Algorithmic Trading and DMA: An introduction to direct access trading strategies

Terminology time: when the average amateur thinks of "Algorithmic trading," he thinks of vast machine intelligences duking it out in microseconds using exotic signal processing techniques. Well, in the business, "algorithmic trading" generally refers to the process of finding liquidity for an instrument using a computer, generally done by a buy side trader. This may seem needlessly pedantic, but it's important, as this is an actual job description, and this is what the book is all about. It also relates to 2009's favorite whipping boy, "High Frequency Trading," and could be considered the premier book on this subject -it's the best one I've yet read anyway. In addition to describing the other end of a HF trade done intelligently, it describes how various arbitrageurs and other prop traders make their money (in ch 13 in particular). The appendices are also excellent, and there is a useful key to abbreviations and acronyms: something sorely missed in many other books. The book is a model of clarity and trading didactics; I have read no better description of this sort of thing, anywhere. While I'm not qualified to say so, as I don't actually do such things for a living, I suspect it's extremely complete and accurate introduction to the subject. In addition to the didactics, it contains plenty of folk wisdom, practical advice, obscure information and good old horse sense. In detail: For part I, ch 1 gives a basic overview of the subject, including necessary definitions (aka DMA versus algo trading versus ...). Ch 2 touches on market microstructure; this is excellent, both for the rank amateur, and the professional looking to be grounded in a clear exposition. Ch 3 a description of the different types of markets, asset classes, dark pools, dealer markets and etc. This is all basic stuff; the nuts and bolts of what we're talking about. On to part II; ch 4 gives a detailed description of the different order types one can use in different markets. Ch 5 gives the basic kinds of trading algorithms; VWAP, implementation shortfall and all that. Ch 6 is on the process of modeling transaction costs; this chapter doesn't give any algorithms for doing so; it is more of a framework for thinking about the problem from the point of view of the algorithmic trader. I originally thought ch 7 was one of the weaker chapters, though upon reflection it may be one of the most useful ones for assessing market behavior; I was focusing on the classical use of the word "optimal." Section III, chapter 8 order placement is a sort of review of market microstructure models of price formation, and a strategic break down of the way a trader thinks about the problem of order placement. That and the sections on dark liquidity: gold. Ch 9 is on tactics; also invaluable stuff. How does a trader fake out other traders, look for hidden liquidity, update the limit book to minimize signaling? Ch 10 can be seen as a collection of ways to use forecasting techniques in your trading algorithms. Lots of practical information mixed in here about "forecastibles" that everyone knows about (dividends, witching days, etc). It's not always obvious how to incorporate known future events into a trading strategy or algorithm: this chapter is very helpful. I'd have liked to see it done in some kind of Bayesian framework, but whatever; this is really practical, useful stuff. Ch 11, infrastructure; this goes over things like FIX (the protocol for talking to the broker), some graphs as to how the actual order process works, ideas on latency, testing, market compliance and so on. I'd have liked a little more information on things like tick databases, trading platforms and trade resolution infrastructure, but maybe such information would be out of date as soon as he wrote it. Anyway, mentioning the words and some problems with typical such software might be useful to the tyro. Part IV, Ch 12 is on portfolio trading. There is a decent introduction to classical portfolio theory, and some good ideas on minimizing portfolio risk using the author's "marginal contribution to risk" metric. I'd have liked some more information here, but perhaps this is an appropriate chapter for a book more or less on Algo trading and DMA, rather than prop trading. Ch 13 is on various multi-asset trading strategies; roughly speaking, forms of statistical arbitrage and prop trading. It's sort of an oddball chapter, as this isn't the primary subject matter of the book, but it's a topical subject, and I'm glad it's there. Ch 14 is on trading the news; also a very interesting topic, and subject of ongoing research. The author gives a lot of practical information here which could be useful to the experimenter. Chapter 15 on machine learning is probably the weakest of the book, though I can find no factual faults with it. It is a reasonable introduction to machine learning and data mining techniques. Personally, I'd have lost much of the stuff on ANN's, and added a bit on reinforcement learning, and perhaps a section on the block bootstrap (one of my favorite hobby horses) used for testing for overfitting. One idea I found really interesting was the notion of using artificial stock markets to test ideas. I've fiddled with these, though I never thought of using them to test ideas! That's a damn good idea. Honestly, I think most of the machine learning papers out there are crap, and their appearance in books like this are more or less smokescreens. Stuff like econometrics and particle filters: way more useful. Don't tell anyone I told you so. The 70 pages of appendices, well, they're all super helpful for figuring out how the actual markets work in detail. No doubt some of the details are already out of date, but the over all structure: priceless. A real map of world markets. Don't know why he wrote it, but I'm glad he did. I'd have paid twice the cover price for the book. I've actually physically worn the thing out (it doesn't do well on beaches), and will probably order another copy.

Barry Johnson's book is a great introduction to electronic and algorithmic trading. The book is so well written that you will find yourself reading it like a novel. The contents are well chosen and the chapters are fairly balanced in terms of length and importance. The approach used in the book is very pedagogic. The author illustrates each and every trading strategy with an example and a figure, which permits to clearly grasp the motivation, the intuition and the ideas behind each trading scenario. He takes a good time explaining the variables that determine prices, liquidity, market impact and volatility. He also provides a lot of references for the readers willing to go deeper on a specific topic, and the summaries at the end of each chapter are an excellent addition. In my opinion it is far better to understand the mechanisms of trading in today's electronic environment than just learning ready-to-use recipes. It is indeed the ignorant use of financial instruments that is at the genesis of the current crisis. Therefore, the author has my full admiration and support because he manages to provide a full understanding and grounding of algorithmic trading. However, I have to put only 4 stars for the following reasons. Algorithmic trading is crucial today not only because it is far more reactive than human traders, but also because it can predict and exploit trading patterns more accurately. Unfortunately, the book only has a subsection on forecasting market conditions and short-term prediction of prices, trading volume and volatility. Each one of these topics deserves a full chapter because they are the main reason why we are switching from human trading to algorithmic trading. Another topic that is crucial for algorithmic trading is arbitrage. Again, the book falls short, just adding a subsection on the topic. Moreover, the author cites a result that seems to show that implied volatility is a more accurate measure than statistical volatility such as GARCH. The empirical evidence says the contrary, i.e. that GARCH and other volatility measures like bipower variation predict better than implied volatility, in particular for high frequency data. This feature is in fact exploited by quantitative hedge funds and proprietary trading desks. The author also skips the stylized facts from the empirical analysis of financial time series: returns are not normal and exhibit high peaks, fat tails, auto-correlation and volatility clustering. It is the evidence of these facts and the necessity to understand and control them that has given to Finance the mathematical and computational trend it currently has. My suggestions for mathematical references are the following classic books: A. Shiryaev Essentials of Stochastic Finance: Facts, Models, Theory P. Embrechts Quantitative Risk Management: Concepts, Techniques, and Tools (Princeton Series in Finance) ****************** EDIT Nov 23, 2013 ****************** Concerning references on market microstructure with other flavours than mathematics and computer science, I would recommend the following two classics: For a trader's point of view, L. Harris Trading and Exchanges: Market Microstructure for Practitioners . It takes you to the basics, from brokers to traders to Limit Orders to Market Orders to market-makers to electronic markets. I would recommend to read this book before Barry Johnson's. For an economist's point of view, M. O'Hara Market Microstructure Theory . It does explain in detail several classical models based on inventory risk and asymmetry of information.

It is incredible how much knowledge is packed in this book. It is well researched, well written, and certainly enlightened me on a topic that intriged me for a long time. You may never get into algorithmic trading & DMA but algorithmic trading & DMA are getting into everything that you (or your financial advisor) do in the market. Additionally, the proceeds from this book are going to a good cause. You realy have nothing to lose at all. To the author, thank you for a quality work.

This is a good book with a lot of details. Over the past 10-15 years, Quantitative OR algorithmic trading has been further researched. Different researchers have come up with algorithms for various strategies. Barry Johnson has done a good job of aggregating the strategies and algorithms. One need not have an advanced knowledge of financial instruments to follow this book. Since he himself was a software engineer, he provides details about the software side as well. I have not fully read the book. Plan on doing so pretty soon.

Although the book is more than 10yrs old, it still provides a great overview of the systems and algorithms used in the industry today. Highly recommend for anyone interested in electronic equities trading.

It's really annoying to read all those references embedded in the text. The author looks like a student who gets a point for each reference in this dissertation. Otherwise the content is good.

Brill HFT Book - a must read if you want to know about trading

great

Good all round introduction