Optimization and Portfolio Construction
- Estimation Error and the "Fundamental Law of Active Management": Is Quant Fundamentally Flawed?
The authors show with intuitive discussion followed by a novel simulation study that applications of the Grinold (1989) "Fundamental Law" theory for optimized portfolio design are often unreliable and self-defeating.
Authors: Richard Michaud, David Esch, and Robert Michaud
Publication: Journal Of Investing June 2020 - Estimation Error and the "Fundamental Law of Active Management": Technical Companion
A technical description for the simulation experiment within the paper, designed to serve as a companion piece.
Authors: David Esch, Richard Michaud, and Robert Michaud - Comment on: Allen, D., C. Lizieri, S. Satchell 2019. "In Defense of Portfolio Optimization: What If We Can Forecast?”
An examination of how and why the findings in Allen et al (2019) are inconsistent with canonical Monte Carlo simulation studies of estimation error in MV optimization.
Author: Richard Michaud, David Esch, and Robert Michaud
Publication: Financial Analysts Journal February 2020 - Comment on: Kritzman, M. 2006, “Are Optimizers Error Maximizers?”
In this comment, Dr. Richard Michaud examines the shortcomings of Mark Kritzman’s 2006 challenge to the Michaud rule, “Are Optimizers Error Maximizers?” and offers compelling counterexamples to illustrate the issue of error maximization.
Author: Richard Michaud
Publication: Journal of Portfolio Management, forthcoming - When Michaud Optimization Fails
This working paper examines the characteristics of risk and return numbers used in simulations where classic Markowitz frontier optimization beat Michaud frontier optimization.
Authors: Richard Michaud and David Esch - Comment on: "The Road Not Taken" by C. French, Journal Of Investment Management 14(4): 4-13
Dr. Michaud offers a thoughtful response to French's piece on the rejection of the Markowitz (1959) critical line algorithm by investment managers.
Author: Richard Michaud - Reply to 'Reply to 'Comment on 'Markowitz versus Michaud: Portfolio Optimization Strategies Reconsidered,' Becker, Gürtler and Hibbeln, European Journal of Finance, 21(4): 2015.'''
The continued academic exchange concerning a critique of Michaud optimization.
Authors: Richard Michaud, Robert Michaud, David Esch. - Deconstructing Black-Litterman (available through JOIM)
Black-Litterman optimization claims to solve the problems of mean-variance optimization in practice, but our analysis demonstrates that it has limited investment value.
Authors: Richard Michaud, David Esch, and Robert Michaud
Publication: Journal Of Investment Management 1st quarter 2013 - Deconstructing Black-Litterman (draft)
Black-Litterman optimization claims to solve the problems of mean-variance optimization in practice, but our analysis demonstrates that it has limited investment value.
Authors: Richard Michaud, David Esch, and Robert Michaud
Publication: Journal Of Investment Management 1st quarter 2013 - Comment on 'Markowitz versus Michaud: Portfolio Optimization Strategies Reconsidered,' Becker, Gürtler and Hibbeln, European Journal of Finance, 21(4): 2015.
New Frontier responds to a critique of Michaud optimization.
Authors: Richard Michaud, Robert Michaud, David Esch. - Portfolio Monitoring in Theory and Practice (available through JOIM)
The described algorithms improve the when-to-trade decision and allow for large-scale automatable, non-calendar-based portfolio monitoring.
Author: Richard Michaud, David Esch, and Robert Michaud
Publication: JOIM Vol 10, No. 4, 2012 - Portfolio Monitoring in Theory and Practice (draft)
The described algorithms improve the when-to-trade decision and allow for large-scale automatable, non-calendar-based portfolio monitoring.
Author: Richard Michaud, David Esch, and Robert Michaud - Morningstar vs. Michaud Optimization
Richard Michaud and David Esch present the first direct comparison as well as an analysis of key differences.
Author: Richard Michaud and David Esch
Publication: September 2012 Newsletter - Non-Normality Facts and Fallacies (available through JOIM)
A summary rejection of normal distributions is almost always ill-advised.
Author: David Esch
Publication: Journal Of Investment Management 1st quarter 2010 - Non-Normality Facts and Fallacies (draft)
A summary rejection of normal distributions is almost always ill-advised.
Author: David Esch
Publication: September 2009 - Are Good Estimates Enough?
Markowitz optimization has been the standard, but it does not correct for estimate uncertainty.
Author: Richard Michaud and Robert Michaud
Publication: Investment Management Consultants Association. January/February 2009 - Estimation Error and Portfolio Optimization (available through JOIM)
Richard Michaud and Robert Michaud review and update the optimization information introduced in Efficient Asset Management.
Author: Richard Michaud and Robert Michaud
Publication: JOIM First Quarter 2008 - Estimation Error and Portfolio Optimization (draft)
Richard Michaud and Robert Michaud review and update the optimization information introduced in Efficient Asset Management.
Author: Richard Michaud and Robert Michaud - Defense of Markowitz-Usmen
Dr. Michaud and Robert Michaud respond to Harvey et al.'s criticism of the simulation tests used by Markowitz and Usmen.
Author: Robert Michaud and Richard Michaud
Publication: NFA March 2008 - Scherer's Errors
Scherer published a critique of resampled efficiency. This paper provides corrections to the invalid conclusions found in Scherer.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: December 2005 Newsletter - Resampled Efficiency Equity Portfolio Optimizer
New Frontier describes the features and theory behind the new Equity Optimizer.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: October 2005 Newsletter - The Information Ratio of Factor Based Alpha
Misconceptions concerning the proper definition of breadth in Grinold’s Active Law of Management have suggested that the information ratio of optimized portfolios increases with the number of stocks in the portfolio. We show that when active return depends on factor bets, the IR has an upper bound independent of the number of stocks, but depending on the breadth of the strategy and some maximum information ratio of the joint factor bet.
Authors: Noah Kraut, Robert O. Michaud & Richard O. Michaud
Publication: October 2005 Newsletter. - Equity Optimization Issues-V: Monte Carlo and Optimization Errors
Improvements in optimization design and resolutions of fallacies in asset management practice are largely due to recent applications of Monte Carlo simulation technology.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: August 2005 Newsletter - Equity Optimization Issues-IV: The Fundamental Law of Mismanagement
The Grinold Law of Active Management is one of the most widely referenced and misused formulas in investment theory and practice.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: July 2005 Newsletter - Equity Optimization Issues-III: Insignificant Alphas, Heterogeneous Errors
Insignificant alphas and heterogeneous estimation error are two issues associated with performance limitations in mean variance equity portfolio optimization.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: March 2005 Newsletter - Equity Optimization Issues-II: Large Stock Universes and Scaling Alphas
In order to obtain the provable benefits of Resampled Efficiency, a number of common ad hoc equity portfolio optimization techniques need to be avoided or corrected. This article focuses on two: the use of large stock universes and incorrect alpha scaling.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: February 2005 Newsletter - Equity Optimization Issues-I
The first equity optimization article is a beginning discussion of difficulties of traditional optimizers and the solutions New Frontier was starting to explore.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: November 2004 Newsletter - Optimization with Non-Normal Resampling
In order to retain some of the normal distribution relevance for MC optimization, we use a multivariate distribution procedure that allows for exogenous specification of skewness and kurtosis.
Author: Noah Kraut
Publication: September 2004 Newsletter - Forecast Confidence Level and Portfolio Optimization
This report focuses on the role and importance of the uncertainty in forecast information in constructing portfolios with the optimal performance.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: July 2004 Newsletter - Resampled Efficiency Fallacies
This report responds to critiques of Resampled Efficiency.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: April 2004 Newsletter - Resampled Efficiency vs. Bayes: Implications for Asset Management
Good inputs, prepared with Bayesian statistics, are no better than bad inputs if the portfolio construction process misuses investment information.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: February 2004 Newsletter - Why Mean-Variance Optimization Isn't Useful for Investment Management
The logic of mean variance optimization is seductive, but the seduction unravels in the investment period.
Author: Richard O. Michaud
Publication: January 2004 Newsletter - Are Good Inputs Enough? No.
Investment institutions tend to focus the bulk of their human and capital resources on developing reliable forecasts of asset risks and return while ignoring the optimization technology they use to transform their information into investors' portfolios. These good inputs are no better than bad if the portfolios that represent the information to the investor have no investment value.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: October 2003 Newsletter. - Resampled Efficiency For Financial Planning and Return Forecasting
An examination of the effect of forecast certainty level indicate that the enormous effort focused on input estimation by many managers and institutions without Resampled Efficient Optimization is misplaced and likely to be ineffective.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: August 2003 Newsletter - Optimal and Investable Portfolios
Optimal portfolios typically include inconvenient and insignificant asset weights, make for impractical investment. This article introduces some of New Frontier's compute-efficient solutions for finding an investable portfolio from the optimal portfolio.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: June 2003 Newsletter - Letters to the Editor: Portfolio Resampling: Review and Critique
The authors responded immediately to Bernd Scherer’s critique of Resampled Efficient Optimization.
Authors: Richard O. Michaud & Robert O. Michaud.
Publication: Financial Analysts Journal. May/June 2003. - Resampled Efficiency Issues
Once resampled efficiency gained prominence, misinterpretations and misunderstandings arose. This 2003 article addresses the most frequently asked questions and misunderstandings of that time period.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: February 2003 Newsletter. - Liquidity and Portfolio Optimization
Liquidity, within the context of defining an optimal portfolio of risky assets, may be viewed as a non-linear return penalty factor that depends on the level of investment and asset size or float.
Authors: Richard O. Michaud & Robert O. Michaud
Publication: Second Quarter 2003 Newsletter - Letters to the Editor: 'An Examination of Resampled Portfolio Efficiency': A Comment
Dr. Michaud responds to the Fletcher & Hillier article (2001) that compared the performance of mean variance efficient asset allocations to resampled efficient asset allocations by means of a back test.
Author: Richard O. Michaud
Publication: Financial Analysts Journal. January/February 2003 - Resampled Portfolio Rebalancing and Monitoring
An enhancement to the original rebalancing procedure is now available that dramatically increases the uniformity and discrimination power of the original portfolio rebalancing and asset weight range procedures.
Author: Richard O. Michaud
Publication: Fourth Quarter 2002 Newsletter - An Introduction to Resampled Efficiency
Resampled Efficiency provides the solution to using uncertain information in portfolio optimization.
Author: Richard O. Michaud.
Publication: Investment Management Consulting Association's Monitor, September 2002
Third Quarter 2002 Newsletter - A Better Way to Use Information
In the second resampled efficiency article in the European Pensions News, Richard Michaud argues for the importance of taking statistical errors into account in asset allocation decisions.
Author: Richard O. Michaud.
Publication:European Pensions & Investment NewsJuly 9, 2001. - Out-of-Sample Tests of Resampled Efficiency
Resampled Efficient Optimization improves the average reward-to-risk ratio of classical asset allocation portfolios.
Author: Richard O. Michaud
Publication: European Pensions & Investment News. June 25, 2001 - Aspects: Resampled Efficient Asset Allocation
This article introduces the advantages and functionality of the newly patented Resampled Efficiency.
Author: Richard O. Michaud
Publication: Frontier News. Second Quarter 2001 - A New Design for Portfolios
Given all the problems associated with the inferior investment technology currently being used, it is little wonder that capital markets appear to be efficient. Only when asset management practice has achieved a level of sophistication consistent with the thoughtful use of investment information is it likely to provide statistically significant risk-adjusted performance.
Author: Richard O. Michaud
Publication: Bloomberg Personal Finance. July/August 2000 - New View of Mean Variance
This article discusses five methods other than mean variance optimization for defining portfolio optimality: non-variance risk measures, utility function optimization, multi-period objectives, Monte Carlo financial planning, or linear programming.
Author: Richard O. Michaud
Publication: Financial Planning Magazine. November 1, 1998 - The Markowitz Optimization Enigma: Is Optimized Optimal?
The major problem with mean variance optimization is its tendency to maximize the effects of errors in the input assumptions. Unconstrained mean variance optimization can yield results that are inferior to those of simple equal-weighting schemes. Author: Richard O. Michaud
Publication: Financial Analysts Journal. January/February 1989