Welcome to the third issue of the fifth volume of The Journal of Investment Strategies. In this issue, we present four papers on portfolio construction methodologies, analysis of the factors that drive investment performance, and models of optimal trading.
In the first paper of this issue, François Soupé, Thomas Heckel and Raul Leote de Carvalho from BNP Paribas present a dynamic model of portfolio insurance with adaptive protection. Portfolio insurance, and investment products incorporating portfolio protection, have been around since their introduction by Leland, O'Brien and Rubinstein in the 1980s. Despite getting (not entirely deserved) criticism for contributing to the 1987 stock market crash, portfolio insurance has remained an important staple of long-term investment management and has led to the development of many innovative quantitative models designed to achieve the protection objectives while reducing various potential shortcomings. The paper from the BNP Paribas team focuses on one important aspect of these strategies, which is the setting of the level of protection for the portfolio. The authors correctly state that the conventional setting at 100% protection is not necessarily optimal, and that depending on investors'true long-term risk versus return preferences, the level of protection may be optimized. What they find is that this optimal level becomes dynamic, and adapts to both the past performance of the portfolio and the current risk projections till maturity. Their methodology, which they call PIWAP (portfolio insurance with adaptive protection), can be considered as a natural logical evolution of the well-known CPPI (constant proportional portfolio insurance) and is just as straightforward to implement. Given its apparently superior performance, many investors will find it more attractive.
In our second paper, Filippo Passerini and Samuel E.Vázquez present their studies on the problem of optimal trading under linear transient costs given alpha predictors. It is well known that the presence of linear costs introduces no-trading zones, where the expected profit from the alpha signal is not sufficient to overcome the cost of trading. Passerini and Vazquez find an intuitively appealing pattern that, when trading with both market and limit orders, the three-layer separation of trading zones (buy zone, no trading zone, sell zone) is replaced by a five-layer structure, where the inner zone becomes the market-making zone (ie, we trade essentially opposite to the alpha signal using the limit orders and with the objective of benefiting from the bid-offer spread); this is followed by having conservative limit order trading zones on the buy and sell sides (where we trade in the direction of the alpha signal, but only if our limit order is executed); and finally, we have aggressive market order trading zones on the buy and sell sides, where the alpha signals are sufficiently strong to allow us to pay for the bid-offer spread. I find these results very believable and would not hesitate to use this approach in practical trading. As always, the calibration of the specific model parameters to empirical market data would be key to the actual performance of the model.
In the issue's third paper, Boris Gnedenko and Igor Yelnik discuss the advantages and challenges of asset allocation using factor portfolios. They address an important area of modern portfolio theory that focuses on multifactor analysis of investment returns, which has become the lingua franca of quantitative investment managers. Factor analysis can be used for all of the phases of the investment process: security selection, portfolio construction, risk management and trading execution. Gnedenko and Yelnik's paper focuses on portfolio construction and associated aspects of risk management but attempts to solve these problems in a manner that is coherent with security selection. The authors introduce a clever two-step optimization methodology that solves the strategic asset allocation problem across factors during the first step and then modifies the asset weights to satisfy various portfolio constraints while minimizing the tracking error to an unconstrained optimal factor portfolio during the second step. The result is a portfolio that is aligned as closely as possible with the investor's strategic positioning and also respects the investor's risk, position and other such policy constraints. Although there is no guarantee that this results in the closest fit, as the two-step optimization may potentially be suboptimal when the second stage constraints are highly binding, we are sure that the result is quite close to optimal, and the practicality and the robustness of the approach more than make up for any residual deviations. Many portfolio managers would find this approach to be intuitive and would be well advised to include it in their toolkits.
The fourth paper in the issue, from the Capital Fund Management team of Jean-Philippe Bouchaud, Stefano Ciliberti, Augustin Landier, Guillaume Simon and David Thesmar, investigates one of the major known sources of equity excess returns: the so-called quality anomaly. A more modern addition to the classic Fama-French-Carhart set of factors that includes the value, size and momentum, the quality factor measures the stability of companies' profitability and earnings, and is one of the three more recent "alternative beta factors", which also include low volatility and low beta. Unlike the established academic view that the value and size factors represent persistent risk premia and are therefore consistent with the efficient market hypothesis, there is no such uniform theory regarding the next-generation alternative beta factors such as the quality. Bouchaud et al contend that the risk premium view of the quality factor is not well grounded in data. Instead, they put forward the behavioral view, which explains the excess returns by persistent suboptimal behavior of investors. Their explanation is based on the evidence of overly optimistic forecasts of analysts regarding low-quality stocks and insufficiently optimistic forecasts regarding higher-quality companies. Recent industry publications claim that the quality and low-volatility factors are not persistent and, moreover, are highly overvalued using conventional value metrics. The paper bridges this gap and establishes an alternative foundation for the persistence of the quality factor, in some ways resembling the behavioral explanation of the momentum factor.
I would like to thank our readers for their continued interest. I amsure they will find useful ideas in this issue of The Journal of Investment Strategies, and I look forward to presenting more such interesting papers in future issues.
Arthur M. Berd
General Quantitative LLC
This paper projects an optimal unconstrained factor portfolio onto a set of all feasible portfolios using tracking error as a distance measure.
This paper studies the problem of optimal trading using general alpha predictors with linear costs and temporary impact.
This paper investigates the optimal design of funds which provide capital protection at a specific maturity.
This paper investigates the causes of the quality anomaly by exploring two potential explanations - the “risk view” and the “behavioral view”.