Michael Lewis’s Flash Boys – and its Accuracy


Ever since Flash Boys was released this past spring, stock market insiders have not been shy in their complaints of a few factual inaccuracies in the book. These complaints have been spewed loudly at market structure conference after conference, in speeches, notes to clients, and even in blog posts. Most troubling is how they have been spewed by senior regulators. At first, we were amused. We privately laughed – knowing that the book had struck truth, and knowing that the best proof that it had done so was the vitriol it inspired among its conflicted critics.

A few months back, a blog appeared called Lewis Fiction, and it repeatedly pointed out small errors in the book. Its idea was/is that these small errors discredit the premise of Flash Boys. We have ignored it for the most part, much as one ignores crazy Uncle Louie at Christmas parties, or gnats in our backyard. Make no mistake – Michael Lewis in Flash Boys eloquently captured a big problem, and big ideas, and told it clearly to investors everywhere. And its critics have attached themselves to small corrections and little ideas.

A month back we guest-published a point by point response  to the false claims made by the Lewis Fiction blog written by R.T. Leuchtkafer. It was titled Guest Post: R.T. Leuchtkafer shows how poorly informed the LewisFiction Blog Is.

The Lewis Fiction blog continued to put out drivel. Its Twitter sent out “promoted” tweets. Lies continued to be repeated, and sadly to be passed around the industry behind the scenes by people who know inside the book is correct, and who do know better. It frankly is grating.

This is why today, we wish to share with you a second point by point response to Lewis Fiction’s false attacks against Flash Boys – again written by R.T. Leuchtkafer. Please read this! And please send it to the SEC’s Trading and Markets (tradingandmarkets@sec.gov), to your friends, and tweet it. Lies by others do not deserve to be spread, and they deserve to be set straight. Without further adieu, here is R.T. Leuchtkafer’s 2nd response unedited and pure:


Our last time picking through a batch of @Lewisfiction’s “fact checking” was a lesson for all of us in how ill-informed Lewis’s critics often are.  Not only did @Lewisfiction get the basic architecture of the National Market System spectacularly wrong, he turned out to be the Emily Litella of the market structure debate, unable to parse simple sentences.  Anyway, here we go again with another batch of it.



“At the behest of the SEC, in turn responding to public protests about cronyism, the exchanges themselves, in 2005, went from being utilities owned by their members to public corporations run for profit.”– Page 35 of Flash Boys

This is complete fiction! Trading in the common stock of The Nasdaq Stock Market commenced on July 1, 2002 on the Over-the-Counter Bulletin Board Service under the symbol “NDAQ”.[1] NASDAQ’s public status was absolutely NOT at the “behest of the SEC”. In fact, NASDAQ became a public reporting company simply as a result of having too many shareholders under the US securities laws. Any such suggestion that it was “at the behest of the SEC” is an absolute fabrication of the true facts. The NYSE went public in 2005 as a result of a merger with Archipelago Holdings, Inc., a public company at the time.[2] Again, the NYSE’s merger was not at the “behest of the SEC” or encouraged by the SEC. This statement is just another example of Mr. Lewis fabricating facts.

Saying Nasdaq went public because it had too many shareholders is as keen as saying The Godfather is a movie about a family business: so much more is left out than explained.  Nasdaq separated from NASD and became an independent firm in part because a 1995 report recommended it.  The hope was to eliminate governance problems and conflicts of interest at NASD – in other words, cronyism.  Those governance problems and conflicts were exposed at length in the SEC’s 1996 21(a) report on the Nasdaq quote rigging scandal, at the time the most far-reaching market scandal in decades.  Then, in 1998, for the first time in history, the SEC was explicit – exchanges could be demutualized, for-profit entities.  The next year NASD decided to spin Nasdaq off and sell stock in the new entity, and NASD soon approved a $1 billion restructuring to turn Nasdaq into a privately owned, for-profit company.  

So, yes, after a decade of scandals, investigations, enforcement actions, Justice Department and SEC settlements, and an historic public policy statement, Nasdaq ended up with more than 500 shareholders and became a public reporting company, at last listing its shares on its own exchange in 2005.  But there’s quite a bit more to the story than @Lewisfiction suggests, isn’t there?

Scandal wasn’t limited to Nasdaq.  NYSE had its own scandals.  In 2005, that magical year, the SEC went after the NYSE, saying that “NYSE, over the course of nearly four years, failed to police specialists, who engaged in widespread and unlawful proprietary trading on the floor of the NYSE.”  This was only two years after an uproar over NYSE Chairman and CEO Richard Grasso’s pay, an imbroglio that saw the NYSE overwhelmed by “criticism over its corporate-governance standards” and then compelled to adopt “a slate of overhauls that will lead to the disclosure of Chairman Dick Grasso’s pay and force exchange officials to step down from the boards of NYSE-listed companies,” according to the Wall Street Journal.  The Journal also reported that the changes would decrease “the potential for conflict of interest on the part of NYSE directors serving on certain committees, some of whose firms are regulated by the Big Board, and on increasing the transparency of the exchange’s governance.”  Part of the fallout from all this was Grasso’s departure after a 35 year career at NYSE.

And so precisely because it was worried about cronyism at the exchanges, the SEC encouraged and welcomed the demutualization of the exchanges.  How do we know?  At the very least because the SEC took credit for it.  This is what SEC Chair William Donaldson said when NYSE announced plans in 2005 to become a public company, “But broadly, it seems to me to be a positive development for investors illustrating just how aggressive the markets will become in competing with one another now that Regulation NMS has leveled the playing field between the NYSE and Nasdaq markets.”  Donaldson attributes NYSE’s decision to the SEC’s approval of Reg NMS, which he believed would force exchanges to compete more aggressively for order flow.  Reg NMS was enacted in no small part because of one form of cronyism at the exchanges, where exchange members regularly refused to honor quotes on other markets, behavior which stifled competition and served exchange monopolies.  

Beginning in the early 1990s in Europe, and soon all over the world, including in the U.S., regulators believed that demutualization would offset the influence of exchange officials and members – cronyism – and force exchanges to compete with new products and services without their members and other entrenched interests strangling innovation.

Michael Lewis is exactly right.




“In general, the exchanges charged takers a few pennies a share, paid makers somewhat less, and pocketed the difference – on the dubious theory that whoever resisted the urge to cross the spread was performing some kind of service.”– Page 36 Flash Boys

This too is wrong and total fiction! Exchanges do not charge “a few pennies a share”. Exchanges are subject to a 2005 SEC imposed cap on access fees under Rule 610 of Regulation NMS. Rule 610 caps access fees at $0.0030 per share (or 3 tenths of a penny per share). In fact, exchanges have not charged “a few pennies a share” over the last 20 years. This claim illustrates how Mr. Lewis misunderstands the market a lot more than he understands it.

In addition, the concept of remunerating the “maker” under the maker-taker model was originally designed to subsidize the market making function associated with providing competitive bids and offers in the market. In fact, if Mr. Lewis performed a simple Google search he would have come across a study called “Subsidizing Liquidity: The Impact of Make/Take Fees on Market Quality” authored by Katya Malinova and Andreas Park of the University of Toronto, who concluded “that the liquidity rebate structure leads to decreased spreads, increased depth, increased volume, and intensified competition in liquidity provision.”[3]

Perhaps Lewis said “share” when he meant “trade”; perhaps he said “few pennies” when he meant “fractions of pennies.”  I’m sure you’ve made him weep.  Of course it has very little to do with the point of the sentence, which is that exchange business models changed to put payment for order flow on center stage, and Michael Lewis is exactly right



“He soon realized that, while RBC would allow him to apply for awards, it would not let him describe publicly what Thor had inadvertently exposed: the manner in which HFT firms front-ran ordinary investors; the conflict of interest that brokers had when they were being paid by the exchanges to route orders; the conflict of interest the exchanges had when they were being paid a billion dollars a year by HFT firms for faster access to market data…” (Emphasis added)– Page 104 of Flash Boys

This statement is fiction and intended to scare and confuse the public. First, all the exchanges in the US combined do not collect “a billion dollars a year” in direct market data fees and collocation revenue. Mr. Lewis does not provide any detail or support for this claim so it is difficult to determine what he means by “faster access”. Regardless, the handful HFT of firms that operate in the US cannot possibly add up to “a billion dollars a year” in market data revenue for the exchanges when the entire industry does not generate that amount for all the exchanges.Mr. Lewis is once again making up facts even when the true facts are available on Google.


According to NYSE Euronext’s 2012 10K, it earned $341 million for “technology services.”  These include connectivity and collocation revenues, as well as software license and maintenance fees.  NYSE Euronext also earned $348 million for “market data,” a figure which includes its proprietary data products as well as consolidated tape revenue.  (We’ll haircut this figure to $200 million to exclude a generous estimate of what NYSE earns from the consolidated tape.)  Nasdaq’s 2012 10K reports a combined $465 million in revenue for “Access Services” and U.S. and European “market data products” (excluding SIP revenue).  CME’s 2012 10K shows a combined $475 million in revenue for “Market data and information services” and “Access and communication fees.”  

By now we’re already at $1.481 billion and we haven’t yet got to the CBOE, ISE, or BATS/Direct Edge.  If anything, Michael Lewis underestimates what exchanges charge for privileged, faster access to market data.



“By late 2011, when Bollerman quit his job (‘I felt there was a lack of leadership’), more than two-thirds of NASDAQ’s revenues derived, one way or another, from high-frequency trading firms.”– Page 163 Flash Boys

This statement is FICTION. According to a 2014 Nasdaq OMX release, HFT revenue only accounted for between $23 million and $29 million of NASDAQ’s first quarter revenue, or 1% of total revenue.[4] Even a report by Raymond James in 2012 estimated NASDAQ’s HFT related revenues at between 17% and 17.8% of its total revenue in 2011 and 2012.[5] Once again, Mr. Lewis uses quotes or a character in the book to introduce concepts in his storyline as facts that are clearly inaccurate. Of course, this approach allows him to weave any fictional story he desires without taking responsibility for accuracy! How long before Flash Boys is re-classified as FICTION?

Using data supplied by Nasdaq, a well-known research paper estimated that HFT firms participated in 70% or more of Nasdaq’s U.S. volume in 2008 and 2009.  And that paper likely underestimated the amount of HFT activity on Nasdaq because the data didn’t identify the HFT activity of large investment banks like Goldman, among other limitations.  

So to the extent that Nasdaq’s revenue derives directly or indirectly from its transaction volume – revenue that includes market data fees, SIP revenue, collocation revenue, non-HFT participant trading volume revenue, listing revenue from issuers attracted to Nasdaq because of its active market – it’s obvious from this statistic alone that the lion’s share of Nasdaq’s revenue derives directly or indirectly from HFT firms.  

In other contexts HFT proponents muse about what the market would look like without them – nothing but the sound of crickets, they say – but huff and puff when Lewis claims HFT firms drive exchange revenues.  Take your pick, friends.  HFT can’t be responsible for all sorts of supposed goodies without at the same time powering exchange income. 

Michael Lewis likely underestimates how much Nasdaq earns, one way or another, from HFT firms.




“Instead this new beast rose up in the middle of the market and the tax increased—by billions of dollars. Or had it? To measure the cost to the economy of Scalpers Inc., you needed to know how much money it made. That was not possible. The new intermediaries were too good at keeping their profits secret.”– Page 109 Flash Boys

Michael Lewis is spinning a fictional tale and this statement – like countless others – is absolutely inaccurate. Mr. Lewis proves the inaccuracy of this statement on the same page in his book by footnoting the fact that “one of the largest high-frequency traders” filed an S1 on 10 March 2014 with the SEC detailing its financials and its historical financial results. Mr. Lewis had to have read the public filing to craft that footnote but yet he mysteriously missed the financial results of the “one of the largest high-frequency traders”. Additionally, Getco, also one of the largest high-frequency traders had filed its historical financial results as part of its merger with Knight Capital in 2013. Mr. Lewis’ claim that it “was not possible” is absolutely inaccurate. It is obvious that he makes this claim because the financial results of the “largest high-frequency traders” did not add up to his claim of “billions of dollars”.

Getco’s S4 for the Knight transaction and Virtu’s S1 were watershed events, the first time ever large HFT firms showed their finances in public.  (Bizarrely, though, Virtu’s S1 noted a “material weakness related to our inability to prepare accurate financial statements.”  How’s that for transparency?)  And what about HFT giants like Citadel, Renaissance Technologies, Jump, DRW, Quantlab, RGM, HRT, Tower and others?  Some foreign regulators might require them to disclose financials for their in-country operations, but their consolidated worldwide financials stay private.  The vast majority of firms in this industry keep their financials private.  

Michael Lewis is exactly right.



“For instance, they bought 10 million shares of Citigroup, then trading at roughly $4 per share, and saved $29,000—or less than a tenth of 1 percent of the total price. ‘That was the tax,’ said Rob Park. It sounded small until you realized that the average daily volume in the U.S. stock market was $225 billion. The same tax rate applied to the sum came to more than $160 million a day.”– Page 52 Flash Boys

This statement is NOT supported by facts. Mr. Lewis’ attempt to quantify this so called “tax” is absurd and reflects his true lack of understanding of the market and unwillingness to learn how the market functions. First, the “savings” calculated by the RBC team is suspect given there is no explanation how they actually saved money. Second, the example used is on the 10 million shares of Citigroup which, at the time, was one of the most liquid instruments trading in the US equity markets. In 2009, CITI was trading on average over 1.2 billion shares per day.[6] In 2009, trading 10 million shares of CITI did not require any special routing technology or routing sophistication. More importantly, Mr. Lewis’ attempt to extrapolate the RBC savings in the most liquid symbol on the market to all trading across the market defies logic. There were close to 6,000 listed companies in 2009. Using the most liquid symbol, CITI, to calculate a savings on each of the other 5,999 listed companies is absurd and disingenuous.


The fact that Citi was one of the most liquid instruments would tend to minimize Park’s estimate of the “tax” paid.  The “tax rate” on less liquid stocks could be far higher.  If anything Lewis’s estimate of the tax might understate the full cost of predatory algorithms that regularly try to trade ahead of investors.  And Lewis doesn’t say $160 million is scalped from the market every day.  He cites an example and extrapolates from it, and is very clear about the figure and how he has computed it.  And of course many others refer to HFT predatory behavior as an HFT tax or rent.  Seth Merrin of Liquidnet did so just recently at a talk in Australia. At a 2010 meeting of the CFTC-SEC Joint Advisory Committee on Emerging Regulatory Issues, Nobel Laureate Joseph Stiglitz said HFT firms “aren’t adding information, they’re just reprocessing other peoples’ information, and in that sense, they almost inevitably are taking away — to the extent that they make a return, they’re taking away rents from those who are providing information.” 

Michael Lewis might well underestimate the full cost to investors of HFT predation.  




“The reality turned out to be a windfall for financial intermediaries—of somewhere between $10 billion and $22 billion a year, depending on whose estimates you wanted to believe.”– Page 135 Flash Boys

This is FICTION! Mr. Lewis conveniently fails to reference his sources for this claim. Contrary to Mr. Lewis’ claim, Rosenblatt estimated that in 2009 the entire HFT industry made around $5 billion trading stocks. Rosenblatt also estimated that in 2012 the entire industry made closer to $1 billion. Tabb Group, which also estimates HFT profits, estimates that the industry made $1.3 billion in 2012, down from $7.9 billion in 2009. Once again, Mr. Lewis seems to be making up facts that align with the story he wanted to tell – not with what he would have learned with a simple Google search.

There’s a sleight of hand here.  Lewis estimates $10 to $22 billion a year for HFT firms.  But @Lewisfiction flips that into estimates of what HFT firms earn from equities to challenge Lewis’s figures.  Of course HFT firms trade options, futures, and currencies as well as equities.  Tabb estimated 2008 revenues of – what do you know? – $22 billion for the HFT industry as a whole, just as Lewis reported.  And while Tabb is a very credible firm, it’s unclear how Tabb conducts these surveys and who it includes.  Does Tabb include such famously secretive firms as Renaissance Technologies or Citadel?  Renaissance’s Medallion fund was rumored to have generated 80% returns in 2008.  How many billions were there?  


Michael Lewis probably understates HFT revenues.




“At the same time, the exchanges were changing the way they made money. In 2002 they charged every Wall Street broker who submitted a stock market order the same simple fixed commission per share traded.”– page 36 of Flash Boys

This statement is factually inaccurate. In 2002, the NYSE and NASDAQ maintained volume based pricing where each broker was charged a price based on each brokers volume. In fact, the NYSE maintained a cap on transaction fees that only the largest brokers reached. Under NYSE pricing, all transactions were free over the fixed fee cap. Mr. Lewis (and his editors) must not have Google.

Across the industry, today’s exchange fees are not only volume-based, exchanges charge for taking liquidity, they charge for providing liquidity, they rebate for taking liquidity, they rebate for adding liquidity, they charge different rates for different order types, they charge different rates for different participant classes, and they divvy up tape revenues, all of this adding up to a Babel of fees and rebates indecipherable to anyone who can’t devote a career to understanding it.  Importantly, firms game these fee and rebate schedules with intra- and inter-exchange strategies that have nothing to do with healthy intermediation or investment.  

Perhaps Lewis meant “a simple fixed commission per share.”  Once again, you’ve surely made him weep.  Of course the point is that exchange fee schedules today are needlessly complex and gameable compared to what they once were, and of course Michael Lewis is exactly right.




“For $300,000 a month plus a few million more in up-front expenses, the people on Wall Street then making perhaps more money than people have ever made on Wall Street would enjoy the right to continue doing what they were already doing.”– Page 18 of Flash Boys

This is another fictional ‘fact’ in Flash Boys! In 2010, Spread Networks began marketing the low-latency line through the use of NDAs. The Spread Networks salesmen violated those same NDAs regularly by telling prospective clients which competitors were planning to lease the fast line. This technique of selling the low-latency line through rampant violations of the Spread Networks NDAs and fear was the reason why so many market participants were angry with Spread Networks.

In 2010, Spread Networks was targeting firms like Getco and Hudson River Trading. Getco recently reported its 2010 total global revenue of $865.0 million. In 2010, the investment banks Goldman Sachs and JP Morgan Chase reported total revenue of $39.1 billion and $104 billion, respectively. It is complete exaggeration by Mr. Lewis to suggest that the Spread Networks prospective clients, i.e., Getco and Hudson River Trading, were “making perhaps more money than people have ever made on Wall Street.”


Let’s talk about Getco and Goldman.  According to filings, Getco’s all time peak average compensation per employee was $1.688 million in 2008.  Goldman’s all time peak compensation per employee was $661,000 in 2007.  

Using these measures, Michael Lewis is exactly right.


(Just as an aside, according to a recent filing Tower Research’s UK affiliate’s average compensation per employee in 2013 was nearly $1 million, as was Jump Trading’s.  Last year Goldman’s average compensation per employee was $383,000.)  




@Lewisfiction, it’s long past time for you to go home and feed your cats.  However sincerely you might feel about it, however much you don’t like his book, however much Michael Lewis hurt your feelings, you just don’t know what you’re talking about.