After an earlier post of mine here about high-frequency traders and the Securities Information Processor, I received a tweet from @LiquidMkt. Actually, I received two.
LiquidMkt, a blogger who believes that HF traders are on the receiving end of inaccurate charges,credits them with valuable work providing the markets with liquidity. The first of his tweets to me said, “Read your blog, but it leaves out some relevant info. Have you checked this out?” and then linked to this discussion of the legality of direct feeds.
I have never contended that direct feeds are illegal, so I didn’t really know what to do with that. Another tweet soon came along, asking me, @ccfaille, “When u say #HFT front run the SIP, how do u mean? Front-running as it is commonly understood is a myth.” There followed another link to the same discussion of the legality of direct feeds.
But let’s focus on the question. What do I mean by “front run”? I use the term as suggested by the metaphor of running-in-front. It refers to a variety of situations, in which any party trades on the basis of advance knowledge, that is, of material non-public information, of an upcoming trade. Click here for a recent post on the subject from the blog of the Columbia Law School.
Fiduciaries or Not
The term initially developed in the fiduciary context for situations such as this: if you ask me on Monday to help you buy a large portion of the world’s aluminum, and I agree, but then before I execute your order I buy some aluminum myself, on Tuesday, I may profit from my expectation that your offer will then drive up the price of aluminum when I finally get around to executing it, on Wednesday.
Obviously there is nothing different in principle if this happens over three days than if it happens over 3 milliseconds or less.
Consider that aluminum/fiduciary context: who is harmed? Well … you are, to start. Perhaps only slightly or hypothetically (depending on how greedy I am). But my own purchase on Tuesday, even if a small one, will in principle make aluminum more valuable on Wednesday than it would otherwise have been, increasing the costs of your trade. Others are also harmed. For example, if you are buying aluminum because you have an operational need for aluminum – perhaps you are a large manufacturer of soda cans – then front running imposes costs that are passed along to people buying those cans, that is, to the public at large.
But in circumstances in which HFT is employed to do something analogous, there is not typically a fiduciary relationship. HFT’s aren’t getting the crucial non-public information from customer/clients. There are getting the information as a consequence of ping orders. This is precisely why such HFT activity is often referred to (by Charlie Munger for example) as “legalized front running.” It is acknowledged to be legalized because the HFT firms can do it without violating a confidence. Observers like Munger acknowledge this legality in a disgruntled sort of way, but they do acknowledge it.
It is with all this in mind that I sent a reply tweet: “’preciate your legal point in link, but it leaves me cold. ‘Front running’ describes events, not legal categories for them.”
Closing the Window
Under standard finance theory, the existence of arbitrage opportunities in general suggests inefficiency. This is true for latency arb as for any other sort of arb. Regular readers will know my views on some of the sources for the inefficiencies at issue.
In the paradigmatic situation, though, the arbitrageurs close the windows that they exploit. That is their social utility. If an aluminum company’s stock trades on a Pittsburgh exchange and on a Philadelphia exchange, and if there is a significant difference in its value from one end of Pennsylvania to the other, then there is inefficiency somewhere. Trans-Pennsylvania arbs will eliminate that inefficiency, buying the stock on the one exchange and selling it on the other. And voila! They make some money in the process.
But this does not seem to be the case with latency arbitrage. With a welter of supportive regulations, from the 1990s shrinkage in the size of minimum ticks to and through Reg NMS, the inefficiencies in question are not obviously self-resolving, at least not for so long as the expensive arm’s race in the associated tech continues. This legal front running preserves rather than terminates the inefficiency that has given rise to it, and that is a systemic concern for the markets.
Thus, my second response to @liquidmkt took the form of two questions of my own. I asked/tweeted: “Are you agreeing with the contention that nobody uses SIP anyway? Are you contending ‘latency arb’ doesn’t exist?”
No answer yet.