Mastering Backtesting for Algorithmic Trading

Machine learning offers a set of powerful tools that holds considerable promise for investment management. As with most quantitative applications in finance, the danger of misapplying these techniques can lead to disappointment. One crucial limitation involves data availability. Many of machine learning’s early successes originated in the physical and biological sciences, in which truly vast amounts of data are available. Machine learning applications often require far more data than are available in finance, which is of particular concern in longer-horizon investing. Hence, choosing the right applications before applying the tools is important. In addition, capital markets reflect the actions of people, which may be influenced by others’ actions and by the findings of past research. In many ways, the challenges that affect machine learning are merely a continuation of the long-standing issues researchers have always faced in quantitative finance. While investors need to be cautious—indeed, more cautious than in past applications of quantitative methods—these new tools offer many potential applications in finance. In this video lecture, we explore the author's research protocol that pertains both to the application of machine learning techniques and to quantitative finance in general.

Data is a crucial aspect of a successful quant model. But what is good data and what are some of the best practice tips for data?

This video discusses the importance of data to a successful quantitative model, what are some of the characteristics of good data, and some best data practice tips.

The video is based on the paper “Best Practices in Research for Quantitative Equity Strategies” by Joseph A. Carniglia, Frank J. Fabozzi, and Petter N. Kolm.

What are some of the different types of quant strategies? And what characteristics are needed for a good quant model?

This video provides a taxonomy of some of the best quant strategies and outlines some of the key characteristics of constructing a quant model that every quant should know.

The video is based on the paper “Best Practices in Research for Quantitative Equity Strategies” by Joseph A. Carniglia, Frank J. Fabozzi, and Petter N. Kolm.

What are some of the best steps to follow when developing a successful quant strategy?

This video discusses the idea and philosophy behind quant strategies and models, and how quant strategies are developed in 6 steps:

1. Formulating investment ideas and strategies

2. Developing signals

3. Acquiring and processing data

4. Analyzing the signals

5. Building the strategy

6. Evaluating, testing, and implementing the strategy.

The video is based on the paper “Best Practices in Research for Quantitative Equity Strategies” by Joseph A. Carniglia, Frank J. Fabozzi, and Petter N. Kolm.

Quant models are always an approximation to some aspect of the markets and, therefore, wrong. Still, we can use models to make accurate forecasts about new financial data.

In this video, we discuss the idea concerning models and the real world. We then present 9 tips for better model development and, finally, we discuss what quant models can mean to quant investors and fundamental investors.

The video is based on the paper “Best Practices in Research for Quantitative Equity Strategies” by Joseph A. Carniglia, Frank J. Fabozzi, and Petter N. Kolm.

This article guides users through essential resources for learning. It recommends a comprehensive book for theoretical understanding and two Python libraries, for practical application, giving readers a solid foundation in the principles and techniques of causality in data science.

We delved into how Lopez de Prado's work makes the argument for utilizing causality in factor investing through causal graphs. These graphs uncover intricate causal ties shaping strategies.

In addition to the paper, we explored a comprehensive beginner's introduction to causality.

This lecture we review the first 4 chapters of Marcos Lopez de Prado's latest paper from ADIA Labs, which lays out the foundation of the scientific discovery process in finance and how Causal Inference is the next step.

Based on: Causal Factor Investing: Can Factor Investing Become Scientific?

In this lecture, we discuss the seven common mistakes investors tend to make when they perform backtesting and build quant models. Some of these may be familiar to our students, but nonetheless, you may be surprised to see the impact of these biases. The other sins are so commonplace in both academia and practitioner’s research that we usually take them for granted.

Machine Learning has been used in the financial services industry for over 40 years, yet it is only in recent years that it has become more pervasive across investment management and trading. Machine learning provides a more general framework for financial modeling than its linear parametric predecessors, generalizing archetypal modeling approaches, such as factor modeling, derivative pricing, portfolio construction, optimal hedging with model-free, data-driven approaches which are more robust to model risk and capture outliers. Yet despite their demonstrated potential, barriers to adoption have emerged – most of them artifacts of the sociology of this inter-disciplinary field. Based on discussions with several industry experts and the authors' multi-decadal experience using machine learning and traditional quantitative finance at investment banks, asset management and securities trading firms, this position article identifies the major red flags and sets out guidelines and solutions to avoid them. Examples using supervised learning and reinforcement in investment management & trading are provided to illustrate best practices.

In this lecture, we discuss the 9 principles of effective machine learning in finance: 

1. Problem definition

2. Modeling assumptions

3. Develop intuition

4. Defensible results

5. Diagnostics

6. Keep it simple

7. Choice of the utility function

8. Solution constraints

9. Is your data biased?

10. Feature Engineering

In this video, we explore the False Strategy Theorem proposed by Marcos Lopez de Prado and David Bailey, which provides a solution for selection bias under multiple testing in investment strategy development.

The False Strategy Theorem argues that most investment strategies fail due to the "false positive" problem, where a strategy may appear successful when backtested on historical data, but fails when applied to new and unseen data. This is because traditional backtesting methods do not account for selection bias and multiple testing, leading to the false discovery of successful strategies.

In this video, we discuss the paper "Detection of false investment strategies using unsupervised learning methods". We discuss several topics such as the Probabilistic Sharpe Ratio and Deflated Sharpe Ratio and how to use them. Practical considerations are also discussed for estimating certain variables within these equations, which is done through the Optimal Number of Clusters (ONC) algorithm.

In this video we discuss the paper "Backtesting", which provides a statistical framework to analytically determine the discount factor applied to the Sharpe ratio when multiple testing is factored in.

Prof. Campbell R. Harvey delivers the keynote address at the 18th Annual Portfolio Management Conference, June 10, 2015 in Frankfurt.

We delve into the fascinating world of Kelly Criterion and explore its four different approaches to bet sizing.

We will enhance your understanding of optimal capital allocation strategies with both theory and code. Discover the discrete-time and continuous-time Kelly Criterion, delve into the optimal allocation Kelly Criterion, and explore the risk-constrained Kelly Criterion.

Don't miss out on this opportunity to deepen your knowledge and gain insights into bet sizing techniques that can enhance your decision-making in the world of investments.

Tackling the intricacies of stop-loss rules was the focus of our latest reading group session, building upon prior discussions like 'Penny-Picking' and 'Skewness-Seeking.' A robust framework guided our analysis of stop-loss impact on expected returns.

The Random Walk Hypothesis, AR(1) and Regime-switching models also entered the conversation, offering insights into the efficacy of stop-loss strategies under varying market conditions. Various performance metrics were considered for evaluating the effectiveness of stop-loss

This lecture discusses the fundamentals of the market structure and trading, along with how these impact the execution of investment strategies.