Flash Boys: A Wall Street Revolt

This is a classic Michael Lewis book. It reads quickly. The topic is fascinating. The content is extremely insightful as the true technicalities of High Frequency Trading are either not covered or not understood even by the investment related media.Michael Lewis book follows three intertwined narratives.First, he opens the black box on what is high frequency trading (HFT). How it works, how it extracts rent profits from investors in the stock markets. There are currently over 50 stock market exchanges: 13 are public, and the rest are dark pools. The more market exchanges there are, the more arbitrage and front running opportunities there are for high frequency traders (HFTs) to exploit.Second, it narrates the history of the Investors Exchange (IEX) founded by a righteous quant type bunch who decided to start a stock market exchange that would eliminate all the HFT rent seeking strategies so to deliver a fairer market price to institutional investors trading on their platform.And, third it follows the strange life and career of the Russian computer programmer Sergey Aleynikov. He worked for two years for Goldman Sachs from 2007 to 2009 to render their computer trading systems faster and more competitive within the high speed world of HFT. He left Goldman Sachs with his computer codes that Goldman Sachs deemed proprietary. Goldman Sachs had him arrested by the FBI in 2009, and ever since he has been either engaged in trials prosecuted by Goldman Sachs or in jail.This third narrative also covers the ambiguous and evolving engagement of Goldman Sachs in HFT. At first, it attempts to become an engaged competitive high frequency trader itself. And, that is when it hired Aleynikov to improve its trading computers’ speed. Later, it will realize that chasing the HFTs in a speed competition is a losing proposition. And, it will become the only major Wall Street investment bank to fully support the Investors Exchange (IEX) to counter and neutralize the nefarious impact of HFTs.Going back to the first narrative, High Frequency Trading extracts rent profits from institutional investors (and their retail investors) in three ways.The first way is by beating the investor to the stock market gateway and quickly buying and reselling the stock to the investor at a small profit. They call it “electronic front-running.” To do that, you need to be fast. That is where the nano second trading speed comes in. The “co-location” of the HFTs servers next to the ones of the exchanges plays a major role by reducing the electronic distance travelled and maximizing trading speed.The second way is by exploiting a complex system of kickback and rebates on trades implemented by the various exchanges themselves. They call it “rebate arbitrage.”The third way appears similar to electronic front-running, except that the HFTs exploit minute price discrepancies between the various exchanges before the exchanges themselves have had a chance of correcting those. They call it “slow market arbitrage”. Apparently, of the three rent seeking strategies this is the most lucrative one for the HFTs.The above strategies are implemented within a market universe that is alien to individual investors and most institutional investors. This market universe has interesting characteristics. Its foundational one is an unfathomable stock trading speed measured in the 1/10000 of a second. Such speed relies on extra fast fiber optic networks and computer servers located extremely closely to the servers of the stock exchange themselves. Another characteristic is the HFTs purchasing customer order flows from the Wall Street brokerage houses. The latter now make more money from selling those customer order flows to HFTs than from trading itself. In essence, Wall Street sells proprietary customer order information to the HFTs, so the HFTs can front run these same customers (their stock orders). And, somehow SEC laws have still not caught up to this apparent infraction of the integrity of the stock markets. That’s even though the mentioned HFTs rent seeking strategies are at least a decade old.So, next time when you think your brokerage house is acting in your best interest, think again. It is acting in the best interest of the HFTs and itself by making money on selling your order information to the HFTs. And, we are talking millions if not billions of dollars in total annual revenues for the Wall Street brokerage houses.Going back to the second narrative, to correct for all those markets flaws exploited by the HFTs, Brad Katsuyama, a former trader at Royal Bank of Canada, will create a “fair” exchange: the Investors Exchange (IEX) in 2012. This exchange takes specific infrastructure measures to entirely eliminate all the exploitative advantages of HFTs including: 1) ensuring market pricing data arrives at external points of presence simultaneously; 2) slightly delaying market pricing data to all customers (no co-location, HFTs servers are not allowed proximate to the IEX servers); and 3) IEX refuses to pay for order flow and does not offer related trade rebates of any kind. The majority of Wall Street banks and HFTs will do everything possible to kill this emerging “clean” exchange in its infancy. This is because they collectively extract yearly rent-profit in the $billions on the back of retail and institutional investors. However, as mentioned one of the main player will break rank as Goldman Sachs ultimately decides to support IEX by routing a good portion of its trades to IEX. Goldman Sachs understands that what IEX is doing to restoring integrity in the equity markets is critical. And, as a result IEX survives. Nevertheless, it is not entirely encouraging when evaluating how much impact IEX has in restoring the integrity of the US equities markets since it captures less than 3% of its volume to this day. In other words, over 97% of such market trading volume still is done under the exploitative rent-seeking system abused by the HFTs (electronic front running, etc.) and the other Wall Street banks (making more money from selling their customer order flows than actual trading).The third narrative about Sergey Aleynikov and Goldman Sachs evolving position regarding HFT is very interesting because of its ambiguity. Aleynikov used mainly open source software to develop his codes to improve Goldman Sachs computer speed. When he accepts an offer to join Teza Technologies (who offered to triple his compensation from $400k to $1.2 million), he decides to copy and take his computer code on a USB drive. At such point, Goldman Sachs aggressively pursues him (gets him arrested by the FBI, tried, and jailed). At the time, Goldman Sachs considered the mentioned computer codes to be proprietary and critical to its competitive position within the HFT environment.Michael Lewis will engage with many industry insiders (HFTs, computer programmers, etc.) and solicit their opinion on whether Aleynikov was truly guilty of stealing proprietary company codes or not. Almost unanimously this crowd of insiders advance that Aleynikov was innocent. And, that his practice of copying his own open source based codes when he moved to another employer is absolutely standard within the computer programming community. Aleynikov also indicated that he had no use for Goldman’s proprietary codes as they were very cumbersome catered to Goldman’s antiquated legacy computer systems. When Michael Lewis talked to outsiders like institutional investors, they were far less lenient. And, they typically considered that Aleynikov was clearly guilty of stealing proprietary codes.As indicated, Goldman Sachs at first vigorously pursues Aleynikov in order to protect its position in terms of trading speed within the world of HFT. Much later, when it decides to give up on the speed competition and decides to do just the opposite by supporting IEX, Goldman Sachs does not pursue Aleynikov as adamantly anymore. But, by then the legal system takes a life of its own. As a result, some of the related lawsuits are still going on to this day. Aleynikov is nearly bankrupt and has an online legal defense fund to raise money to mount his defense and reclaim his innocence.

This book is an interesting "story" ... but it leaves me very suspicious. If I were sitting in the meeting where it was first proposed that HFT be tried I would have been able to immediately describe how it could be (and should be) neutered as a tactic ... at virtually zero cost.HFT requires two things: (1) timing predictability and (2) being able to "cut in line". Both of those needs are easily made impossible. Thus, a process can be designed that makes both impossible. And just like the whole HFT process requires quick adaptation, the process could have easily imposed a fix as an adaptation. Further, IEX could have designed their process with such an adaptation even if the others didn't. Their strategy would have remained the same: Attract business by being fair to everyone ... i.e. treating everyone the same ... i.e. unpredictable and sufficiently long latency ... not just in the order time stamping, but throughout the system responsiveness. Investing is not something about changing trading decisions within mili-seconds of committing to them. It is, by definition, a much longer term decision. Thus, any latency making all players equal would have no effect on the principal purpose of the process ... to bring buyers and sellers together in a fair, non-gamable, process.I'll now just give some of the annotations I made to the book in the order I made them (with the hope the author sees them):Pg. 174: "for instance one professor suggested a 'randomized delay'" ... and later: "Someone will just flood the market with orders" was mentioned as why this would not work. But it obviously would work. If no one can predict the process timing, no one can benefit from that timing. Focusing just on the orders was not the solution. But here, randomizing was just dispelled by this simple assertion and never brought up again. A $300 million secret effort taking over a year (i.e. the fiber from Chicago to NYC) would have been neutered in a matter of weeks ... or even days of becoming visible. Way before it could repay itself.Pg. 177: "To create a 350-microsecond delay, they needed to keep the new exchange roughly thirty-eight miles from the place brokers were allowed to connect to the exchange".With randomized latency this was unnecessary. With a "known" latency, it isn't even effective. I had seen the "coil the fiber" when working for a major telecom carrier six years earlier. They had the fiber on a spool and just tapped into the ends ... no big deal at all. Again ... that really fixes nothing, just moves the issue 350 micro-seconds down the road. Nobody wins by betting on the instant replay if all everyone sees is the instant replay.Pg. 194: "You couldn't see the plane," he said. "You just saw the explosion".Did "anybody" see the plane? Why did WTC7 mysteriously fall down? These people keep behaving like they believe the government's 19 cavemen with box cutters conspiracy theory tale.Pg. 277: "The authorities evidently felt the need to leap into actions, or to appear to"The "or to appear to" speaks volumes. The "authorities" are well known to be co-conspirators.Pg. 281: "These savings were fully realized by 2005 and were enabled less by high-frequency market-making than the Internet, the subsequent competition among online brokers, the decimalization of stock prices, and the removal of expensive human intermediaries from the stock market."The part regarding the "internet" is nonsense. If you didn't subscribe (and pay a fee to a broker or a data feed), you had a 15 minute delay in price reporting ... on the TV and on the internet. You were betting on the instant replay. Do you think that fact was there to protect the little investor?Pg. 281: "the cost to investors of trading in the U.S. stock market has, if anything, risen... possibly by a lot."This is absolute nonsense. Just digitalizing the spreads removed enormous cost. Then reducing and removing human interaction removed still more cost. All that HFT was about was front running and picking up crumbs. In the oil field, a business called "gathering" tapped into flaring gas. In the accounting business, diverting to a special account rounding to the penny residue did the same thing HFT is doing. It's not new. And it's not going away. Government regulators "always" enable such things.Pg. 282: "The attention became focused almost entirely on high frequency trading..."The fact that high frequency trading was cost effective was proof it was a bad guy. It could have been neutered by introducing random delays...within days of the scheme going live. Such a simple fact would have kept someone from trying to go live with it in the first place.Pg. 282: "but perhaps it's too much to expect Wall Street traders to worry about the social consequences of their actions."People will "always" act in their own self interest. And that is "always" enabled by government when it comes to really big scams. Being a "nation of laws" instead of "a nation of principles" makes it legitimate. If it's not against the law, it's ok. Thus we get 40,000 new laws each year.Pg. 282: "I honestly don't fee that strongly about high-frequency trading."Oh really? How do you feel about "gaming" the process.Pg. 282: "If I didn't do more to distinguish "good" HFT from "bad: HFT, it was because I saw, fairly early on that there was no practical way for me or anyone else without subpoena power to do it".There is "no" good HFT. Follow the money! He got all the information he needed in simple conversations ... no subpoenas at all. It would never have been tried if proper transparency and process existed. It was obviously gaming...masquerading as arbitrage. Random delay would have rendered it impossible from the beginning!Pg. 282: "... firms need to reveal the contents of their algorithms"No they don't. That just opens a new attack vector. Who gets to see those algos first? Make it impossible for any algo to have an advantage. That's the solution! Random delays deliver that solution.Pg. 283: "They designed IEX to eliminate predatory opportunities created by speed".And all they had to do was introduce random delay to those using their process. This neutered all HFC advantage yet doesn't resort to knowing what that advantage might be. Interestingly there's no mention of the tactic of placing orders and immediately canceling them. The process allowed that.Pg. 286: "If this story has a soul, it is in the decision made by its principal characters to resist the temptation of easy money and to pay attention to the spirit in which they livce their working lives."This is what hit me first about HFC and gaming processes in general. It sanctions cheating in a game as a valid tactic. This is why we need to be a "nation of principles"; not a "nation of laws". People violating principles will be naturally ostracized. Laws just deem everything not against the law to be lawful.Pg. 286:"That some minority of Wall Street is getting rich by exploiting a screwed-up financial system is no longer news".As if it ever was news. The government scheme of looking the other way and then fining as a new income stream is becoming government (and those they regulate) modus operendi. Ostracize both!Pg. 286: "All they need is a little help from the silent majority"What they need is a little help from plain ole common sense. The silent majority is the "clueless" majority. By my poll, less than 6% of USA voters know anything about WTC7.