Evidence-Based Technical Analysis: Applying the Scientific Method and Statistical Inference to Trading Signals

by David Aronson

Evidence-Based Technical Analysis examines how you can apply the scientific method, and recently developed statistical tests, to determine the true effectiveness of technical trading signals. Throughout the book, expert David Aronson provides you with comprehensive coverage of this new methodology, which is specifically designed for evaluating the performance of rules/signals that are discovered by data mining.

Genres: Finance, Nonfiction, Business

544 pages, Hardcover

First published November 3, 2006

Reviews

[David · Rating 4 out of 5]

This book is a scientific approach to technical analysis. This is quite possibly the most objective book on this subject you may ever read. Highly recommended for anyone using techniques of TA as it helps dispel common myths and biases. It is very researched-based which is excellent since most other books are hocus-pocus. Quite unfortunate that it's essentially a 500 page textbook but it is useful nonetheless.

Pros:
Very objective
Covers tons of topics required to understand TA (eg. statistics)
Unique book
Helpful ideas

Cons:
Much longer than it needs to be
Conclusion is underwhelming
Can be very mathematical

[Jad Malaeb · Rating 4 out of 5]

A scientific exploration of truths is as alien to common trading literature as it is to astrology. This, in part, may explain the horrendously high failure rate in the trading industry.

Offering refreshing insight and hope for the aspiring trader, Evidence-Based Technical Analysis dismisses the conventional wisdom of subjective TA and ushers in objective TA as an antidote to chaos.

The author explores, in depth, why subjective TA is akin to financial astrology and how our brains are programmed to assume this path over its more rigorous counterpart. He then introduces statistics as the “language of data” and hails statistical inference as perhaps “the only method” to find true trading edges in the market.

This book is not perfect in many ways. For one, I found the important parts of the book (statistical inference, applying statistical inference to trading strategies and results) are rather short while the less-important parts are too long. Also, I was disappointed that the case study highlighted data mining rather than process of taking an idea and making into a statistically-significant strategy. Finally, while I agree that systemization is an essential element to serious trading, the author’s extreme position that any discretionary element in a trading plan turns the plan into sh*t is one that’s hard to support.

As an aspiring professional, this is still one of the best books I’ve read on trading and one I will reread many times over.

[Warren Mcpherson · Rating 4 out of 5]

Is it too much to ask that people actually test their theories before putting other people's money to work?

[Andrew · Rating 3 out of 5]

Sensible, but more of an inspiration than highly technical. Undue amount of space dedicated to basic statistical concepts. May serve as a kick-start for completely lost.

[Jairo Fraga · Rating 3 out of 5]

Um livro ok sobre o assunto Análise Técnica objetiva, acredito que o único já traduzido para o português.

O autor mostra como a análise técnica subjetiva (ondas de Elliot, padrões gráficos de figuras, etc) não passa de uma fé que os grafistas tem, e que não pode ser utilizada objetivamente com o fim de efetivamente ganhar dinheiro.

Acho que gasta muito tempo explicando conceitos básicos de método científico e vieses, mas que pode ser útil para aqueles não familiarizados com o assunto. Interessante que o autor, tão cético e "científico", toma como garantido suas alegações sobre a teoria da evolução. Prossegue nos seus dogmas da "ciência moderna", falando inúmeras bobagens em relação à Aristóteles, criticando sua metafísica, que ele teria sido um obstáculo ao progresso científico. Ataca também a Igreja Católica e propaga distorções sobre Galileu, mas nada de novo aqui, basta ver, por exemplo, o comportamento dos proponentes dogmáticos do "evidence based medicine" hoje em dia.

Após pelo menos metade do livro passando por assuntos que não são necessários para as conclusões que se quer chegar aqui, começa a parte mais interessante do livro, explicando sobre Monte Carlo, Bootstrap, testes fora-da-amostra, fatores de correção Markowitz/Xu, "walk-forward optimization".

Dedica uma parte do livro a refutar a facilmente refutável hipótese dos mercados eficientes. E prossegue novamente com mais conversas sobre o motivo dos mercados não serem um passeio aleatório, citando diversos vieses, como aqueles propostos pelas finanças comportamentais.

Uma parte interessante explica o surgimento de tendências e como, por exemplo nas commodities, nada mais é do que a consequência da oferta e da demanda dos hedgers, e que a tendência é o prêmio de risco, coisa que é mais difícil existir nas ações.

Outra parte, que foi a mais interessante que achei no livro, foi a referente aos métodos estatísticos para considerar o viés de mineração de dados no contexto em que, se milhões de regras forem testadas para se achar uma estratégia, não se pode ter a mesma significância estatística de se fizéssemos as análises a partir de uma regra gerada manualmente através de um único backtest.

Finaliza analisando mais um pouco a questão da subjetividade, e como ela em questões científicas é inferior aos critérios objetivos, sendo mostrado no livro diversos exemplos de estudos que provam isso para diversas áreas científicas.

Eu esperava muito mais desse livro, e só extraí de bom mesmo a parte do viés da mineração de dados. Recomendo outros livros ligeiramente diferentes mas que são mais úteis para o trader, como os do Ernest Chan e especialmente Brent Penfold, ambos em inglês.

Tempo estimado de leitura: 11 horas

[Piotr Karaś · Rating 5 out of 5]

This book is amazing! The author explains his point so clearly that one can easily understand even the most sophisticated problems of statistics and econometrics pertinent to the topic of testing the trading systems. Having read this book, it's hard not to be convinced that continuing with the subjective technical analysis makes no sense, and that successful trading requires at least the use of objective pattern definitions allowing for consistent judgements, if not computer-based automation. The best take-aways are in the area of the objective technical-analysis research and the data-mining process. I particularly liked the author's ideas on how to make best use of the limited data sample to produce reliable testing results. The author also drives the point home that even when you find the best profitable system during the data-mining process it is very likely that it's superior performance is just due to luck.
It's certainly a must-read for every aspiring trader. The sooner one reads it, the better.

[Scott Constantine · Rating 5 out of 5]

If you still think the jury's out on the validity of most securities trading technical analysis methods (beyond moving averages and trend lines), this book debunks this pseudoscience so rigorously and boringly that it'll send you to sleep with nightmares about why you ever thought this stuff worked in the first place. Pro-tip, skip to the final chapters where he aggregates his findings because you don't need to listen to him prove each and every one of hundreds of technical indicators is no better than reading tea leaves. Or maybe you do. I don't know your trading history. But if you're a quant trader and want to really kick the technical analysis bug once and for all, this is your golden opportunity.

[Noud · Rating 4 out of 5]

Definitely one of the most authoritative books within the realm of sceptical empiricism applied to financial market prediction. The book covers a lot of statistical, psychological and philosophical topics to build a foundation for the actual application of TA rules which is quite useful. However, some readers (like myself) might find the discussion of basic statistical topics and/or the discussion of heuristics and biases redundant.

[utkal · Rating 3 out of 5]

This book is practically a 600 page research paper which not only give you the outcome of the research but spends considerable time (70%) on explaining the research methodology in detail. The author goes deep into explaining fundamentals of statistics - from normal distribution to t-test and confidence intervals. The book ultimately raises valid questions against subjective TA and makes the case for objective TA.

[Ryan · Rating 3 out of 5]

If you want to learn how to trade better, you're not gonna find any useful indicators or TA here.

Written in an academic format. 80% of it is apply cognitive biases / critical thinking / epistemology to TA to show how there are flaws and explaining what scientific groundwork needs to be laid to actually use TA

Author was a prop trader for 5 years before he became a professor

- Proves Head & Shoulders pattern is BS, through heavy research using the whole scientific method

[Lime Street Labrador · Rating 4 out of 5]

A discussion of backtesting methods and issues, and then finally runs a test on various undefined technical analysis indicators. While many discussions are basic (what is hypothesis testing etc), some points are relevant (walk-forward, multiple comparison etc) and he includes references to dig into further. Worth a re-visit when coding my own backtest framework. I wish he could include a few ways he defined stock patterns such as double tops or flags.

[Ethan Drower · Rating 5 out of 5]

Trading must read.

[Ifor Williams · Rating 5 out of 5]

Absolutely essential

[Javier Villar · Rating 2 out of 5]

This is academia. No real trading utility in this book.

[Murilo Spino · Rating 5 out of 5]

Great book for those who are interested in trading automation.

[E · Rating 5 out of 5]

If you try to predict a coin-toss using past coin-tosses, the best information you can learn from it is that half the time coins show heads and the other half they show tails. If it seems like you are able to call heads when it actually happens most likely it is just a fluke. To be convinced that you can actually predict coin tosses, you have to be able to predict so many heads correctly it is not even a joke at that point.

This is how it is with trying to predict prices using prices.

The correct way to predict prices is by using the causes that cause prices, and even then you'd be only right in the long run. While it is true prices can cause prices (see momentum), without a notion of a fair price derived from other external factors, the momentum cannot be relied upon.

[Mariano · Rating 3 out of 5]

Empirical look into technical analysis. Dispels a lot of myths surrounding TA and makes the case that well-researched trading rules can work.

[Keith Lansford · Rating 4 out of 5]

Very thourough submission on statistical inference and data mining, as well as the latest in behavioral finance. Interesting read.