Feature Stories | Apr 09 2014
By Greg Peel
Prologue
Last week US company Virtu Financial decided to postpone its scheduled IPO. The postponement was a direct response to the furore created by the release of acclaimed author Michael Lewis’ new book, Flash Boys, which declares the US stock market to be “rigged” because of high frequency trading. A claim by Virtu, an HFT specialist, that the company made money in 99.9% of transactions it engaged in, only fuelled the anger Lewis’ accusation had whipped up.
I personally have great respect for Lewis as an intelligent and knowledgeable author on the subject of the truth behind financial markets. From his 1980s expose of the excesses of Salomon Bros (Liar’s Poker) to his more recent disclosure of the realities of the Global Financial Crisis (The Big Short), Lewis has proven he knows what he’s talking about and can relate that to the layman in an informative and entertaining fashion. I am yet to get my hands on a copy of Flash Boys but I have seen the US 60 Minutes interview in which the author makes his “rigged” accusation. I am sure his book will make a compelling case but I do believe he has chosen the wrong word in “rigged”.
“Rigged” implies illegal, fraudulent or underhand. HFT is none of these.
It is typical in the wake of a major stock market crash for wounded investors to look for scapegoats. In 1987 it was “program trading” which highlighted to the world that computers were trading on their own volition. The computer debate reached a new peak following the 2008 GFC when high frequency trading was singled out as a scapegoat.
In 2009 I was commissioned to write a speech on subject of HFT for a colleague to deliver at a symposium in Singapore. At the time, protest against HFT was rife. It flared up again in 2011, when I was granted permission to republish that speech on the FNArena website.
The speech concludes with the line, “HFT is here to stay – get used to it”. Five years later, I’m yet to be proven wrong. But that is not to say I don’t have any problem with HFT or that I am a vocal supporter of HFT. I do believe that HFT is just one of the latest elements of a natural progression in the history of financial markets that sees technological advance running well ahead of regulation. But I also suggest HFT is a response to a US stock market system that is flawed. Regulation, or lack thereof, allowed that system to become flawed. HFT simply takes advantage of that system, legally.
Here follows the text of that speech. Bear in mind it was written five years ago, and that in that time no regulation has been forthcoming and HFT has exploded. Following the text, I will provide my epilogue in which I will discuss the arguments being put forward in 2014 as to why HFT indeed needs to be addressed, even if it is not accurate to conclude that the market is “rigged”.
The Speech
High Frequency Trading
The Global Financial Crisis of 2008 is the most recent financial market “bust” in the long line of booms and busts that dot history, including the Crash of ’29, the Crash of ’87 and the Tech-Wreck of 2000.
After every such crisis, fingers are pointed, accusations are made, politicians vow revenge, and the media has a field day. Such a response is simply part of human nature – the same human nature that determines the following trading tenet:
If I make money it is because of my skill. If I lose money it must be someone else’s fault.
Which brings me to the true topic of today’s presentation:
High Frequency Trading – Every Crisis Needs its Scapegoats
Let me start with a brief historical tale.
In the late eighties, the Australian Stock Exchange was in a transitional phase which was to see the old open out-cry and chalk board system gradually replaced by an automated trading system.
The Australian stock trading model never included “specialist” market makers as had always been the case on the New York Stock Exchange. However the accompanying exchange-traded option market relied almost entirely on specialist market makers. The ETO market was also open-outcry in the late eighties, and whenever a specialist quoted a price spread on an option, he would do so based on the prevailing price spread on the underlying stock as displayed on a screen.
If the specialist was hit on a price, he would then run across the floor to the chalk board, find his designated operator, and attempt to cover his delta hedge. If the stock in question was one of those already automated, he would instead run into his designated booth, pick up the phone, ring his computer operator, and attempt to quickly snatch that bid or offer upon which his option price was originally based.
Frustration would result if the phone at either end was already busy.
If several options market-makers were simultaneously hit at the same price, a scramble ensued. The market-maker who was quickest to the phone would get the best stock price on his hedge. The laggards would have to settle for something less.
Macquarie Bank –with whom I’m sure you’re all familiar – operated the largest team of option specialists. While having the biggest team was an advantage, it was also a disadvantage. Like all specialists, Macquarie’s team was only allowed one phone line out of the exchange. This often resulted in Macquarie specialists also racing each other to transact delta hedge trades.
(Remember that this was long before the advent of the mobile phone.)
Macquarie solved the problem, however, by giving each of its specialists, and the computer operator, a walkie-talkie. Macquarie overcame reception problems by running a cable from a transmitter at the exchange to a receiver at its trading desk. As luck would have it, the Australian Stock Exchange and Macquarie Bank shared the same building.
From that point, the Macquarie specialists could immediately place stock orders from where they were standing. From that point, Macquarie specialists were the first to every delta hedge trade.
Macquarie’s competitors, understandably, immediately cried foul, and petitioned the exchange to ban the walkie-talkies. The Exchange took a quick look at its rule book and noted every rule pertaining to fair and equal communication access specifically referred to telephones. There were simply no rules against walkie-talkies.
The walkie-talkies were thus deemed permissible.
For a brief period, Macquarie ruled the options market. But eventually, every specialist had his own walkie-talkie. The playing field was once again even.
I relate this anecdote today because the determining elements within it are “speed”, “technology” and “advantage”.
Let me now jump forward to the present day, and highlight this recent quote:
“Powerful computers enable high-frequency traders to transmit millions of orders at lightning speed and, their detractors contend, reap billions at everyone else’s expense.” (New York Times, July 2009)
Since the “sub-prime credit crisis” of 2007 became the “Global Financial Crisis” of 2008, many fingers have been pointed, accusations made, and scapegoats sought. The usual suspects have been trundled out, including greed, fraud, derivatives, leverage, hedge funds and short-selling.
But when the angry mob is frustrated that no progress is being made in lynching one particular culprit, it will quickly move on to another. As always, ignorance is the dominant force. That is particularly the case when the culprit is deemed to be a computer.
Computer-driven high-frequency trading has become the most recent target of critics, the media, politicians and regulators. HFT now appears on all the fresh “Wanted” posters.
One might be forgiven for thinking HFT is a new concept that was created just in time to bring down Lehman Brothers. But HFT has been with us for many years in one shape or form. Indeed, HFT may have given philosopher Plato another reason to suggest that “necessity is the mother of invention”, scientist Charles Darwin further proof of “natural selection”, and economist Joseph Schumpeter further evidence that “money will always pass from weak hands to strong”.
HFT is not just some passing fad.
“In recent years, HFTs have been the single greatest influence on many aspects of market structure globally, including exchange pricing, new order types, competition among market centres and the technology that handles everything from matching trades to spitting out market data to customers.”
(Rosenblatt Securities, September 2009)
And to maintain the historical theme, perhaps King Canute might suggest at this point that “you can’t hold back the tide”.
HFT is the subject of various criticisms, all of which lead back to the simple concept of “unfair advantage”. To assess whether accusations of “unfair advantage” are in themselves fair, one must first appreciate the evolution of HFT.
Let’s go back in time again, back to when stocks were traded by open out-cry at designated exchanges, and orders could only be placed with stockbrokers by telephone.
Various exchanges across the globe adopted differing trading models. On the New York Stock Exchange, specialist market-makers were licensed to ensure a level of liquidity and a constant bid/ask spread. Specialists were compensated for the risk they assumed by being granted the “first look” at all orders.
On Australian exchanges, only options markets boasted market-makers. Stocks traded only when broker-operators matched orders with their peers. At varying times, some stocks were liquid, and some were not.
Whether or not an exchange adopted a market-maker model, the successful trading of any market – stocks, futures, options, forex, interest rates etc – required a level of discretion, intelligence, experience, talent and overall “street-smarts” in order to be consistently profitable. And just as if it were a poker game, good traders never immediately “showed their hand”.
If an operator needed to fill a big buy-order, for example, which he knew would move the price, he might sell a few contracts first to confuse his rivals.
Sell a few before buying many.
Perhaps he might buy only a few contracts first, hoping he might unearth a big seller who could fill his order quickly.
Buy a few to see who’s selling in volume.
Perhaps he might have been instructed to test the water first, and if there appears to be no real sellers, cancel the buying.
Fill or kill.
Or perhaps he might decide the best way to achieve a good price is to work the order over the course of the session and be happy with the ultimate volume-weighted average price.
Work a VWAP.
These are just a few examples of what might be called “tactics” or “strategies” used by traders since around about…um…the dawn of time. Funnily enough, no one in the market on the day – not the rival operators, nor the clients, nor the exchange officials, nor the regulators, nor the media, nor politicians – complained about such practices. Indeed, an operator who could achieve the best price on an order – by any stealthy, but not illegal, means – was much admired.
He might be considered a “gun trader”. And the best operators attracted the most business for their firms.
Such operator strategies were even more important when, as was often the case in the stock market, the contract being traded was somewhat illiquid. In the US, illiquidity in stocks was resolved by the specialist market-maker system, although dealing with a market-maker often meant crossing a wide bid/ask spread.
Given the history of computers, it is surprising to appreciate that the first computer-based, electronic stock market was established as early as 1971, when the US National Association of Securities Dealers created a computer bulletin board where buyers and sellers could publicly post their bids and offers.
Importantly, this initially over-the-counter market served to reduce the spreads on listed stocks. This pleased the buy-side, who could buy or sell at a better price, but not the sell-side, which relied on spreads to make profits.
But the buy-side won, and from this early market was born the National Association of Securities Dealers Automated Quotient, or NASDAQ, stock exchange.
1971 Nasdaq created.
On the New York Stock Exchange meanwhile, the broker-dealers continued to keep their orders a secret from the market, thus exploiting the advantage of exclusive information in return for offering a spread. But in 1997, the rules were changed, and the broker-dealers were forced to disclose their limit orders. The immediate effect was that bid/ask spreads became tighter.
1997 Disclosed orders tighten NYSE spreads
The new rules, along with the exponential growth of computer power by this time, gave rise in 1998 to Electronic Communications Networks, which were off-market facilitators of trade matching, and which evolved with the blessing of the US Securities Exchange Commission.
Importantly, in order to attract business the ECN’s re-engineered the old model that every trade attracts a fee and a commission, instead offering rebates to anyone placing an order and funding the rebates by charging more for anyone hitting an order. The concept of liquidity rebates was thus born.
1998 ECNs offer rebates for liquidity
The development of greater computer power at this time also offered a simple means of dealing with those fiddly smaller orders. Nasdaq created the Small Order Execution System, allowing orders of 500 shares or less to be routed directly to market-makers for automatic execution.
1998 Automated execution for small parcels
Which brings us to arguably the most fundamental development in the evolution of high-frequency trading. When small order execution was automated, wily traders with fast computers realized they could pick off market-maker prices in small parcels faster than the systems used by market makers would adjust their prices. The faster the computer, the more incremental profits were available.
The ultimate response was that Nasdaq shut down its Small Order Execution System in 2001. All this meant was disgruntled traders shifted their business to the Electronic Communications Networks, which by now were building even faster order-matching systems.
2001 Fast computer traders hook up with fast off-market exchanges
Arguably the second most fundamental development in the evolution of HFT occurred one year later.
From its early beginnings, stock trading in the US had always been executed on increments of one eighth of a dollar, meaning the tightest possible spread was a full 12.5 cents. In the late nineties, this tradition was under immense pressure from liquidity providers who would happily offer prices inside the spread if allowed. And so in 2002, decimalization was introduced.
Suddenly the tightest possible spread became a mere one penny.
2002 Decimalization closes the spread
At this point the US was suffering in the aftermath of the tech-wreck and 9/11. Many hotshot traders had blown themselves away chasing the tech bubble. But computer-trading of stocks on America’s plethora of competing on-market exchanges and off-market automated matching services continued to grow, as did the speed of the computers themselves.
Now that market-makers had lost the risk/reward safety of their spreads, speed – not risk tolerance – became the successful traders most important tool. Using speed and small parcels, traders could break down their risk into tiny increments and thus build consistent profits over the course of a session, without needing to take much of a punt on actual market direction to do so.
But there remained one more hurdle. America’s biggest stock exchange – the NYSE – remained stoically open out-cry.
The final domino fell in 2007 when the SEC moved to eliminate price mismatches across the various exchanges and trading platforms by introducing rules to ensure buy-side orders must receive the best available price. On the introduction of this rule, the NYSE was forced to join the electronic age.
2007 The NYSE goes simultaneously electronic
By now, computers were fast enough to place, execute, and cancel orders within milliseconds – far faster than the human brain could ever respond. With the biggest exchange now ripe for the picking, high-frequency trading exploded.
But if computers could act faster than humans, then computers needed to armed with the trading tactics and strategies which humans had developed in order to profit from trading. Over the course of this same time line, another evolution was occurring.
It was the evolution of algorithmic programs, or “algos”, which instructed the computers how and when to place and cancel orders. These algos have been heavily criticized by those on the buy-side for using supposedly “predatory” tactics and strategies which disadvantage and indeed compromise honest orders.
And what might some of these tactics be?
Sell a few before buying many.
Buy a few to see who’s selling in volume.
Fill or kill.
Work a VWAP.
And there you have it.
In the stock and other financial markets of 2009, high-frequency trading using super-fast computers and clever software programs are estimated to account for up to 70% of any day’s turnover in US stocks.
The geeks have inherited the earth.
Let’s now revisit our earlier quote.
“Powerful computers enable high-frequency traders to transmit millions of orders at lightning speed and, their detractors contend, reap billions at everyone else’s expense.” (New York Times, July 2009)
It is true that HFT now dominates turnover – a feature of 2008-09 that not only relates to the growth of HFT trading, but to the plunge in natural interest over the course of the GFC.
HFT has meant that millions of orders are now generated at millisecond speed – and apparently now at even microsecond speed – the bulk of which are then just as quickly cancelled. The remaining turnover generated has maintained overall volumes but significantly reduced the average size of each trade. HFT reduces risk by trading only incrementally, which is why HFT has continued to thrive through the riskiest period most of us has ever experienced.
And the more liquidity HFT traders provide, the greater the rebate they receive.
To those attempting to lay blame as to why they lost so much money over the past two years, HFT is an easy target. Those ignorant of the evolution of HFT can only assume that evil computers are instructed by evil geeks to trade on evil algorithms designed to rape and pillage the unsuspecting, honest investor.
This, of course, is rubbish. As our little trip through history shows, nothing much has actually changed at all. Stocks and other financial instruments are traded the same way they always have been – just a lot faster.
Yes – the computer has replaced the human for execution purposes. Geeks have indeed found a purpose in life, and a profitable one, but realistically computers are no better than the humans who program them in the first place, and resulting software is no better than the humans who have been trading by stealth since the dawn of time.
While HFT may be a way for traders to diffuse risk, at the end of the day the risk always remains. Just because computer programs are fast does not mean every trade must be a winner. At the end of the day, all stock market investment will generate winners and losers, just as it has throughout time. Even liquidity rebates are not sufficient compensation for being forced to cross a spread.
The ignorant assume that the remaining 30% of stock market turnover not attributable to HFT must, by definition, be the losers. Computers are hunting in packs, they assume, picking off the slow movers in the herd.
This, they suggest, is not fair. But what has realistically ever changed? Traders have been looking to exploit price mismatches for profit since the beginning of time.
Moreover, winners and losers in the stock market cannot ever be determined at the end of one day’s trade. While a day-trader may have profited on a quick turnaround, the investor with a longer time horizon is unconcerned that his one, two, or maybe five-year investment meant paying up a couple of extra pennies.
Winners and losers are only determined over time, and time will always roll on.
Furthermore, the fact that HFT turnover volume exceeds 50% suggests a mathematical fact – there must equivalently be winners and losers among the HFT traders. Thus the concept of “reaping billions at everyone else’s expense” cannot, by definition, be true.
It is true, nevertheless, that HFT traders have been reaping the spoils of their technological innovation over several years. But just as the Macquarie options traders of the late eighties enjoyed only a brief window of dominance when they alone introduced walkie-talkies, so too will the first-mover advantage of HFT be lost once the practice becomes generic, if it hasn’t already.
There are those that contend that HFT has no place in the market. But one must ask the question: if US stock volumes are no greater today than in recent times, and HFT is accounting for up to 70% of the volume, what would liquidity be like in 2009 if not for HFT?
HFT provides massive liquidity
Should computers be allowed to dominate liquidity, thus clouding what we might call the “natural” market?
Well consider that for decades, 90% of all daily foreign exchange positions are closed out at the end of a session. In other words, only 10% of turnover represents the “natural” market.
Consider further that for decades an average of only 3% of all deliverable commodity futures contracts ever result in actual delivery. This means 97% of all futures trades do not represent natural buyers and sellers.
And finally, note that on any given day, turnover volume in the Comex gold futures contract can exceed the entire volume of physical gold ever to be extracted from the earth in the course of history.
Yet we don’t here too many complaints about these markets.
Indeed, we don’t ever hear much complaint about anything going on in the stock market – be it derivatives, short-selling, algos, HFT or computers in general when the market is going up. Such complaint is only reserved for when the market is going down –
– for when those who have lost money are looking for scapegoats.
HFT is not, however, without its problems, and nor is it totally impervious to abuse. But then throughout time every new development in financial markets has brought with it previously unforeseen problems, and opened up loopholes which can be exploited before the regulators catch up.
Regulation is only ever imposed in hindsight, and most often by knee-jerk reaction following crises. In many cases, regulations are later relaxed, and the cycle resumes.
HFT is only another step in the evolutionary process.
There are no doubt those among you sitting here today who despise the concept of super-fast computers taking over from humans in the business that we know as financial market investment. You might even pine for the day when trading and profiting meant living and dying simply by the seat of your own pants.
And no doubt those same people also own a computer, an internet connection, a smart phone, and maybe even an iPod.
Can we ever force technological innovation to move backwards? Not often.
HFT is here to stay – get used to it.
Thank you for your time.
Epilogue
Back now in 2014, let me first disclose that the person who introduced walkie-talkies to the ASX options trading floor, way back when, was me.
In the context of the above speech, I now reiterate my suggestion made in the Prologue that the US stock market is not rigged but flawed. The simple fact that HFT specialist Virtu Financial can claim to profit on 99.9% of transactions is clear evidence. If Joe Average the investor could boast that sort of track record, he would eventually make Warren Buffet look like a pauper. Financial markets are the manifestation of the capitalist concept of risk. Financial markets globally function on the basis that there is no such thing as a pure “arbitrage”, at least not for very long, that guarantees no-risk profit. If Virtu can make what is essentially no-risk profit, capitalism is failing.
The US stock market has found itself in a capitalist conundrum. Competition is a cornerstone of capitalism, but in the case of publically listed financial instruments, transparency is another cornerstone. Given the myriad of quasi-exchanges that have been legally allowed to propagate in the US, the iconic New York Stock Exchange – that famed institution that sits on the corner of Broad Street and Wall Street – now only accounts for around 15% of stock market turnover. The bulk of the balance is now spread among the various electronic order-matching facilities at which transactions occur without transparency. There is no irony intended in the expression “dark pools”.
Not that HFT could ever specifically lend itself to “transparency”. HFT involves the exploitation of computer speed (and indeed in many cases, physical server location) to generate small orders within milliseconds to jump in and out of the market before any human could possibly notice, picking off bits of new orders or hopping in between the spread to entice new orders. If an HFT order is unsuccessful, it is just as quickly cancelled. Indeed, some 75% of all HFT orders placed are just as quickly cancelled, faster than the human eye could detect on an exchange ticker.
HFT is accused of being a form of “front running”. Front running in my days as a trader was an illegal practice in which, for example, a trader receives a large buy order from a client which he knows will move the market, first buys a few contracts for himself or “the house”, then works the order and saves the last few contracts to be bought back from “the house” at a virtually guaranteed profit. Front running in HFT terms involves exploiting the myriad of exchanges the same stock can now be traded on. An unsuspecting buyer can place an order which would take out apparent offers in aggregate across different exchanges, except that HFT can move faster, see the order coming in and switch to taking out other offers. The buyer then has to pay up further if the order is to be filled. Exactly the same tactics were once employed day in, day out on the old trading floors, albeit not at microsecond pace. A good operator on a trading floor could sense a big order coming in just by the body language of the operator holding the order. It was not illegal then and is not illegal now to attempt to move ahead of that operator. The issue now is that "real" investors can't see this happening.
Now, I suggested in my speech above that if more than 50% of the turnover of a stock market on any given day can be attributed to HFT, and given HFT trades are not carried past the closing bell, there must also be losers amongst the winners in this game. This is true, but competition amongst exchanges means HFT firms are paid to place orders in the form of liquidity rebates.
Since 2009, HFT activity has unsurprisingly grown exponentially in the US. Were HFT traders not to receive liquidity rebates, but had to pay brokerage like the rest of us, one presumes the practice would likely not be commercially viable. The rebates are offered because there are so many competing exchanges, or quasi-exchanges, in the US. “Real” investors will not use an exchange that appears to be illiquid. HFT provides for tight buy/sell spreads and thus attracts orders.
But” liquidity” is actually a bit of a misnomer in this case. A market is liquid if the volume of orders leads to a tight buy/sell spread. Individual HFT traders may tighten a spread but they provide no volume to speak of. The concept is thus misleading. Only as a pack do HFT traders add meaningfully to volume, and given they are simply skimming the day's orders to close square, standing in between the bids and offers for a penny here and a penny there, the volume they represent is actually providing no "liquidity" at all.
HFT evolved because the system allowed it to. HFT simply exploits ever evolving technology to gain an advantage within the regulations. There is little point in suggesting HFT should be banned. There will always be ways to exploit the system and the technology to provide for such exploitation. Just like walkie-talkies. What needs to be addressed is the system itself – the multiple exchanges, the lack of transparency, the liquidity rebate system, even the server proximity issue – such that the playing field again becomes level. If Virtu is making money 100% of the time then the playing field is clearly not level. And Virtu is doing nothing wrong.
One suggestion I have heard made this week has merit, that of charging a transaction fee on cancelled orders as well as filled orders. The “rest of the market” won’t mind, given not that many orders are cancelled in the real world. But in the virtual world of HFT, some 75% of orders are cancelled. It would be a rule for all rather than a singular attempt to directly undermine HFT, but the impact would be the same.
One thing is certain. Michael Lewis has opened a wound. A wound that has been festering since at least 2009 when the most in the world ever became aware of this thing called HFT. The vast bulk of commentary I personally have heard on US business television over the past week suggests a majority dismissal of the word “rigged”, but a majority agreement that something needs to be done nevertheless. Many agree that Michael Lewis has done the US stock market a favour by drawing attention to the matter.
So now it’s up to the regulators. We are reminded that we will mark the sixth anniversary of the Fall of Lehman this September, but the bill proposed soon after, intended to regulate US banks so the same thing could not happen again, is yet to be enacted.
So good luck on hoping for a swift regulatory response to HFT courtesy of the controversy Mr Lewis has sparked. It looks like HFT is still “here to stay” for a bit longer. And yes, the practice is rapidly growing in Australia.
Find out why FNArena subscribers like the service so much: "Your Feedback (Thank You)" – Warning this story contains unashamedly positive feedback on the service provided.
