Quantitative Trading Systems

I am Dr. Howard B. Bandy.

This is my blog about technical analysis, including Statistics-based Technical Analysis.

It is based, in large part, on my books “Quantitative Trading Systems” and “Modeling Trading System Performance”

This blog coordinates and consolidates my thinking, analysis, and writing related to trading financial markets.

The focus is on technical analysis — techniques that can be expressed unambiguously, tested with historical data, and used to identify potentially profitable trading opportunities. Analysis that can be described using statistics.

There are several threads, including:

1. A forum for discussion of trading system development techniques — methods for design, testing, validation, and analysis of trading systems.

2. Analysis of specific trading systems — think of this as Statistics-based Technical Analysis.

3. Forecasts, stated as probabilities, of future price changes of specific tradable issues based on specific trading systems.

Readers are encouraged to ask questions and make comments. This is a moderated forum. Comments may be posted as submitted or edited before posting. Please be civil, brief, and clear. Make as many comments as you wish, but only one comment or question per posting.

My background

My background includes both educational and practical experience.

I have university degrees in mathematics, physics, engineering, and computer science. My graduate work focused on artificial intelligence, statistics, and modeling and simulation. I held academic positions as professor of mathematics and computer science, department chair, and university dean.

I worked as a programmer, analyst, program developer, and chief information office in a variety of companies.

I designed, programmed, and documented a commercially successful program to select stocks and give buy and sell signals.

I was a senior technical analysis for a Commodity Trading Advisor (CTA) firm, where I held a Series 3 license. I designed, developed, tested, and analyzed trading systems.

I tried retirement, during which my wife (a registered nurse with multiple biology degrees) and I traveled in an RV. It was during that time that we managed a fishing lodge on Kodiak Island, Alaska.

Throughout the period of retirement, I continued to study the financial markets and to develop methods to trade them profitably. As I read, I was often frustrated by books and articles that outlined some idea and claimed some result, but did not provide enough specific detail to allow the reader to replicate the author’s results.

In 2007 I wrote and published the First Edition of “Quantitative Trading Systems” (QTS), in which I outlined the technical analysis methods I use, recommend, and teach. The book takes a systems engineering approach — specify the goal, choose the metric by which success will be measured, develop trading models using an in-sample set of data, validate those models using an out-of-sample set of data.

In order to enable every reader to replicate the ideas and techniques I describe, I needed to implement those systems using a specific trading system development platform. I chose AmiBroker because it, almost alone among retail-level platforms, is powerful enough to support all the features I needed. QTS contains 80 some trading systems — both complete systems and segments — used to illustrate specific ideas. Although the book uses the AmiBroker platform, readers who use other platforms should have little difficulty translating. QTS has been very well received. It has sold several thousand copies and has been shipped to over 50 countries. The revised and updated Second Edition was published in 2011.

Critical questions about trading

As a trading system moves from development to implementation, several critical questions arise, including:

1. What results can be expected from live trading?

2. What are the indications that the system is broken?

3. How should position sized be determined?

The first is answered through walk forward testing, which is explained in detail in QTS. In essence, each of the walk forward steps is an example of that transition — a practice.

The second requires some statistical analysis. It is addressed briefly in QTS, and in more detail in the sequel, “Modeling Trading System Performance” (MTSP). There are many thousands of trading systems, any of which could be used to give buy and sell signals for any of many thousands of tradable issues. If all of these combinations are tested, some will be truly profitable and tradable, some will appear to be profitable in backtesting due to chance, some are clearly not useable. Determining whether a given system is healthy and may be safely traded or is broken and should not be traded is the most important decision a trader must make. Because of the high variability of results of most trading systems, it is difficult to separate luck from skill. But MTSP outlines some procedures that help.

The third is position size, which determines the growth of the trading account. If a system is broken, the correct position size is zero. If a system is healthy and working, position size is related to the variability of the system and to the risk the trader is willing to take — that is, tolerance to drawdown. MTSP addresses this third question in detail. The analysis begins by expanding beyond single value results such as mean, standard deviation, and largest loss. Examination of the distribution of results, together with Monte Carlo simulation, allows estimates to be made that relate account growth and expected drawdown to position size. MTSP addresses these issues in detail. Analysis is done using Excel and is independent of the trading system development platform.

Contact

Private email: howard@BlueOwlPress.com

main website: BlueOwlPress

book websites:

        Quantitative Trading Systems (Special pricing for Blog readers)

        Modeling Trading System Performance (Special pricing for Blog readers)

        Introduction to AmiBroker