Optimal Portfolio Modeling: Models to Maximize Returns and Control Risk in Excel and R

By Philip McDonnell

Quick Summary

Philip McDonnell provides a quantitative, hands-on guide to portfolio optimization and risk management using mathematical models implemented in both Excel and R. The book covers modern portfolio theory, utility functions, the Kelly criterion, variance and covariance analysis, Monte Carlo simulation concepts, and practical methods for constructing portfolios that balance return maximization with risk control -- all with downloadable software for implementation.

Detailed Summary

Modern Portfolio Theory Foundations

The book builds from the foundations of modern portfolio theory (Markowitz mean-variance optimization, James Tobin's capital market line) through advanced quantitative methods. The mathematical treatment is rigorous but accessible, with each concept accompanied by implementation in both Excel (for those comfortable with spreadsheets) and R (for those seeking more powerful statistical computing capabilities).

Utility Theory and Risk Preferences

A distinctive feature is the thorough treatment of utility theory -- the mathematical framework for modeling investor risk preferences. Multiple utility functions are examined: logarithmic utility (which leads to the Kelly criterion), square root functions, and monotonically increasing concave functions that model risk aversion. The relationship between utility theory and the practical question of "how much should I bet?" is clearly drawn, connecting Von Neumann and Morgenstern's foundational work to practical portfolio sizing.

The Kelly Criterion

The Kelly criterion -- which determines the optimal fraction of capital to wager on a favorable bet to maximize the long-term growth rate of wealth -- receives substantial treatment. McDonnell covers Ed Thorp's application of Kelly from gambling to financial markets, the mathematical derivation, the distinction between full Kelly and fractional Kelly (which reduces volatility at the cost of some expected return), and the practical challenges of estimating the inputs required for Kelly calculations in real markets.

Variance, Covariance, and Correlation

The mathematics of portfolio risk -- variance of individual assets, covariance between asset pairs, the formula for portfolio variance incorporating correlation -- is explained and implemented computationally. The book shows how the variance of a portfolio of correlated assets can be dramatically different from the sum of individual variances, providing the mathematical basis for diversification.

Statistical Testing and Behavioral Finance

The t-test and other statistical methods for evaluating whether observed performance is statistically significant (rather than attributable to chance) are covered. The integration of behavioral finance insights -- particularly the work of Kahneman, Tversky, and the concept of loss aversion -- with quantitative optimization methods acknowledges the gap between mathematically optimal portfolios and psychologically comfortable ones.

Practical Implementation

The emphasis throughout is on practical implementation. The CD-ROM (and downloadable materials) includes working Excel spreadsheets and R scripts that readers can use immediately with their own data. The goal is to move portfolio optimization from theory to practice, enabling readers to construct, test, and refine their own portfolio models.

Categories

• Risk Management
• Investing
• Quantitative Analysis
• Portfolio Management

Key Takeaways

• Modern portfolio theory provides the mathematical framework for balancing return and risk through diversification
• The Kelly criterion determines optimal position sizing to maximize long-term wealth growth
• Variance and covariance analysis reveals how correlation between assets affects portfolio risk
• Practical implementation in Excel and R makes quantitative portfolio optimization accessible to individual investors
• Behavioral finance insights must complement mathematical optimization because psychologically tolerable portfolios differ from mathematically optimal ones