract

This paper demonstrates that short sales are often misclassified as buyer-initiated by the Lee-Ready and other commonly used trade classification algorithms. This result is due in part to regulations which require short sales be executed on an uptick or zero-uptick. In addition, while the literature considers "immediacy premiums" in determining trade direction, it ignores the often larger borrowing premiums which short sellers must pay. Since short sales constitute approximately 30% of all trade volume on U.S. exchanges, these results are important to the empirical market microstructure literature as well as to measures that rely upon trade classification, such as the probability of informed trading (PIN) metric.

Trade classification accuracy for the BIST

“…Like these authors, we find that long sellers appear to consume liquidity more often than they provide it while short sellers appear to provide liquidity more often than they consume it. Like these studies, we find that short sales are predominantly buyer-initiated which is consistent with the expectation advanced by Chakrabarty et al (2012) but not Asquith et al (2010) whose prior is that short sales should be predominantly seller-initiated. This interpretation depends on the somewhat strong implicit assumption that any differential immediacy costs associated with the short-side of a trade are not material.…”

“…We find significant differences in the accuracy rates of the trade classification algorithms that generally are higher for long versus short trades. Unlike Asquith et al (2010) and Chakrabarty et al (2012) who examine US data for June and December 2005 provided under the SEC's Reg SHO initiative, 5 we find accuracy rates of at least 90% using one-second lagged quotes for both long and short trades for the quote, at-the-quote and LR algorithms. Like these authors, we find that long sellers appear to consume liquidity more often than they provide it while short sellers appear to provide liquidity more often than they consume it.…”

“…The test period is also centered on the month of the collapse of Lehman Brothers (namely, September 2008) to provide a first test of the accuracy of the five trade classification algorithms during a period including the effects of a financial crisis and the Great Recession. 2 Our choice of time period extends the work of Asquith et al (2010) and Chakrabarty et al (2012) who use two months in 2005 that the former authors (page 472) characterize as: "These two months also offer the advantage of capturing a recent, relatively "normal" time in the markets -after decimalization, but before the financial crisis and the scrutiny of short sellers that followed . .…”

“…This interpretation depends on the somewhat strong implicit assumption that any differential immediacy costs associated with the short-side of a trade are not material. To further illustrate the importance of this implicit assumption, we note that Asquith et al (2010) report that the proportion of buyer-initiated trades involving short sellers is 2 Turkey and the following countries/territories went into economic recession in the third quarter of 2008: Euro-using nations in the European Union as a whole, Germany, Hong Kong, Japan, Italy, Singapore, Sweden, and the United Kingdom. They were followed into technical recession in the fourth quarter of 2008 by Spain, Switzerland, Taiwan and the United States.…”

“…Asquith et al (2010) find that the tick, quote and LR algorithms often misclassify short sales as buyerinitiated based on a sample of 100 stocks from each of NASDAQ and NYSE during the three months of March, June and December 2005. The LR algorithm performs the best with a 33.4% accuracy rate for short sales when measured against a prior that short sales should be seller-initiated.…”