- Tapa blanda: 448 páginas
- Editor: Blue Owl Press, Incorporated; Edición: 1 (24 de enero de 2015)
- Idioma: Inglés
- ISBN-10: 0979183855
- ISBN-13: 978-0979183850
- Valoración media de los clientes: 1 opinión de cliente
- Clasificación en los más vendidos de Amazon: nº101.932 en Libros en idiomas extranjeros (Ver el Top 100 en Libros en idiomas extranjeros)
Quantitative Technical Analysis: An integrated approach to trading system development and trading management (Inglés) Tapa blanda – 24 ene 2015
|Nuevo desde||Usado desde|
Comprados juntos habitualmente
Los clientes que compraron este producto también compraron
Descripción del producto
Reseña del editor
An integrated approach to trading system development and trading management
No es necesario ningún dispositivo Kindle. Descárgate una de las apps de Kindle gratuitas para comenzar a leer libros Kindle en tu smartphone, tablet u ordenador.
Obtén la app gratuita:
Detalles del producto
Si eres el vendedor de este producto, ¿te gustaría sugerir ciertos cambios a través del servicio de atención al vendedor?
¿Qué otros productos compran los clientes tras ver este producto?
Opiniones de clientes
Principales opiniones de clientes
Ha surgido un problema al filtrar las opiniones justo en este momento. Vuelva a intentarlo en otro momento.
Opiniones de clientes más útiles en Amazon.com
A bit about my background so you know what I'm writing about:
I'm an individual trader. I have about 10 years of trading experience and all of it was using trading system development platforms (not machine learning). After experimenting with many platforms such as Tradestation, Ninjatrader, Telechart, StockFinder I settled on Amibroker as my platform since 2011. I consider myself a competent user of Amibroker. I also have Dr Bandy's prior books and consider them outstanding.
I have a software development background and early in career (almost 20 years ago) was exposed to some machine learning concepts. My application of machine learning was however, not for trading or any financial services. My machine learning experience was using C and C++ as programming languages. Since 2012, I elected to learn Python and have come to use it and love it!
Given my machine learning exposure I have been on the lookout for books that bridge the gap between theory and their application to trading. Quantitative Technical Analysis does that extremely well. Just to be clear, this book is a great asset if you use Amibroker and have no intention of using machine learning methods, but I'll focus my review on the topics of my interest -- namely the application of machine learning using Python to trading.
Here is what I loved about the chapter progressions:
- The opening chapters discuss the key concepts with respect to key risk in trading -- the risk of drawdowns and provide a framework for managing this risk algorithmically. I have learnt tremendously from Dr Bandy's prior books and welcome this summary of the concepts along with more and detailed explanation of the concepts.
- Next, the design of a trading system are considered. Assume a system accuracy and maximum holding period, what is the profit potential and draw down?
- Armed with the knowledge from the earlier step, now the implementation of the actual trading system is tackled. This is where the books is unique. It provides the classic (what Dr Bandy call's the indicator based approach) way of developing systems using a platform like Amibroker and in *parallel*, introduces the use of Python as a platform for trading development. For me, this was a great way to bridge the concepts of system development using Amibroker to Python. Amibroker with its rich library of indicators and charts and access to data sources makes many things easy. Python, on the other hand, is a great language for complex analysis of data and monte carlo simulations. Comparing the two approaches side-by-side makes it easy to decide the approach you want to take. (I'll take a hybrid approach for now).
- Next, the use of machine learning to predict prices is introduced. For me, this was the part I have been waiting for a long time. Dr Bandy takes the classic Iris classification problem to illustrate the various machine learning algorithms (from the now widely available scikit Python ML library). He then follows up with a real trading system (using logistic regression as the algorithm). This is a fully functional system and if you follow the Iris examples, you can easily experiment with different learning algorithms. If you have prior exposure to machine learning concepts, you should be up and running with a working system in a few hours (like me). If you are new to machine learning, this is the best book I know of that takes the concepts of machine learning, combines them with freely available Python code and libraries and free data sources and puts you in a position to advance your learning about real trading systems.
- A Python implementation of the dynamic position sizing technique is introduced. The concept of dynamic position sizing, once explained, seems intuitive and obvious. What is challenging is the actual execution of the concept. This book provided a complete working example. Heck, the "for education only" examples here are a lot better than many production systems.It provides a daily management of the position size of the trade based on changing risk conditions. It is brilliant!
Very few books can boast of providing working examples and fewer can boast of working examples that work for the reader. Dr Bandy has set a high bar with his prior books so we've come to expect this. Still, getting all of the Python code to work withing hours of getting the book is an amazing testament to his attention to detail. If you are at all interested in applying machine learning to trading, getting this book is a nobrainer!.
I just wanted to update my review after having spent considerable time using some of the techniques in this book and learning more about quantitative technical analysis of market data from other sources. I still stand behind my original review that is below this edit.
First, I think the title of this book is a bit misleading. This book does not really cover quantitative technical analysis. The book is really an introduction to developing automated trading systems using both the more traditional indicator approach and the newer machine learning approach. There is not much information about statistics in this book, and statistics are really the foundation of quantitative analysis. I think that the contents of a book with this title should be at least half full of statistics, particularly how they should be applied to market data and to analysis of trading system returns.
Second, while using python was fun and interesting at first, I ran into a lot of issues trying to do statistical analysis with it. The best part of python is that you can use the interpreter with compact commands and do some powerful number crunching and generate charts on the fly. The problem is that a lot of the commands and syntax for both SciPy and pandas are very poorly documented. I spent days trying to figure out how to do things and sometimes I could not find any good examples or explanations of how to use certain methods. Keep in mind that I have been programming for over 15 years and I am quite good at figuring things like this out.
Finally, I wanted to provide some links to resources that I have found helpful. I discovered another author who does a very good job at explaining quantitative analysis, his name is Adam Grimes. He wrote a book that sells on Amazon called The Art and Science of Technical Analysis (http://www.amazon.com/The-Art-Science-Technical-Analysis/dp/1118115120). I took his free 35 hour trading course and found his approach to analyzing trading systems superior. He also has a free ebook (Quantitative Analysis of Market Data: a Primer) that was quite helpful. You can find these on his website (http://adamhgrimes.com/TAAS/). Finally, if you are inclined to use C#, I found a very promising open-source library that has plenty of statistical and machine learning components. It's called Accord (http://accord-framework.net/). The library comes with many samples, but none of them are trading related.
*** End Edit ***
I was really excited when I bought and received this book. It seemed like just the thing I was looking for - a new book on state of the art trading system development that included machine learning, coupled with downloadable source code.
While I can't say that the title, table of contents, or index were inaccurate, the content was a little less robust than I had hoped. It was as if titles were given to sections just to beef up the table of contents and many buzzwords were included to make the index more sexy. There were several instances where the author only gave a cursory overview of a topic and directed the reader to some webpage or book to learn more, without even trying to give a basic explanation of what the topic means.
I was slightly offended when the author chose to include program listings and results for fifteen different machine learning algorithms that were nearly exact copies of each other, except for 2 lines of code to call the featured learning algorithm. Each of these copies contained a note telling you to go to the scikit-learn webpage to learn more. What a waste of 30 pages in a book that costs so much! I would have much rather seen the author use those 30 pages to explain these different learning algorithms and describe which ones to use for different types of trading systems.
Another area that was lacking is the application of the book. All the examples and sample code assume that you are working on a stock/ETF trading system that enters/exits trades once a day. I want to build a system that trades highly leveraged futures many times intraday, and there isn't much information on how to handle that. I thought "Trade Management" would cover techniques to manage live trades, adjusting positions while in a trade. This book uses the term to mean end of day assessment of a trading system after a trade is closed, or marked to market. Then using the results to compute position sizing and decide whether you should take the next trade. The DynamicPositionSizing program provided in the book takes a few hours to run on a set of 3000 trades. This might not be an issue for an end-of-day system, but that definitely won't work for a system that generates 100 trades a day.
The book recommends not including stop losses in your trading system. With leveraged futures products, many people cannot afford to do this. If you are building a high frequency futures trading system, you really need to use stops, and this doesn't mesh well with the sample code in this book that is supposed to calculate risk, a safe position size, and the objective function. I'm sure it can be modified to work, but it is a bit disappointing that I have to do these additional steps on my own.
One final gripe is that this book uses Amibroker for the indicator-based portion of the book. Amibroker can be powerful, but it's language and platform are difficult to use, even for a seasoned programmer such as myself. It would have been much better if a more widely used trading development platform such as NinjaTrader were used. Or even better, if a custom platform was used and provided as a download.
That said, I still think there is some value in this book. It gives a pretty good overview of trading system development and things to look out for. This book also gives an introduction to using machine learning for trading system development. For someone new to machine learning like myself, I find this useful - at least now I know which python libraries and websites to look to for more information (hint: scikit-learn). The downloadable python code is probably the best part of this book, especially if you want to build a stock/ETF end-of-day swing trading system.
For those interested, the python libraries used for the machine learning stuff are:
"For more information, refer to the scikit-learn webpage." - Howard Bandy