Group Co.
Rational Arguments on Market Structure & Unintended Consequences
Traders Magazine Online News, July 31, 2017
In a segment on John Lothian News, Peter Maragos of Dash Financial calmly makes two important points: First, market data costs need examination, since it is illogical that, as technology costs per message have dropped, data costs have continued to go up, and, Second, calls for banning rebates are illogical since institutions, armed with proper information, can benefit from our market structure, which is the most liquid in the world. Before delving into the content, I want to point out how sad it is that Peter’s intelligent, common sense arguments fail to gain mainstream press coverage, while the hysterical bleatings of Brad Katsuyama and his paid minions become major news stories whenever they are repeated…
With that rant out of the way, let’s examine both points. First, market data, where Peter’s observation of lower cost to produce the data not making into user costs, is completely valid. Of course, basic economics teaches us that oligopoly power will result in this pathology, so we should not be surprised. His second point, that the SIPs must be improved to “fix” this problem deserves amplification, however. On this point, there is a not-subtle issue, namely that the geographic dispersion of the main data centers needs to be addressed to make the SIP competitive with direct feeds.
To explain this simply, consider a router, which resides in Secaucus, making a routing decision about a Nasdaq stock. It gets SIP data from one of the Bats exchanges, which also reside in Secaucus. Currently that data needs to traverse the 40 miles to Carteret to be processed by the tape C SIP and return the same distance. This means that even if the SIP was infinitely fast, the delay caused by the speed of light over the 80 miles renders it insufficient for routing purposes. (The same scenario exists for routers in all three data centers.) This effectively grants a monopoly to the direct feeds for routing users. Thus, the ONLY way the SIP can effectively compete is for there to be THREE SIP instances, one in each main data center. I should point out that, while he was not the first to say this, Michael Blaugrund of the NYSE (ICE) made this point publicly, at an industry event last year. That deserves recognition, as it showed both intellectual honesty and a true concern for improving the industry from a leader in the SRO community.
For more information on related topics, visit the following channels:
Comments (0)