logo

Chrilly's Book Reviews

The number of really good finance books can be counted on the fingers of one hand. At least if one counts with the Chinese-number-gestures. In comparison there is almost an infinite amount of junk. The chance of picking randomly a good trading book is in an epsilon around zero. Once upon a time one could filter out the bad books by looking at the publisher. A John Wiley book was not always excellent, but it was never really bad. In finance it's the other way round. A John Wiley-Finance book is almost always junk and only in exceptional cases worth the money. It's not only the content which baffles me. In Pairs-Trading one reads of Norbert Weiner, a Weiner-process ... I don't blame the author. He certainly knows, that it is a Wiener-process. He has certainly checked his manuscript several times. But an author is the least-qualified person to find errors in his own work. It's the job of the proofreader to spot such annoying mistakes. But no one at Wiley-Finance ever reads a single line of Wiley-Finance books. It's all about marketing and selling, but not about producing good books. The situation is not better at other publishers. I just mention Wiley-Finance, because Wiley was once a flagship in publishing.

The books are divided into 3 categories:

In case of junk, please believe me, it's really bad. The Weiner book is under Maybe. Save your money and the trees and don't buy these books. If you speak German, the extensive list of book-reviews (not just finance) at Chrilly's-amazon.de-Reviews could be of interest. For further differentiation within the 3 groups the books are rated from 0 (terrible bad) to 10 (ingenious).

  1. Euan Sinclair: Volatility Trading. Rating 9
    Most finance books are either full of measure theory or the author avoids any mathematics at all. This book belongs to the rare species which follows - in my view - a reasonable middle ground. The author filters from the finance-paper glut the practical relevant topics and presents them in a comprehensible way. E.g. there are hundreds of papers about measuring volatility with high-frequency data. The mathematical properties of these methods are very nice. I have played around with these methods. The practical benefit was humble. In real life there exists microstructure, overnight jumps and a strong seasonal daily pattern. The author states about these models: "So as a long-term forecasting tool, using high-frequency data is possibly the wrong approach". This is in line with my own experience. The author clearly states: There is no silver bullet. All methods and models have pros and cons. He lists explicitely the good and bad points of each major method. Options (and OTC variance swaps) are the traditional way to trade volatility. With the introduction of VIX-Futures and VIX-based ETNs a new and very interesting volatility playing ground has been created. The 12th chapter deals with VIX-Trading. I liked especially the subchapter about VIX based ETNs. The author covers a very interesting model of Ch. Donninger :-) Generally this part of the book can be expanded in future editions (the easiest way is just to follow the working-paper deluge on this site) . The book can not be recommended for absolute beginners and mathematics-grumps. But it is by far the best book I know off for experienced traders and quants. It has a clear leitmotif and is well written.
  2. Carol Alexander: Market Risk Analysis, Vol. I to IV. Rating 8
    This series is one of the few impressive books in finance. It is certainly the best introduction to the field. In Volume III and IV also advanced topics are covered. Most quantitative finance books are written by mathematicians who are not really interested in finance. Finance is - in the bast case - only used to illustrate mathematical concepts. In these books mathematics is a tool to solve financial problems.
    Volume I, Quantitative Methods in Finance.
    is an introduction in the basic methods like Linear Regression, Numerical Methods. The only directly finance-related chapter is the last one about Portfolio theory. If you have a good background in statistics and numerics you can skip this volume. Personally I have this background and skimmed over most topics. But it's really a nice intro into the finance toolbox.
    Volume II, Practical Financial Econometrics.
    presents advanced statistical methods which are directly related to finance. The central theme is correlation and volatility. There is an outstanding chapter about GARCH models. My favourite chapter is the introduction to Copulas. The discussion is balanced. Copulas are presented often as a panacea. The book mentions also the pitfalls of this method.
    Volume III, Pricing, Hedging and Trading Financial Instruments.
    discusses the essential models of the fundamental asset classes: Bonds and Swaps, Futures and Forwards, Options, Volatility. The last chapter is about portfolio mapping. Carol Alexander treats Volatility as an own asset class. The series is generally heavy on volatility. Personally I also think that volatility and not prices is the interesting point in financial markets. Or to put it in another way. One can't model prices. All the stuff of the so called technical literature is voodoo. But one can model volatility. There are now also a lot of easy to trade volatility products like the VIX related ETF's.
    Volume IV, Value-at-Risk Models.
    discusses the different approaches to model risk. Risk-modeling is an attempt to avoid a frontal crash by looking in the back-mirror. There is no silver bullet. The book presents the best methods to tame this beast.

    The books have an exceptional content, are well written, have an appealing layout and graphics. But they do not deserve 10 points for one simple reason. Modeling is done in Excel. I don't understand why Carol Alexander did not use Matlab. Matlab is for an hold hacker already a mollycoddle language. But using Excel is really disgusting. I can't give an Excel book 10 points.

  3. Manfred Gilli, Dietmar Maringer, Enrico Schumann: Numerical Methods and Optimization in Finance. Rating 9
    Before entering the quant business I have build the chess supercomputer Hydra. I expected that Hydra is a disk-calculator in comparison to the systems used in finance. To my ecstatic terror I found a strange mixture of measure-theory, Excel-sheets and trader voodoo. As an old hacker one gets the impression: The computer is not yet invented. This book is different. The authors state in the introduction. We don't know much in finance. No, we are not modest here. "We" does not only refer to the authors, but to finance and the people working in finance in general ... The perfectionism that numerical analysts employ to refine methods is not just unnecessary in finance, it is misplaced

    Instead of thinking about an approximation of the 4th degree one should better spend the effort to improve the model. Instead of using the most arcane mathematical tricks to develop analytical solvable models one should develop models which are of practical use.

    To follow these guidelines one must master the art and handcraft of programming. The authors are without doubt experienced Matlab hackers. It is usually straightforward to write running Matlab code. But it is an art to write efficient routines. One has to vectorize the model. Vectorized Matlab code can almost match with real programming languages like C. The authors deal also with such mundane but essential questions like the quality of the Matlab random-number generator for normal distributions. The statistical properties of this generator are surprisingly bad. I hate books, were the author lists any method which was ever proposed in literature, without giving a hint which is the best. Such authors don't take the risk to be proven wrong. Gilli et al. are in contrast very brave. They clearly state: We have good practical experience with this simple method (e.g. threshold-accepting for heuristic optimization) and we advise you to use it. Other authors mention all sorts of random-number generators. Although there is currently a clear winner. The Mersenne-Twister. This book gives the clear advice: Use the Mersenne and forget the rest. Other books contain numerous examples. But none is explained well. "For further details see Ref. X". In this book four representative problems are treated with scrutiny. E.g. in the last chapter the parameters of the Heston- and Bates- options model are calibrated. With the sobering result: The vol of vol parameter converges only very slowly, the jump-component in the Bates-model does not converge at all. The time-series must start at the big-bang to get reliable results.

    Having detailed knowledge of numerical methods is no prerequisite to follow the book. But one needs some programming experience and some understanding of trading. If one does not know the purpose of option models, the book is of little use. But after reading Carol Alexander this is the best follow-up I am aware of.

  4. Paul Embrechts, Claudia Klueppelberg, Thomas Mikosch: Modeling Extremal Events: for Insurance and Finance. Rating 7
    This is an excellent mathematics book. But it is not really about modelling. It is without doubt the most detailed monograph about Extreme Value Theory. It is perfect as a starting point for a mathematical excursion of this vast and important field. But a model is usually a simplified concept of reality. Reality plays in this book a minor role. It is only used to illustrate mathematical concepts. The applied part of these 600-pages book can be easily condensed in a 20-pages paper ( Applying Extreme Value Theory to Finance). The authors stress in the preface, that they have developed the concept of the book with practitioners from insurance and finance. I doubt this claim. The book is for practically working people much too academic. But it is from the academic point of view an impressive work. If the title would have been "The Mathematics of Extremal Events" I would have rated it with 10 points. But with the current title it is somewhat off topic.
  5. Euan Sinclair: Option Trading: Pricing and Volatility Strategies and Techniques. Rating 8
    The author has studied theoretical physics and has changed then - as usual - into the quant business. This is a problematic background for a trader book. Many of this wannabee Einsteins write books just to show how ingenious they are (see Wannabee-Einstein). Sinclair is different. He just wants to demonstrate that he can write and explain well. The target audience are readers with some practical experience in stocks or futures trading. The mathematical requirements are low. The author simply knows the fundamental publishing rule: Each formula halves the sells. This constraints the level of complexity. E.g. the calculation of volatility is restricted to daily data. One can improve the accuracy with high-frequency data. Using HF-data is not without its own problems. Individual investors usually also have not access to HF-data. Sinclair avoids the complicated discussion about the pros and cons of HF-data by skipping the topic at all.

    The book starts with a gentle introduction to options. The author gives some nice historical examples for options trading in ancient time. The history of options trading starts long before 1973 and Black-Scholes.

    The pace is increased from chapter to chapter. Personally I have skimmed over the first ones. But in the later chapters I have studied the book very carefully. There are a lot of nuggets to discover. I have especially liked chapter 10 about Market-Making-Techniques and chapter 11 about Volatility Trading. If you look at the Working-Papers section you will find several Volatility Trading ideas. Chapter 11 was the original inspiration for these efforts.

    The book is not perfect. Sometimes the author stops when it gets complicated and interesting. Sometimes he avoids complicated topics at all (e.g. High-Frequency-data). But it is a well written and informative advanced introduction to the miracles of options and volatility trading.
  6. Gonzalo Arce: Nonlinear Signal Processing: A Statistical Approach. Rating 9
    I have developed two strategies which use the VIX futures term structure as trading signals. The trembling of the invisible hand was smoothed with a median-5 filter. This simple idea increased the performance considerable. I have known from the image-processing literature that median-filters are used to remove salt- and pepper-noise. It was evident that the same can be done for short-term market overreactions. On occasion I decided to study this type of problems more systematically.
    This book was for this task an excellent choice. The author starts from the family of stable-distributions. He analyses the behaviour and the optimality of smoothers and filters in dependency from alpha. From 2.0 to 1.7 linear filters are the best choice (they are in the strict mathematical sense only for alpha=2.0 aka the normal-distribution optimal). Median filters excel between 1.7 down to near 1.0. For even lower alpha (fatter tails) the Myriad filters are best. Median filters are strictly speaking only for the Laplace-distribution optimal. But the book does not restrict itself to rigorous mathematical results. It gives also practical advice of the sort presented above. The author covers several important refinements to the Median- and Myriad filter approach. One can use them - like in the linear case - not only as smoothers, but also as high- or mid-pass filters. The effects are nicely demonstrated with image-processing examples.
    The only drawback is from my point of view: All the filters are symmetric. For filtering the point t one uses the points t-k,...,t,...,t+k. This is for image processing and other applications the natural choice. For trading purposes only strictly asymmetric filters are applicable. One can't wait for tomorrows result to trade today. An asymmetric smoother introduces always a lag. I don't know of any literature which covers this sort of filtering problem in detail.
    Included is a CD with a collection of 60 MatLab routines. The book is well balanced between mathematical rigour and practical considerations.

Maybe of interest

  1. Brian Johnson: Option Strategy Risk/Return Ratios: A Revolutionary New Approach to Optimizing, Adjusting, and Trading Any Option Income Strategy. Rating 6
    B. Johnson is according "About the Author" an old investment fox for income strategies. He is now an independent who lives from selling investment strategies and advertisement for other products. One could consider the book as a sort of advertisement-folder for his activities. But the book is besides the lurid headline "A Revolutionary New Approach" quite solid. There is no promise of "getting rich in 21 days".
    The first chapter is a general intro to option income strategies. Next are the greeks. The author avoids almost all of the mathematics. He advises the reader to use OptionVue instead. The 3rd chapter develops criteria for a reasonable risk/return ratio. Chapters 4 to 6 are about his own measures Delta/Theta (DTRRR), Vega/Theta (VTRRR) and Rho/Theta Risk-Return-Ratio (RTRRR). The ratios are simple linear functions of the risk Delta+Gamma, Vega and Rho. Theta measures the return. The derivation of these trivial formulas is rather redundant. B. Johnson spends 55 pages on a topic which can be easily explained in 2 pages.
    In the following chapters he analyses these ratios for Iron Condors, Calendar Spreads, Iron Butterflies, Double Diagonals and a hybrid combination of these basis strategies for RUT (Russel-2000) options on 18. April 2013. These chapters are also very redundant. Instead of repeating the same stuff over and over again he should have analysed his measures for different market-regimes. Sticking to just on day is not very enlightening. His Risk/Return Ratios are too simple to model the non-linear dynamics of options. It is not useful to separate Delta and Vega Risk. There is a direct but rather complex relation between these two Greeks (GARCH models are an attempt to model this relation).
    The last chapter addresses these shortcomings. According the author one can avoid some of the problems by using the "True Delta, Vega ..." functions of OptionVue. This improves somewhat the Greek estimates, but it does not solve the principal shortcomings of his approach. He presents in very general terms how one could improve and optimize the model. I have done a similar research. I don't believe that B. Johnson has really implemented such a model. His programming skills seem to be restricted to Excel-Sheets (which is in my view not even a programming skill). It is not possible to implement in Excel any sophisticated model. He has probably also not the necessary mathematical background.
    B. Johnson sells on his homepage a simple asset allocation strategy. If he would have done something similar for options he would try to sell this too. At least he would give some performance measures. The rating of the book is a matter of taste. One could publish the contents of this book in a 10-pages paper. But it is - in contrast to many trading books - not complete idle talk. The author does not promise the moon. He adresses the right questions, his answers are too simple but they are not wrong. Reading the book was an incentive to adress this question myself once more.
  2. Colin Bennet: Trading Volatility: Trading Volatility, Correlation, Term Structure and Skew. Rating 6
    The author is according the verso a top quant. "Colin Bennett is a Managing Director and Head of Quantitative and Derivative Strategy of Banco Santander. Previously he was Head of Delta 1 Research at Barclays Capital". There is no doubt on his high professional qualification. Nevertheless I found the book somewhat disappointing. Everything is written in an enpassant style. Bennett shows the reader his memos, but nothing is explained in more detail, with more a ccuracy. E.g. He remarks that the term-structure of VSTOXX-Futures is different to VIX-Futures. There is less market pressure from ETFs and ETNs like VXX. Hence it pays to go VSTOXX-Futures long, VIX-Futures short. End of memo. There is no analysis of the term-structures, no trading-strategy is specified and hence no performance measures are given.
    I have done this detailed analysis: The term-structure is indeed flatter for the VSTOXX and also the term-structure of the Futures Beta is somewhat different. For VIX-Futures beta(maturity) = 0.99 - 0.15*log(maturity) gives a good approximation.
    For VSTOXX 0.95- 0.10*log(maturity) is more reasonable. The fit of this function is generally for the VIX better than for VSTOXX.
    The long VSTOXX- short VIX-Future strategy performs reasonable till end of 2012. But in the last 2 years the performance is for a naive model (always trade the 1st future) rather poor. I have delevoped a model based strategy which performs somewhat better. But there are a host of other VIX-Futures and Option strategies which have by far a better performance. Bennett has obviously not checked his memo on the latest available data. This lack of craftsmanship can be seen at first glance. The layout and the graphics doesn't meet a professional standard. It was not edited by a publisher like Springer. It's home-made and printed by Amazon.
    The head of Eurex-Marketing praises the book on the verso with "The book could be seen as the volatility bible!" It could have been a bible, if the author had devoted more time and energy for writing a solid and well crafted book. In the current state it is a loosely sorted collection of memos.
  3. Paul Wilmott: Frequently Asked Questions in Quantitative Finance. Rating 6
    The thesis of P. Wilmott was about the diffusion of waste water in a river. This is mathematically equivalent to Black-Scholes. Waste water Pauli has then changed to the more fashionable quant-business and plays there the motley bird. In the meantime he has become the head of a whole motley crew. The Wilmott-Magazine is a prototype of scientainment. Wilmott has a good sense for fine sarcasm. It's fun to read his small pieces as a bedtime story. He makes especially fun of the mathematical overkill in many academic finance papers. He has coined the term "Measure-Theory-Police". There is no doubt that many academic papers are written in an absurd way. But the world of traders is mathematically very simple minded. They draw with a ruler some resistance lines and do some technical voodoo with Excel. Wilmott addresses this issue under the FAQ "best-kept secret". His answer is: "The inventors/discoverers/creators of models usually don't use them. They often use simpler models instead". The book gives the impression the Measure-Theory-Police rulez. In fact the MTP rules only in academia. The cover quotes the infamous N.Taleb with "P. Wilmott is the smartest of the quants, he may be the only smart quant". He is in my view the best entertainer. The scientific achievements are not remarkable. The rating of the book depends on ones expectations. From the entertainment point of view it is a 9, from the scientific point a 3.
  4. Russel Rhoads: Trading VIX Derivatives: Trading and Hedging Strategies Using VIX Futures, Options, and Exchange Traded Notes. Rating 6
    Rhoads is lecturer at the CBOE Options-Institute. He does his best to explain the concepts in a pedagogical way. Sometimes the book suffers even from an educational overkill. E.g. he explains at length the Dow-Jones. There is a well known book-selling rule: Each formula halves the sale. Consequently Rhoads avoids any mathematics. Even Black-Scholes is missing. Instead the author presents a lot of material about VIX related topics. He considers VIX-ETF's like VXX and VXZ. There is also some interesting material about VIX-Futures and Options strategies. According the basic intention of this book in a qualitative manner.

    The book offers some interesting starting points for an own analysis (Selling-Volatility-Insurance). But trading VIX-related products with just the knowledge Rhoads assumes and presents is a sure recipe for disaster. One can trade stocks with a few voodoo rules, because nobody has any good rules and models. It does not matter if one uses voodoo or something else. But one can model volatility and especially volatility-clustering. Volatility is by an order of magnitude more volatile than price. Volatility explodes in a crash. One should avoid volatility if one does not have a sound understanding of this behavior. The presented info is certainly too less. If the book would contain such a clear warning message, I would have put it in the recommended section.

  5. Ali Hirsa: Computational Methods in Finance. Rating 5
    This book has some qualities. It is well written, has an extensive reference section and gives a good overview about volatility and option models. But it's practical merits are in an Epsilon-neighborhood around Zero.There are a lot of integrals, but there is no single line of code. There is also no homepage were one can download one. The author is a mathematician who is probably not interested in this mundane stuff. One could have written this type of book long before the computer was invented.
    The book has emerged from lecture notes. Although it was published in 2013, there is a lot of legacy. There are only a few recent references, example-data are typically from 2000. One of these examples is calculating forward-rates. Eurodollar futures are used only for the first 2 years, because according to the author, ED-futures with longer maturities are illiquid. This is not any more the case. One can nowadays use Eurodollars for at least 5 years. The author could have simply looked at the CME ED-homepage to check this fact.
    If you like an elegant mathematics book, this one is for you. If you want to solve practical problems, it can not be really recommended.
  6. Andrei N. Borodin, Paavo Salminen: Handbook of Brownian Motion - Facts and Formulae. Rating 5
    Some time ago a friendly hedge-fund manager called me. What's the distribution of the maximal drawdown for a Brownian motion? I would say, it is the usual suspect, the Gumbel-distribution. But don't ask me for details. The question has certainly by researched. Yes it has. It is in the Handbook of Brownian Motion. The book stands in the library of the Institute for Financial-Mathematics in Vienna. But he has not yet found a person who understands it. This was a challenge I could not resist. That can't be so difficult at all. I will order it from Amazon and give you the answer next week. The 171 Euro price tag at amazon was a slight shock. But I found a cheap antiquarian offer. According the stamp at the inside the handbook was once standing at the GME-OPL department of Deutsche Bank, Alexanderstrasse 5, Berlin. The DB-quants had obviously the same problem than their Viennese colleagues. They could not decipher the contents and made it too money again.

    Normally I am quite good at finding within the mathematical calligraphy the practical relevant information. But this book is different. It is filled from the first to the last line with dense and hard-core mathematics. A lot of information is transferred in the mathematical cloud. One only finds references to mathematical papers. No idea, where the formula for the maximum drawdown is listened. I am sure, it is in the book, because everything one knows (at the time of publishing) about the Brownian is there.

    As a mathematician I admire the authors for their mastership in this theoretical and practical important field. As a practical working quant I can only shake my head about their absolute incompetency to present their knowledge in a general accessible way. This book was written only for a dozen of peers in mind. The book probably deserves from a purely mathematical point of view a 10+. From the practically point of view it is without doubt a 0-. My rating is the mean of these two extremes. But I will not resell it like the DB-quants. I have a dream: To become the 13th who can read it.

    But at least I could solve the initial question (Maximum Drawdown) The problem is due to the reflection-principle for a Brownian without thrift relative easy to solve. With a thrift it gets nasty. The initial guess - it converges to a Gumbel - was also right. But this was too easy: There are only 3 candidates to choose from and out of these 3 candidates it is almost always the Gumbel.
  7. Ganapathy Vidyamurthy: Pairs-Trading: Quantitative Methods and Analysis. Rating 6
    The author tells in the preface the story of the 6 blind man and the elephant. Each of them touches the elephant on different parts of the body. As a result for each of them the elephant "looks" quite different. The purpose of this book is to add yet another sight at the topic. I found this modesty very likeable in an otherwise bragging business. The author keeps this tone during the whole book. He never promises "get rich in 21 days".

    The first chapters are a very gentle introduction to CoIntegration. CoIntegration is the central concept behind pairs trading. If one can show that two assets are CoIntegrated, pairs trading is almost a save bet. This introduction is well written. But it is for my taste too gentle. A more in-depth treatment can be found in Market-Risk, Vol.II. If one is serious about pairs trading, one has to read this more detailed exposition. But one comes to the sad conclusion, that most interesting pairs are not CoIntegrated. I have analyzed this for the Dow-Jones and the S&P-500 from May 2006 till April-2010. The Dow and the S&P-500 are highly correlated, but not CoIntegrated. Correlation measures the short-term, CoIntegration the long term relation between time-series. Dow-Stocks are for short time-periods (e.g. half a year) CoIntegrated with the Dow. For longer periods they are not. But the short-term statistical test is almost meaningless. CoIntegration is by definition a long-term feature. The method works relative fine in stable markets. But the crash of 2008 changed all the relations. This is a general problem of finance models. The statistical relations are only valid during a stable market-environment. There is no inherent mechanism why the Dow and the S&P-500 should stay together (to stay together is the loose meaning of CoIntegration). According the author one can justify CoIntegration from the CAPM and APT market model. But CAPM and APT are at the end based on the idea of correlation and not of CoIntegration. There is no economic reason why the stocks of Apple, IBM and Microsoft should be married with each other. There is certainly a strong CoIntegration between the S&P-500 index and the S&P Futures. Index-arbitrage exploits the deviation of the current futures price to the fair-value. But index-arbitrage is far beyond the scope of individual investors. It can only be played by the big market-makers.

    The second part of the book deals with risk-arbitrage-pairs. One tries to exploit mergers. I can't give a qualified judgment about this part, as I have neither knowledge about nor experience with this trading strategy. Its like the first part easy reading.

    Wiley-Trading books are in relation to volume quite expensive. Once upon a time every renowned publisher had an editorial-office. Wiley management has obviously closed this traditional publishing institution. One reads in the book consistently about Norbert Weiner, a Weiner process. Also in the index there is a Weiner, but no Wiener. I don't blame the author. One does not see one's owns mistakes. But it should be an easy task for an experienced proofreader to spot this annoying mistake.

  8. Jeff Augen: The Volatility Edge in Options Trade: New Technical Strategies for Investing in Unstable Markets. Rating 6
    The author was according the book-cover a leading member of IBM's Life Sciences Computing business. The business is to find sequences in DNA-data. He had the idea to apply these methods to finance. At the beginning it was just a hobby. But he realized soon, that it can't be done on the weekend. Now he digs already since 10 years like once Alexey Stakhanov through the immense amount of finance data. He has build up an impressive mining infrastructure. The intensive work with real data distinguishes Augen positively from market-evangelists and theoretical academic researchers. The book contains several nuggets. But there is also a fundamental problem with his approach. One finds also in completely random data always some patterns. It's just a matter of searching long enough. Some of his results suffer also from overfitting. Augen does not address this critical point in his books (he has written several similar to this one). I have the impression that he is not even aware of the problem. The book is useful as a source of inspiration for one owns investigations. But one should not take any of his results for granted. Augen does not cheat. But many of his results are of the sort Butter-in-Bangladesh predicts the S&P

Junk

  1. Irene Aldridge: High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems. Rating 2
    The book reads like a MBA-thesis. It is a lecture about the academic standard trading literature. There are relative few academic papers about HFT. Hence the HFT specific part is in an epsilon around zero. Instead it is an overview about the classical investment theory. There is for example an own chapter about performance- and risk-measures. This is nicely done, but it is not directly related to HFT. I assume that the Sharpe-ratio is pretty useless in HFT trading. The majority of the references are from the far trading past where HFT was not yet invented.
    Once Mrs. Aldridge deals with technical details, she makes (almost all) a severe blunder. E.g. FPGAs are used for HFT in accelerator-cards. I have some experience in FPGA based HPC applications. It is obvious that she has not the slightest idea about FPGAs. This was noted by other reviewers too. It is quite strange that she denies in a comment heir incompetence. The book is full of odd technical remarks: "Genetic algorithms learn from past forecasts via the so called Bayesian approach". In fact Genetic algorithms and Bayesian methods are two independent concepts. There are any details missing, how this learning is done. There is only one HFT-specific chapter. Mrs. Aldridge defends the HFT-business against allegations of other market participants and market-regulators. This part is also superficial and basically HFT-PR: HFT provides additional liquidity and liquidity is beneficial.
    In my view nobody understands the market-mechanism. According the Geometric Brownian-motion standard model there should be regular market disasters, because a Brownian-motion is non stationary. One would have to establish on each day a commission to find out, why the markets have worked fine today. This is the miracle and not the (flash-)crash. The Pro- and Cons- about HFT are therefore sentiments and not sound scientific conclusions.
    Trading is a zero-sum-game. The wins of the HFT-traders are paid from the pockets of the underpowered gamers. But nobody acts in this game to the beneficial of the real-economy or even humanity. Practically all of trading is speculation. It happens also on the poker-server that the pros rip off the amateurs. Who laments about the bad HFT-guys should simply engage Chrilly. He will build him a trading-Hydra.
    Mrs. Aldridge's website is as scanty as the book. One has to enter a password from page 320 of the book. But the second editions ends already at page 306. Obviously nobody cared to update the website for the second edition. The references mention several working papers of Mrs. Aldridge. But only a few of them are available for download. I assume that the entropy of these papers is anyway close to the thermodynamic equilibrium.
  2. N. Taleb: The Black Swan. Rating 0
    As a trained statistician I get the blues when I read all the elegies about this book. Some praise the author even as new giant like Newton or Leibniz. He has without doubt a talent for self-portrayal. But according this criterion Lady Gaga even dwarfs God. His only scientific accomplishment is to recycle the term Black-Swan from Sir Karl Popper. He develops first a burlesque version of modern statistics and fights against his own phantom like once Hercules against Hydra. "Look guys, how brave and smart I am".

    He argues, that statistics deals only with the arithmetic mean and the normal distribution. But - believe Messias Taleb - it is not. Crashes happen, and it is not the mean, but the extreme values which counts.

    In fact there is a vast statistical literature which deals with extreme values. See Modeling Extremal Events and the list below. Black-Scholes is the standard-formula for calculating options values. The normal-distribution of returns forms indeed one of the assumptions. The inventors have of course known that returns have fat tails. But the assumption leads to an elegant and easy to calculate result. The formula became due to its simplicity a de facto standard. It is better to have a simple first approximation than a very complicated slightly more accurate solution. The traders know of course also that the formula is based on wrong assumptions. They adjust for the fat tails by increasing the volatility parameter. One enters in a wrong formula wrong parameters to get the right result (P. Wilmott). Dozens of more realistic options formulae have been developed. But they are very cumbersome to handle and far less intuitive than Black-Scholes.

    The critique of N. Taleb is not very new and original. Already in 1889 Sir Francis Galton remarked sarcastically about the Charms of Statistics: It is difficult to understand why statisticians commonly limit their inquires to Averages, and do not revel in more comprehensive views. Their souls seem as dull to the charm of variety as that of the native of our flat English counties, whose retrospect of Switzerland was that, if its mountains could be thrown into its lakes, two nuisances would be got rid at once Galton-Quotations. But Sir F. Galton has made - in contrast to Taleb - significant contributions to statistics.

    Taleb's attacks with even more furor Plato. He certainly has not read a single line of Plato's work. Instead he vulgarizes some of Sir Poppers polemics against Plato. Popper's arguments against Plato are in fact directed against Marxism. But all of this intellectual context gets lost in the Black Swan trash.

    I am in general an optimistic person. But there must be something wrong with modern culture, if Taleb is considered as a keen thinker and if Lady Gaga or Madonna are considered as great artists.

    A short list of methods which have to be according N. Taleb invented but are in fact long known.
    1. L.v.Bortkewitsch (also written as Bortkiewicz): The law of small numbers. A classical book, first published in 1898. The book deals with rare and not normally distributed events (in case of the normal-distribution one speaks of the law of large numbers). A famous example is the number of killed soldiers by horse kicks in the prussian army. The methods of this classical work form a central starting point for developments in modern finance theory (Levy-processes).
    2. Qi Li, J.S. Racine: Non Parametric Economics, Theory and Practice. Non Parametric Statistics avoids any distributional assumptions. It is an own, large research field within statistics. Some nonparametric-methods are known since more than 200 years. Non-Parametric methods pre-date the normal-distribution.
    3. R. Koenker: Quantile Regression: In quantile regression the lower and upper 10% of a distribution are of paramount interest. Sometimes the method is used to estimate the median. The usual linear regression deals with the arithmetic mean. Quantile Regression was invented 1978.
    4. R. Maronna et al.: Robust Statistics, Theory and Methods. The goal of robust statistics is to find methods which are not influenced by single outliers ("black swans"). In many investigations one has the inverse black-swan problem. One wants to measure the white swans. But the data contain a few black-ones. The arithmetic mean is a gray-swan. The median is a robust measure. The median of many white and a few black-swans is a white one. Such methods are especially important if one investigates a highly skewed distribution like income. Robust methods are known since more than 200 years.
    5. R.B. Nelsen: An Introduction to Copulas. One can model with Copulas any relation between random variables. Pearsons-R which is based on the normal assumption is just a special case of the Copula approach. Copulas were invented in 1959.
    6. D. Sornette: Why Stock Markets Crash. Published in 2001. Sornette introduced - almost 10 years before Taleb - the term King-Dragons. King-Dragons are the same than Black-Swans. His original scientific field are earthquakes. He tries to apply methods of this field to financial crashes and has also established the financial crisis observatory. One can discuss about his methods. But Sornette and his peers publish since 15 years about this topic.
  3. Jim Gatheral: The Volatility Surface: A Practitioner's Guide. Rating 3
    At the age of 17 I had wanted to be another Einstein; at 21, I would have been happy to be another Feynman; at 24, a future T.D. Lee would have sufficied. By 1976, sharing an office with other postdoctoral researchers at Oxford, I realized that I had reached the point were I merely envied the postdoc in the office next door because he had been invited to give a seminar in France (E. Derman: My Life as a Quant).

    Gatheral has studied theoretical physics in Cambridge. He shares the same sad and unfair fate. Instead of being another Einstein, he ended as a quant in Wallstreet.

    The book are notes from lectures at the Courant-Institute in New York. The lectures gained according the preface of N. Taleb cult status. Taleb and Gatheral were engaged in intellectual cockfight. As there were probably only roosters and no chickens in the auditorium, a biological useless activity. The book can of course not transport the atmosphere of these lectures. One gets only the impression of a peacock who wants to demonstrate his mathematical brilliance. He does not miss any mathematical buzzword. Not matter if it is for the topic at hand necessary or useful.

    The topic is the modelling of the volatility-surface. The surface is an artefact of the shortcomings of the Black-Scholes formula. In a perfect model there would be no volatility surface at all. Implied Volatility would be always the same. But traders correct the defiances of Black-Scholes by adjusting the volatility parameter. Puts have a higher implied volatility than calls. Far OTM Puts a higher than near the money ones. Short-term options have another implied volatility than long-term ones. The result is a rather involved implied volatility surface. Gatheral comes to the not very surprising conclusion that Stochastic Volatility with Jumps (SVJ) models the surface best. There is not much difference between the numerous variants of SVJ models. To get to this point he is juggling with the Ito-Integral, the Focker-Plank equation, Green-functions and Fourier-transform. After all he is a Wannabee-Einstein and for a Wannabee Einstein the Ito is just as simple than for other people adding 1+1. But he does never address the mundane question: How can we estimate the parameters of these models. He just assumes, that they are given by God. In fact, the parameters can't be estimated at all (Numerical Methods and Optimization). But okay Einstein did also not care about how the speed of light can be measured when he postulated his famous E=m*c2. But one gains fundamental insight without knowing c up to the 10th place after the comma. Modelling the volatility surface without knowing the parameters is in contrast pretty useless.

    If you want to read the book because you ever wanted to know what Ito, Focker-Planck ... is, be warned. Gatheral has not the slightest intention to explain it. He wants to show that he is a genius and it's the destiny of people like you and me to admire him. If you want to admire a Wannabee-Einstein, the book is a must read. If this is not according your intention, save your money and the even more valuable space on the bookshelf.

  4. Tristan Yates: Enhanced Indexing Strategies: Utilizing Futures and Options to Achieve Higher Performance. Rating 2
    The author is a relative well known finance blogger. I have read his comments about Leveraged ETF's and found them interesting.

    What to say about a book which is published in 2009, but whose time-series end all in Dec. 2006. There is not a single word about the 2008 crash. There is one simple explanation for this oversight: All the strategies with fairy-tale wins would have been busted in 2008. One could argue: Okay, in 2008 almost all strategies have failed. 2008 was a tsunami. If one expects or fears all the time a tsunami it's better to stop trading at all. But the presented strategies would have also capsized in August 2011. August 2011 was not exceptional. Such crashes happen relative frequently and a playable trading-strategy must survive such an event.

    The author tells a lot of stories and avoids any details. This is okay for a blog. If I buy a book, I expect more. I was especially disappointed about his treatment of Leveraged ETF's. The blog is in this case even more detailed than the book. Yates iterates the well known fact that LETF's are a money trap. But money does not disappear. If the naive investor throws his money into the sink it should be a source of income for the informed investor. This is the really interesting point of LETF's. Yates either does not know how to turn the sink in a source or he does not want to publish his trading secrets. In this case, he should not write a trading book. But I assume he just does not know. In case you are interested, keep an eye on the working-paper section of this home-page. The Sibyl-fund traded with some success the Johnny-Walker strategy. It will be published soon.

  5. Linda Bradford Raschke: ETF Trading Strategies Revealed. Rating 1
    The complete! contents of this book is.
    1. Take a list of ETF's. Small-, Largecap, Value, Growth, US, International, Sectors.
    2. Calculate the Returns of the last 6 months aka momentum.
    3. Invest your money 50:50 into the best two.
    4. Repeat the calculation 1 month later. If the return of one of the invested ETF's is below the median, sell the position and buy from the remaining ETF's the best one.
    5. for further details visit the home page of the author.
    6. make minor modifications to the ETF-list and goto step 1.

    If you don't believe me, that this is the complete contents of a whole book, buy it and check for yourself. I kindly ask all other readers of this review to denote half the book price to a charity organization. After all they have already read Chrilly's improved version of the book. The author is not willing or able to present her ideas in a systematic manner.

  6. Khalid Ghayur et al.: ActiveBeta Indexes: Capturing Systematic Sources of Active Equity Returns. Rating 0
    The contents of this book is normally a 4-page folder. The authors have blown up this folder to a full book by the prayer wheel method and a grand scale layout. Their idea is simple. One should classify stocks not according value and growth, but according value and momentum. They claim, that growth is usually just the complement of growth. They iterate the well known fact: In the long run value beats growth, in the long run value beats growth... If one classifies according value and momentum, a stock can belong to both camps. This intersection forms the ActiveBeta index. ActiveBeta outperforms according the authors almost always growth and sometimes also value. This is indeed true for the S&P-500 Growth ETF IVW. The authors stress in the introduction their concept is especially important for the Russel-2000. But when they come finally to the numbers part, the IWO (Russel-2000 Growth ETF) is missing. There is a simple reason for this "oversight": The IWO invalidates the whole book. The Russel-2000 Growth index is NOT the complement of the Russel-2000 Value. It is a momentum index. There are several stocks which are part of the Value- and the Growth-index. The authors have reinvented the Russel-2000 Growth wheel and broadcast this in an own book. There are thousands of bad trading books. The shocking fact about this one is: The renowned Andrew W. Lo praises the scrap in an own preface to the skies. In the middle-ages poets were paid to praise the amateurish poems of their landlords. Cervantes makes fun about this practice by writing in the Don Quixote absurd praises of the book for himself. Obviously there has not changed much since.