Flash Boys, a short review

Flash Boys, a short review

If you’ve ever taken an economics class, you probably know about arbitrage: exploiting the price differences of an asset on different markets by buying low, selling high, and pocketing the difference. I recently finished reading Michael Lewis’ Flash Boys, a book that sheds light on the many faces of arbitrage in the 21st century. Flash Boys is a story about the extent to which Wall Street banks, hedge funds, and other financial institutions apply high frequency trading (HFT) techniques to gain an edge over other players in the most competitive financial markets.

The book tells three separate, tangentially related stories. First, the story of how one unassuming trader - Brad Katsuyama - observed the irregularities that big players caused on the stock market, and what he did about it. Katsuyama left his job at the Royal Bank of Canada to start IEX, an exchange with built-in delays that would guarantee a level playing field for all market participants. His story takes up the bulk of the book. Second, the story of Sergey Aleynikov, the Wall Street programmer who was prosecuted for stealing trade secrets (aka, copying code) from his ex-employer, Goldman Sachs. Lewis’ Vanity Fair profile of Aleynikov is worth reading on its own. Lastly, the book discusses Spread Networks and its infamous fiber optic cable. Running from New York to Chicago, Spread’s line promised to shave off fractions of a second from each transaction, providing those willing to pay for it with a minuscule but significant advantage. By braiding these three narratives, Lewis provides the reader with a set of broad explanations of the complexity of modern finance, and the peculiarities of the policy and technological decisions that got us where we are.

The book is well written, very entertaining, and extremely controversial in the finance world. A lot has been said about the objectivity and bias of Lewis’ portrayal of the high frequency traders, and I’m not going to try to defend or deny either view in this post. Instead, I’ll try to summarize what I thought were the most interesting aspects.

HFT is a bad name

If there was one thing I learned from reading this book, it was that HFT is an awfully named concept. As a software engineer with a degree in economics you’d think I’d know better, but until now I really thought that HFT stood for making many trades, as the “high frequency” would imply. However, the reality is that HFT is not about the quantity of trades, but making higher quality trades by capitalizing on faster assymetric access to information, higher precision, and extremely low latency decision making, to realize gains. The high frequency does not refer to the trades, but to the granularity of the data in play.

Counterintuitive Game Theory

In one of the scenes, there is a discussion about Spread Network’s pricing. When asking the big banks how much they’d be willing to pay, the buyers tried to negotiate for a higher price! This seems counter intuitive, but a “13.10-millisecond round trip between New York and Chicago” is only worth it if you’re the only one who can exploit it, so you’d better make sure you’re the only one able to pay for it before you pay for it.


“The best minds of my generation are thinking about how to make people click ads.”

-Jeff Hammerbacher1

Or, if you are to believe Flash Boys, the best financial engineers of our generation are thinking about how to scalp a penny per share by being a microsecond faster than the rest. Whether HFT provides real value to the market is unclear, but I am certain that whatever value it provides is less than the input cost of building the necessary infrastructure for it. Not too long ago, the stock market operated in 1/16th of a dollar, and things were just fine. The added precision gained by accurately pricing to the penny by the microsecond is, in my view, not worth it.

Front Running

Traditionally, a broker could profit by making trades based on the knowledge of their clients’ pending orders - front running their own books. While unethical, the practice seems to be quite common. Katsuyama’s grand discovery was that the HFT firms had come up with a new way to do this.

HFT front running Image taken from Wikipedia

Since information reaches different exchanges at different times, a fast enough player could see that an order coming to exchange A would only be partially fulfilled. By assuming the parallel order would reach exchange B next, said player could use the information from the trade in exchange A to place the opposite end of the trade at a higher price in B, and pocket that spread. By timing when orders were routed to each exchange, Katsuyama’s team could work around this phantom liquidity, and get the securities they wanted at the advertised prices.

Fake Orders and Complexity

A corollary of the above front running example is that HFTs require knowledge and decision making power to be as close as possible to each exchange. The orders on the other side of the trades Katsuyama and team were missing were only meant to probe the market. They were other investors’ traps to gain information about their intentions, and not at all to be traded.

The orders at A were placed knowing they’d lose money, but would risklessly get them back trading in B.

For this added complexity, Lewis blames “Reg NMS,” an SEC directive that intended to ensure the buyer gets the “best available price” from a consolidated “national best bid and offer” (NBBO) feed. Since centralizing said feed takes time, the HFTs could capitalize on that added latency to make their own guesses as to what the NBBO will be before the “real” prices were ready. Thus, the traders were economically incentivized to add direct links to all the exchanges, and once those were in place, they could have an inside view of the trades before anyone else.

The Victims

One thing that Lewis seems to skip over is the fact that the average retail investor is not really affected by high frequency trading, or at least not directly. I presume the book is fuzzy on this on purpose, to make the reader feel outraged about mom and pop getting screwed by the big banks.

However, if you’re buying 100 shares of Apple on E*TRADE, no one will front run you, and your trade will probably be filled just fine2. If you’re buying into an ETF, then sure, you might be losing a few pennies to the dark pools via your broker, but then we’re talking about the big boys fighting each other over those pennies on extremely large volume trades. Why your broker still thinks that being in a dark pool, and trusting their owners, is a good call is a question I have answers for.

Anyway, go read it. The interconnectedness of all these auto-triggering algorithms lead to strange places, and learning a bit about how these self-reinforcing feedback loops work was definitely worthwhile. If anything, it surely showed me how much more I need to learn about how markets work.

Photo: “Charging Bull - New York City” by Sam Valadi, taken from Flickr.

  1. This Tech Bubble Is Different, by Ashlee Vance 

  2. Matt Levine’s interview with Katsuyama has a section on this. 

Want to see more articles like this? Sign up below: