Back to Portfolio for the Future

Metamorphosis in HFT: Update with Brad Katsuyama of IEX, Central Figure in 'Flash Boys'

September 18, 2014

Doug_bradAs Gregor Samsa awoke one morning from uneasy dreams he found himself transformed in his bed into a gigantic insect.  "The Metamorphosis”, by Franz Kafka.

Brad Katsuyama awoke one morning in early spring, 2014, to find he had just become the object of greatly mixed opinions and emotions upon publication of Michael Lewis’s work Flash Boys.   Michael Lewis had reached out to Brad earlier in the preparation of the book, but as the book took shape, and Michael spent more time at Brad's offices, Brad began to suspect that he might be featured rather more than he first thought.

When publication bore out that notion, the floodgates opened.  Brad Katsuyama did seem rather like a pesky cockroach who shouldn’t be crawling through the hors d’oeuvres to the HFT firms who had financial interests to protect.  On the other hand, there was a growing chorus of institutional investors who, for the first time, could see unnecessary costs generated by HFT operations which apparently served no useful purpose.

We asked Brad what he had learned about his fellow man from the experience of being a lightning rod for the entire discussion. Brad had thought going in that there would be more of a sense of outrage over matters of right and wrong versus a relatively bloodless discussion about who was entitled to the money that HFT funds might be siphoning off.

Brad said:  “I thought arguments would be made more at a higher macro level, rather than a complete deflection away from the broader issues of propriety into minutia of specific market structure details.  Brad felt that HFT apologists were  “attempting to divert attention away from big thematic questions, by talking about single counting, double counting, or broker priority – a host of side issues which could be seen as curveballs designed to confuse the everyday person trying to get the kernel of truth about HFT."   Brad went on to note the volume of argument cloaked by anonymity, with “people hiding in the bushes and throwing rocks at us. But general public response has been very positive, and many people inside the system also believe that changes on Wall Street are badly needed."    But many who endorsed his and IEX’s value to investors oft preferred to share their views in private, usually for reasons having to do with their current employment.

Brad overcame his closet Canadian tendencies to concede that on Wall Street, “People respond to incentives.  The right decision (for them) is the one that fulfills their self interests.”   Since the publication of Flash Boys, IEX has received many resumes of highly qualified people  for posts earning less at IEX than at their competitors.   IEX volume has tripled since Flash Boys was released. As Brad observed, “It’s been a whirlwind.”

Who’s Zooming Whom?

We asked Brad about a criticism leveled at Flash Boys  on Zerohedge by Christopher Whalen, who wroteIndeed, the real scandal is that all of this has been entirely blessed by the SEC, FINRA and the major exchanges and is described in the voluminous public documentation for permitted order types.  But suffice to say, virtually nobody in the Big Media or at most Wall Street firms understand any of this or knows, for example, that there are over 100 different order types allowed by the SEC and FINRA under current law and regulations.

Brad couldn’t comment on Michael Lewis’s (or others’) editorial decisions.  He did say in response, “If I were sitting at the SEC, I wouldn’t have been able to figure it out.”  He further noted that regulators are in a tough spot because they rely on market participants to inform and explain to them what’s going on.  Which could mean that the foxes are running the chicken coop indirectly.

The intellectual mismatch between Wall Street and government regulators isn’t particularly new.  Nor should it be surprising when guys making six and seven figures to play the game win against a team lucky to make five figures for the most part, with so little critical experience that they have to rely on the regulated to advise how and what to regulate.  The only real question in this scenario is why on Earth HFT firms would object to government regulation when they have so much input.  One might also wonder why non-HFT investors, once they begin to suspect that they are disadvantaged by the rules, don’t begin to demand a seat at the regulatory table, so as to combat the gaming of the system.

Students of Wall Street history will know that this sort of thing happens a bit.  One such example was the Drexel Burnham junk bond episode, where well-paid salesmen helped to market toxic, illiquid paper to smaller institutional customers in the American outback who were grateful for the slightest attention paid by their good looking Wall Street contact with shiny hair and shoes, enhanced by generous expense accounts.

The HFT Apologists’ Narrative in Its Simplest Terms

The marketplace needs High Frequency Trading because it provides badly needed liquidity to institutional and individual investors.  It's sort of like Soylent Green for investors, with apologies to an interesting, useful new food company named Soylent, which, they hasten to point out, isn't people.

HFT:  the Emperor’s New Liquidity?

We began our conversation with Brad Katsuyama with the objective of getting to the heart of the HFT story in a manner that could be understood by people who don’t do algorithms.  To torture a paraphrase of Will Rogers:  we never met an algorithm where we could tell whether we liked it or not.  Many readers may share this point of view.  The sad truth is that there are going to be few outside the HFT world who can figure out all the connections.  Even the savants who create the algorithms may not themselves always understand the uses to which their work is being put.  It’s rarely their job to know.

When humans are involved in something, they will often leave fingerprints that show intent.  With careful inspection, one may be able to find elements in the HFT story that show the intent of the practitioners.  Revealing intent provides clarity on this question:  is the purpose of HFT to provide liquidity or to provide a generous revenue stream to its practitioners while providing the appearance of liquidity?  Or maybe some of both?

Fingerprint #1

One indicator of a flaw in the system was mentioned in this 2011 article:

“It is by now common knowledge that High Frequency Trading comprises 70 to 75% of market trading in North America and western Europe.  The primary usefulness of algorithmic trading is to provide liquidity to the investor class….It is comforting to know that 70% of the marketplace activity is there just so that you, part of the 30% investor class, can sell your shares a teensy bit better on the day you decide to go to the souk.   Still, residents of the virtual world we will call Algorithmia do seem to be providing liquidity mostly to each other and making money doing so.  Which would make one wonder, do Algorithmians need Investors more than Investors need Algorithmians?”

That question was answered somewhat by Manoj Narang of Tradeworx, an HFT firm.  He rather disagreed with Brad Katsuyama during a 2014 Bloomberg interview.  In a 2011 interview, Manoj Narang had this this to say: “High-frequency traders can’t generate volume in a vacuum.  If you put a bunch of HFTs on an exchange together, most likely there will be zero volume.  You need large investors seeking liquidity to enter the market – this is what creates all kinds of opportunities, based on correlation, in all kinds of stocks.  HFTs need significant participation by long-term investors.”

Brad weighed in on the subject thusly:  “People say the market has never been more efficient, based on the premise that HFT is there to correct pricing or structural inefficiencies.  How do you call the market efficient if they are the largest participant by volume?  That IS an inefficient market.”  We asked if there might be a percentage of HFT volume as part of total trading that would suggest a clean and efficient role on the part of HFT firms.  Brad thought intuitively it might be around 25%, but admitted that an exact number is really hard to determine. It’s actually easier for him to say that “the right number is definitely not half of all trading volume”.

We observed that investors have been taking positions forever, without HFT.  Brad responded, “HFT can be used to make the market more efficient.  And is.  However, it is quite possible for some HFT firms to make the market more efficient with 20% of their trading and point to that as justification, while in background, another 40% of trading makes the market less efficient, producing a net result of reduced market efficiency.  And how would the typical investor know?”

We wondered whether HFT firms could be following a time-honored supermarket strategy of having loss leaders or less profitable products to get shoppers into the store whereupon shoppers would also buy the higher margin products.  Brad thought it likely.

Brad told us about a misconception about HFT which was found lingering in the halls of Congress when Brad appeared before the Senate to discuss HFT.  One Senator cited the decline in investor trading costs through online brokers from $50 per trade to $8 per trade.  This struck him as proof positive of the value of High Frequency Trading.  While he had learned his talking point well, he was less well equipped to understand that the shrinkage in brokerage commissions was related to totally different online improvements.  It seemed to us as we listened that the facts rolled off the Senator like water off a duck’s back, although Brad believes that the Senator was within shouting distance of the truth of the matter after some twenty minutes of dialogue during which Brad used simple words and spoke slowly.

Fingerprint #2

A second fingerprint exists, it appears, in the myriad of order types that HFT firms typically place.  This is up from the half dozen or so types of orders available decades earlier.

From Flash Boys:  “The old order types were simple and straightforward and mainly sensible.  The new order types…maybe 150 of them… that accompanied the explosion of high-frequency trading were nothing like them, either in detail or spirit...A Hide Not Slide order---one of maybe fifty such [types]---was a way for a high-frequency trader to cut in line, ahead of the people who’d created the line in the first place, and take the kickbacks paid to whoever happened to be at the front of the line…The many (unusual) order types...had one thing in common:  They were designed to create an edge for HFT at the expense of investors….Their purpose was to hardwire into the exchange’s brain the interests of high-frequency traders—at the expense of everyone who wasn’t…That is why, though they made only half of all trades in the US stock market, they submitted more than 99% of the orders, (which) were a tool for divining information about ordinary investors.”

We had heard something similar as we spoke to Brad, but mercifully, we’ve been relieved of having to sort through and explain the intricacies of how these orders worked, probably to the reader’s relief as well.  Brad observed that the ultimate object of these orders was to put the HFT firm at the front of the queue and on the bid side so that it could act as a ‘maker’,  a passive buyer even though it was anything but.  Some order types appeared to work magic in arranging for the HFT firm to occupy that position even though it might not actually have shown up first in the normal way.

One order type worthy of mention, close kin of HNS in Brad’s view, was called, poetically, Partial Post Limit with Maximum Removal Percentage.  The reader can let his imagination run wild as to what it might actually mean, how it works, and more importantly, how much longer it took to say the order than to execute same. And whether they came up with a word like pipplemurp, which is the closest pronunciation of the obvious acronym.

Makers and Takers

As regards HFT, makers are those who provide, or make, liquidity and takers are those who absorb, or take, liquidity.  In other words, if you’ve placed a limit order that gets taken by somebody who’s gotta have it, you’re the maker and your counterparty is the taker.  To the degree that HFT orders have been designed to allow HFT firms to position themselves to be makers at others’ expense, there’s an argument that the system has some inherent flaws. That part of the conversation went this way:

Q:  What’s smallest number of shares I can buy to get myself FIRST on the bid side?

BK:   The objective is to get on the bid side to get the rebate.  At most markets 100 shares is the minimum lot to be in the official queue.

Q:  Are there stats that show trading volume by order type?  The order has to be marked, doesn't it?

BK:  To my knowledge there isn’t a database of volume by order type. Also, that database would only count orders that receive an execution, not orders that don’t. This is the big fight:   in the industry right now.  If I send 1000 orders but none are executed, there would be no record of me using that order type.

Q:  So the industry wants no one to know about these “spy” orders whose purpose is to position the HFT firm to position itself as a “maker” to capture the order and get paid a commission for their “passive” execution?

BK:  From our research we have noticed that many firms place small 100 share “maker” orders in the market to get information about what is happening in the market. But once one of these orders receives a fill – that same “maker” of liquidity, may immediately turn around and start taking liquidity now that it has the signal.   So the “maker” immediately becomes the “taker”.

Q – Why are there so many order types?

BK – Many complicated order types were designed to only trade in very specific conditions, most of these conditions are not that relevant to traditional investors that are looking to increase or decrease holdings in companies. For example, let’s say the order sender wants a rebate.  But he also wants to be first in line to ensure that he increases his chance of getting the rebate.   He could use an order type that gives him a higher probability of achieving that outcome versus other people who aren’t using that order type (many of whom don’t even know that the order type exists).   His rationale: I only want to do it if it suits me, which the exchange and market conditions will determine.

Q:  It sounds like the real story isn’t in the details; it’s the creation of an intentional cloud of dust so that almost no one can figure it out and see the purpose of all the activity.

BK:  That’s the world we occupy.  We live in a dust cloud.  Although the exchanges are meant to be neutral, this can blur the lines of neutrality.

Q:  One has to ask:  when an HFT firm SAYS it’s providing liquidity, is it really?

BK:  I think some are - but it's hard to tell across all HFT firms.  The technological value add in smartphones or medical products is clear. The common man’s understanding of the technological value add  on Wall Street boils down to $8 trades down from $50 per, due to HFT, as the proof that all the technology is beneficial.

Other voices provide similar insight.  In Zerohedge, Haim Bodek of Decimus Capital revealed a number of order types some of which he enumerated with telling nomenclature:  queue jumping; cannon fodder; fish food; bait & switch; catch me if you can.

Maker/taker terminology also has a political currency, where makers are said to have the moral high ground over takers, who are presumed to be absorbing the efforts of the makers whether they are or not.  If a taker games the rules to be able to appear as a maker, profiting with no discernible value-add to the client, the moral high ground becomes rather fuzzier, perhaps.  That's what is alleged to be happening in the HFT world.   Long ago, we lived in New York City in an apartment billed as having "old world charm".  We learned how cockroaches behaved when we went into the kitchen in the middle of the night and turned on the lights.  They scurried for cover and hid as quickly as they could.  While it may be difficult to determine who the roaches are in the HFT story, if any, it seems likely that aversion to disclosure will be one of the best indicators of roachlike behavior.

And it leaves us, finally, with this maker-taker question:  if a taker has the lobbying clout to game the legal system, pretend to be a maker, and accumulate significant wealth as a toll-taker on a bridge to nowhere, is he entitled to the same moral high ground as the business owner who provides demonstrable added value to his customers?

We didn't ask Brad Katsuyama this question.  We can guess at his answer.  But then, he's Canadian, and is charmingly naive.