A possible solution for latency arbitrage
Latency arbitrage has been a big problem in the eyes of buy-side traders for years now. Such arbitrage refers to the sheer speed of high frequency traders' systems, which can react quickly to orders sent by traditional investors. Latency arbitrageurs, once they sense the order price in one venues, can essentially cancel bids and offers quickly in other venues and re-submit them at a price less favorable to traditional investors, essentially forcing them to buy at a slightly higher price or sell at a slightly lower price.
The issue was big enough that RBC was able to garner lots of mindshare for its Thor offering, which aims to combat such arbitrage on behalf of the traditional buy-side.
Latency arbitrage was the topic of a recent study by the University of Michigan engineering department, which concluded that "latency arbitrage not only reduces profits of the background traders, but also diminishes" efficiency surplus overall. It also found, somewhat surprisingly, that market fragmentation per se does not harm efficiency. In fact, "some degree of fragmentation mitigates inefficient trades that are often executed by a continuous mechanism."
The study also proposed a solution: a "discrete-time call market" that features short, finite clearing intervals. Aggregating orders over small, regular time intervals "prevents high frequency traders from gaining a latency advantage," thereby increasing efficiency for the entire market. These benefits appear to overshadow the gains attributable specifically to neutralizing latency arbitrage.
For now, it's unlikely that the SEC will opt for such a market, given the historical presence of a continuous market. To some, it would seem like a retrograde move. Buy-side traders will have to get by with Thor-like services. But this is certainly an issue that the regulators have to take seriously. In the end, nothing good can come of gaming the NBBO.