RT Journal Article
SR Electronic
T1 Queue-Jumping & Strategic Limit Order Routing
JF The Journal of Investing
FD Institutional Investor Journals
SP joi.2021.1.208
DO 10.3905/joi.2021.1.208
A1 Hitesh Mittal
A1 Kathryn S. Berkow
A1 Johnson Zachariah
YR 2021
UL https://pm-research.com/content/early/2021/10/18/joi.2021.1.208.abstract
AB Selecting the optimal venue for limit order placement depends on the specific exchanges competing at the NBBO at the time of order placement. The speed of a fill at NASDAQ depends not only on NASDAQ’s market share, but whether orders at the same price are also posted at IEX or EDGX, for example, or some other combination of the fifteen other exchanges—and 65,535 possible combinations of exchanges that could be competing simultaneously at the NBBO. This paper outlines specific examples of “queue-jumping” and describes quantitative methods that can be used to identify the exchanges with the highest likelihood of filling limit orders in real time. A ranking system called ELO used to rank chess players and online video gamers is applied; rankings are assigned to exchanges and used to determine which exchange is most likely to receive the next marketable order for any of the possible combinations in competition. Exchanges are ranked under varying conditions and the rankings are used to address the problem of queue-jumping.Key Findings▪ Execution algorithms’ limit order routing infrastructure is often based on heuristics rather than empirical analysis of competitive passive fill rate and speed at each venue. This inefficiency and market fragmentation create opportunities for “queue-jumping”—execution of limit orders before those placed at an earlier time. Queue-jumping strategies can be used to improve algorithms’ passive fill rate and/or speed. ▪ The ELO method developed for ranking chess players is applied to US equity exchanges to aid in making decisions about where to post limit orders to best jump the queue and increase an algorithm’s passive fill rate and/or speed. ELO allows for a simple list of exchange rankings per stock but is not suitable for aggregation across stocks. ▪ A conditional fill rate matrix (the “WinMatrix”) is also applied for ranking exchanges, providing a relative likelihood of fill given that a particular exchange has a limit order present at the National Best Bid (NBB) or National Best Offer (NBO). The WinMatrix can be created for individual stocks or groups of stocks.