Bests & Worsts Reviews from Amazon

Disclaimer: I have a lot of experience trading electronically (although I don't do HFT), and I try to keep up with everything HFT-related, exchange market micro structure, nooks, academic articles, news, SEC/CTFT/FINRA meetings, etc. Flash Boys presents a story about a trader, Brad, who's not technical, and who's trying to learn about the market. The trader find out about a new kind of investors, the high frequency trades, and Brad "figures out" that speed is very important and that his trades are being gamed. Enraged by this, Brad decides to create an alternative trading system, ATS, that would be more "fair" to the long term investors. The book tries to accomplish 2 things: 1. tell a story in an entertaining way 2. raise flags about how to stock market is "rigged" On (1), Michael Lewis does an ok job, although the story is forced. Some chapters have very little relation to each other. (the Sergey Aleynikov chapters for example) However, Michael Lewis is not the first to this story of HFT, and in particular Scott Patterson has a more entertaining story in his book "Dark Pools". Also, the book's entertainment value is much lower than some of Michael Lewis' other books, in particular Liar's Poker, Moneyball and The Big Short. The second part of the book is dry and it reads like a commercial for IEX. (the new ATS started by Brad) On (2), Michael Lewis has too many wrong facts, and basically he loses credibility because of this. Moreover, the whole book is only focusing on the speed aspect of HFTs, and latency arbitrage, that's only a small part of HFTs and market making. In addition, he only presents facts that support his point and many facts where they are blatantly false. Next, I will go thru some of these facts presented in Lewis' book False facts presented in the book: 1) On the first page, Michael Lewis sets up Brad as a trader working for RBC, a very conservative bank -- "It was stable and relatively virtuous, and soon to be known for having resisted the temptation to make bad subprime loans to Americans or peddle them to ignorant investors" Yeah, right... this is an actual misconception that is seen from time to time in the news. The Canadian banks, including RBC, received 114 billion dollars in bail-out money (which is more per capitalization than the American banks), but their bail out was secret. The people only found out in 2012, and this is why this misconception exists. Michael Lewis should have known about this. 2) On the second page, Lewis continues to set up the Canadians (and Brad, the hero of the book) as being super conservative "Debt was a foreign concept in Canada. Debt was evil". This is pure BS. As of March 2014, the household debt-to-income in Canada is 165% and about 105% in America. Their real estate is in big problems right now, and may burst soon. Canadians are, on the contrary, in more debt than Americans. 3) Lewis complains several times that the are some exchanges that pay traders to take liquidity. He says: "For instance, the BATS exchange, in Weehawken, New Jersey, perversely paid takers and charged makers." "Why would you pay anyone to be a taker?" At some point he mentions SIRI, and how after the BATS with the new reverted fees, the volume in SIRI tripled. He claims this is proof of high frequency trading taking fees. I hope Michael Lewis is reading this review. He's why there may be a competitive advantage in reverted fees and in particular for SIRI: basically, when you execute an order, your cost is going to be price+rebate/fee. if you have a bit at 100.14 and ask at 100.15, including the regular rebate/fees, the price would be roughly 100.1370 and 100.1430 assuming a fee/rebate of about 0.0030, which is roughly what Nasdaq has. Now, for a reverted fee exchange, these prices would be different: 100.1430 and 100.1470. What is important to notice is that the actual spread is much smaller in a reverted fee configuration (0.0040 vs 0.0160). SIRI is an extremely cheap stock -- a penny stock where the spread is huge by regulation. With inverted fees, the reduction in spread is very important, because it reduces the costs of trading. Less costs, more activity, more trading. 4) Michael claims there is no paper/trade/quotes trail: "You want to see a single time-stamped sheet of every trade. To see what followed from what. Not only does it not exist, it can't exist, as currently configured. No one could say for sure what caused the flash crash -- for the same reason no one could prove that high-frequency traders were front-running the orders of ordinary investors. The data didn't exist." Actually, the data does exist, and is recorded by many institutions on every little details. You can buy level 2 data (which includes every order on the public exchanges, places, cancellations, repricings, executions, etc) from Bloomberg, Reuters, 7 ticks, FTEN, and many others. You can actually look and see what happened during the flash crash. You can actually check when your orders get on the book, etc. 5) "Back when human beings sat in the middle of the stock market, the spreads between the bids and the offers of any given stock were a sixteenth of a percentage point." Wrong, the tick was 1/8 of a dollar, and in reality it was 25 center (1/4 of a dollar), because the human market makers had a gentleman's agreement to keep the spreads high. Higher spreads, higher profits for the market makers. This gentleman' agreement was a big part of the reason why the SEC force the market to go electronic -- because the old human brokers and market maker were cheating the investors. 6) "There used to be this guy called Vinny who worked on the floor of the stock exchange, said one big investor who had observed the market for a long time. After the markets closed Vinny would get into his Cadillac and drive out to his big house in Long Island. Now there is the guy called Vladimir who gets into his jet and flies to his estate in Aspen for the weekend. I used to worry a little about Vinny. Now I worry a lot about Vladimir" -- The implication here is that Vladimir makes more money than Vinny, and the investors must be paying more to the market makers. But the electronic market making scales easily, and to start there's 1 Vladimir for 1000 Vinnies. 7) "The price volatility within each trading day in the US stock market between 2010 and 2013 was nearly 40 percent higher than the volatility between 2004 and 2006" You need to be very precise when you define volatility. Usually people refer to the VIX implied volatility, and you can see the graph historically in yahoo finance. Take a look. Can you tell that the market now is in a different state? No, you can't. There were some higher volatility periods, but those periods were driven by big macro economic issues: 2008 crisis, 2011 debt ceiling debate and the following debt ceiling debates. In fact, the VIX has been close to historical low in 2013. 8) "It was what investors most noticed: They were less and less able to buy and sell big chunks of stock in a gulp" As market makers compete and get better at reading the market, they will provide a tighter spread. However, at the tighter spread they will not be able to offer as many shares as at a wide spread. Cheap stocks for which the spread is big (1 tick is a bigger percentage of the price) have long queues, and expensive stock will have very thin queues. SIRI probably has queues in the millions of shares and Google in 100-200 shares range. This makes perfect sense since a price in smaller increments (percentage wise) is much easier to ran over. When the market maker is ran over, he loses money. Why is this relevant? The decimalization, going to 25 cent spreads to 1 cent spreads has the same effects. When the spread was 25 cents you can offer a lot, but on 1 cent spread, you'll lose your shirt quickly. This is why the mutual fund execution should change when the tick size is small -- you just can't do the entire quantity in 1 shot anymore. Also, these guys are supposed to be professionals and know how to execute. 9) "A Hide Not Slide order was a way for a high-frequency trader to cut in line, ahead of the people who'd created the line in the first place, and take the kickbacks paid to whoever happened to be at the front of the line" Actually, anybody can use a hide not slide, not just HFT. In particular, hide not slide reprices automatically, and it levels the plain field between fast and slow investors. If used, this order is good for the slow investors. 10) On IEX, "the obvious starting point was to prohibit high-frequency traders from doing what they had done on all the other exchanges -- co-locating inside them" With co-location, you can have hundreds of investors, all at exactly the same distance, with a leveled plain field. Without co-location, the investors will fight for the spaces close to the exchange -- across the street, a block away, 2 blocks away, etc. The plain field will not be leveled and their costs will be higher, plus they will have to maintain their own little datacenter. This is bad -- it's going back to a worse time. 11) "What the **** is the point of a Post-Only order?" On most exchanges when you take liquidity you pay a fee, with post-only you get a rebate, so in effect an order at 100.10 post only would be an order priced at 100.0970 and not post-only priced at 100.1030. So the answer is: they have a different effective price. 12) "For instance, one day, investors woke up to discover that they'd bought shares in some company for 30.0001 Why? How was it possible to pay ten-thousandths of a penny for anything? Easy: High-frequency traders had asked for an order type that enabled them to tack digits on the right side of the decimal, so that they might jump the queue in front of people trying to pay $30.00." Completely false, pure fabrication. The exchanges can only trade in tick increments. (modulo the fees/rebates which are fixed for the exchange) This is against the SEC rules. 13) "In August 2013, the Goldman automated trading system generated a bunch of crazy and embarrassing trades that lost Goldman hundreds of millions of dollars..." Yes, but this was on the Options market, on Options exchanges, and has nothing to do with the rest of the book. 14) "They played only when they had an edge" All investors "play" when they have an edge. No investor plays to lose. 15) Brad's claim that when he did a sweep to take the liquidity on all exchanges, it first hits BATS, and the high frequency players can see that and take the liquidity from the other exchanges, then selling it back to Brad at a higher price -- it's very fishy. Why? Because if it was the case, you can easily game it. Put sell order at the ask, the hit BATS with buys, forcing HFT to buy your sells, in effect forcing sells at the ASK. Similarly, you force buys at the BID and make the spread. Moreover, this conjecture can be easily proved by looking at the L2 data after the failed sweep. Did somebody rush in and take that liquidity in front of you, or were those orders cancelled? Trivial to check. Did Brad and Michael do their homework and actually check this before making this big allegations? Since they didn't bother to actually check with the L2 data, it's hard for me to take them seriously. At last, what bothers me is that the second half of the book is really a marketing material for IEX. I went to the IEX website, and here's what I found about how to prioritize orders: "Among all orders at a given price, precedence will be given first to orders marked Agency or Riskless Principal, in time priority, and then to orders marked Principal, in time priority, belonging to the same Subscriber as the order being processed, and then to all other orders at that price in time priority; the oldest orders having the higher precedence." In other words, their broker friends and other friends (agency and riskless principal) can jump the queue in front of the average investors. It looks to me that the system is set up to screw the investors, not to help them. But the edge is given to their friends. This is much worse than the lit exchanges, where there's just time-priority and nobody has an advantage. This is disgusting. Conclusion: The style of the book is somewhat entertaining and the book is more than readable given the dry subject. Michael Lewis presents a lot of misconceptions and wrong facts about the market. His main point is that there's a particular case when somebody rushes to take all the liquidity at the best bid/ask, but can't because of fragmentation. He claims that HFTs are faster and after you take the liquidity in one exchange, HFTs can "front-run" you to the other exchanges because they are faster. Sounds believable, but the main problem with this assertion is that it can be easily checked by looking at the L2 feed -- in fact this would be the first thing to do to debug the problem. Michael and Brad don't do that. Instead they claim the data is not available, which is clearly a lie (real time and historical data is available in great detail) Further, today, the investors are using the same technology as HFT, such as co-location and fast feeds. E-Trade and Ameritrade flow is executed by Knight/Getco, Virtu, and a couple more aggregators, which are just as fast and sophisticated as HFTs. Moreover, Michael promotes Brad's "solution" to the market, IEX. IEX's innovation is a delay on the feed of 350 microseconds, a speed bump, so that HFT's don't have an advantage. Sounds good at first, but at a second look it resembles a marketing gimmick. Say you put an order on the book, if the market moves and it's clear that you're behind, you would try to cancel the order, but since you'll be slowed down to, there's still a race between you and all other market participants to the stale order. If cancels are delayed less, then you have another problem: the HFT now has the order on the book and he can cancel easier, so he'll only clear when it's more convenient for him. In any case, the relative speed difference between players will be the same, and reacting fast to the market will be just as important, only that with the 350 microsecond delay, now you have another dimension to consider, and the complexity goes up. A smarter investor will be able to navigate a complex market better than a less competent investor. HFTs are the smartest on that time-frame. Looks to me that they will benefit from this "innovation". ----------- A lot of commenters asked more details about high frequency trading. There are a lot of misconceptions and I think this book promotes some of these. Here is what it's known about HFTs: - HFT profits were estimated at 1.3 billion dollars in 2013, down from over 7.2 billion in 2009, according to Larry Tabb, consultant, who has been vocal against HFT's in the past. To put this in perspective, the profit of a single big bank is up to 20 billion per year. - HFT main competition are the old school brokers who are now extinct - HFT drove the spreads down by a factor of 25 since decimalization. The low spreads means low margin per trade, which means human specialists can't compete anymore. - many HFTs have a broad view of the market and react very fast - HFTs use many strategies. The common ones are all sorts of arbitrages and market making (providing liquidity with investors that don't want to wait in the queues) - HFTs have a small edge per trade, but they do many many trades. Because the law of large numbers, the result is that they generally have high sharpe (risk adjusted returns). The way they provide liquidity either works or it doesn't, and they can figure out if it works very quickly. If it works, they will rarely lose money. - the risk of HFTs is that (1) they are not competitive anymore (somebody can read the market better and provide tighter liquidity, ie smaller spread) and (2) computer error (such as the trading error Knight Capital had, that lost 440 million dollars) - some HFTs are still profitable. Virtu comes to mind. However, if you look where they make their money, they haven't grown in their established markets. - some HFTs are much less profitable. Getco, according to SEC filings (it's now a public company), loser 95% of the profits recently. - some HFTs had to close doors. Eladian Partners was a spinoff the trading shop inside Citygroup -- their technology used to work, but they became not competitive enough and closed. Goldman bought the market maker Spear, Leeds & Kellog in 2000 for 6.5 billion dollars. Today they are selling it for 30 million. Tradeworx used to be a prop shop, but now it makes money by consulting for SEC and handling orders for retail. - HFT are good at reading the market, especially bad execution, like what Brad was doing in the book. This forced the brokers and investors to obfuscate their orders. Now, they all use algos to trade. - when mom&pop trades in the ETrade account, they benefit by the small spread. They get ripped by the broker fee. It's funny that people talk about the $0.01 spread (and an HFT only makes a small fraction of that on average) but ETrade charges around $7-8 per trade. - HFTs are not banks. Banks lose to HFT because of their culture. Technology people don't like working in the bank environment. Not even Goldman Sachs can keep up. ------------------- The biggest problem with this book is that it attacks the wrong properties of the market. For example, it attacks co-location which creates a leveled playing field, as everybody buys space at the same distance from the exchange. Before co-location, brokers were buying real estate around the exchanges, and nobody was equal. This was also very expensive, and gave advantage to brokers with a large bank account (mainly the banks like Goldman). Another big problem is that this book focuses on the speed issue. It's never been a better time than today, when more brokers and investors were connected at the same speed. Putting the speed issue forward, the book barely mentions the practice of some big market makers to buy flow from E-Trade, TD Ameritrade, etc. This has nothing to do with HFT, and this practice is orthogonal to HFT issues. While from a business point of view this is ok (it's a free world), the practice of hiding flow hurts the market and price discovery. Every single study that has been done on this shows that price discovery and other market properties are being visibly hurt even if only 10% of the flow is in the dark. This is why Australia, Canada and now Europe have put the brakes on the dark pools. These aggregators are brokers that specialize in internal matching and together with dark pools use lobbying to keep their advantage: dark pools are allowed to match at a price under a tick size, while exchanges are not; This is a regulatory advantage because they can market their service by saying that they offer a better price to retail, also called "price improvement". Moreover, some of the complexity of the exchanges is to go around this problem. Since their price levels are fixed, the only way to change the price level is do offer different fees/rebates (Bats opened 2 exchanges with different feeds, and DirectEdge also 2, and Nasdaq has 2 as well) and exotic orders (that have different fees/rebates).