I've spent more than a few words bashing Flash boys so I won't try to do it anymore.<p>In a way I feel kind of bad for Michael Lewis. He's had such success with earlier books that it must be hard for him to find a target for each new book. In my opinion this book was his first awful book.<p>Instead of finding data and following it through to reach a conclusion he starts with a trendy conclusion, HFT is bad, and then really contorts and reaches with his data to try and make his case. He was also hit by the issue of HFT starting to wind down around 2010 and really starting to wind down around 2012. Many HFT firms no longer concentrate on the US Equity markets alone, they do alot more volume in the futures, options and bond markets where there are alot more mathematical correlations, and hence more opportunities for mispricings to correct.<p>If you want a sound rebuttal, and in my opinion a better book have a look at this book:<p><a href="https://news.ycombinator.com/item?id=8577237" rel="nofollow">https://news.ycombinator.com/item?id=8577237</a><p>I find it more balanced, better researched and it provides a much better understanding of the HFT industry.<p>Having said that, credit where credit is due, he's a tremendous writer and the book is a fun read that you can consume in 2 days. If you like it then I recommend his entire back catalog, they are all fun and informative reads.<p><i></i>EDIT<i></i> someone asked for an example of why I didn't like the book so....<p>One of Lewis' biggest issues was what he termed HFT front running. This alone was a pain as front running already had a well understood meaning( it originally was meant for a intermediary who took your order and executed its own order ahead of yours, often buying the stock lower and selling it to you at a slightly higher price).<p>His example was that a HFT system would see an order for say Microsoft trade on the exchange BATS, say a buy of 1000 shares. The HFT system would then run ahead to all the other exchanges and buy up the remaining shares on those other exchanges, an ammount that might be 1000 shares or it might be 100,000 shares, so when the remaining part of the order to buy Microsoft got to those exchanges it would have to buy the shares from eh faster HFT system at a higher price.<p>This is down right silly for a few reasons.<p>1) it assumes that the order that got filled at BATS was either a market order or a limit order that was priced for more than the fill price, because if it was a limit for the same price as the fill then buying up the remaining shares won't do the HFT firm any good:)<p>2) it assumes that the order was for more shares than what got filled on BATS, so now the HFT firm is exposing itself by owning shares it's not even sure anyone wants to buy at the price its willing to sell.<p>3) it assumes that the HFT system can both buy the remaining shares at the original price and get to the top of sell side of the order book to sell those shares back. The HFt firm now has to take the risk that it won't be at the top of the order book and even if it was right about steps one and two, a big if, it might not be able to capitalize on it as it can't jump ahead of anyone who had previous sell limit orders at the price it now wants to sell at.<p>There are just so many unknown factors there that there is almost no way this type of trading system could be profitable.<p>Remember HFT firms worship at the alter of the law of large numbers, they make fractions of a penny per share traded but only do it if they have an edge ie 90% upside to 10% downside. There just isn't any edge to be had in the above scenario, its just a heaping pile of risk.