Journal of Investment Strategies

Ali Hirsa
Professor, Columbia University & Managing Partner, Sauma Capital LLC

Welcome to the first issue of the twelfth volume of The Journal of Investment Strategies, which contains three research papers.

In the first paper of this issue, “The realized local volatility surface”, Yuming Ma, Shintaro Sengoku and Kazuhide Nakata present a novel Bayesian nonparametric approach for estimating local volatility, which, unlike Dupire’s stochastic local volatility method, involves fitting a Gaussian mixture model. The accurate estimation of a realized volatility distribution is crucial for formulating an options trading strategy and for portfolio risk management. The key contribution of Ma et al’s paper lies in its proposal to use a Bayesian nonparametric estimation technique to reconstruct a hypothetical generalized Wiener measure using historical price data. To achieve this, they employ a stick-breaking Gaussian mixture model (SB-GMM) to compensate for the return distribution across both time and price dimensions. The SB-GMM parameters are set using the Hamiltonian Monte Carlo method to estimate the maximum a posteriori probability. Samples are then drawn from the posterior distribution to compute the standard deviation and construct the realized local volatility surface with a 95% credible interval. Ma et al also include a numerical experiment using high-frequency ticker-level data for Tesla, Inc., to demonstrate the construction of a counterfactual realized local volatility surface. They also highlight the potential use of the realized volatility surface in compensating for implied volatility in hedging, thereby managing risks associated with sudden movements in the underlying price.

The issue’s second paper, “Sherman ratio optimization: constructing alternative ultrashort sovereign bond portfolios” by Karim Henide, explores the construction of duration-matched portfolios using a benchmark ICE BoA 0–2 Year AAA Euro Government Index (EG1E) universe. Henide’s aim is to optimize the portfolio’s Sherman ratio: a measure that indicates the portfolio’s potential for maximizing nominal yield for hold-to-maturity investors. He demonstrates that a dynamic strategy (which maintains a duration exposure identical to the benchmark) can deliver better total return performance than the benchmark despite having virtually homogenous risk exposures. His research indicates that Sherman ratio optimization has potential in portfolio construction. Further, Henide suggests that combining Sherman ratio optimization with constraints on individual bond weights can be an effective strategy. This approach allows the strategy’s risk–return profile to be adjusted to better align with investor utility, offering a more pragmatic, post-Markowitz perspective on portfolio construction. This focus is particularly significant in the context of systematic portfolio construction for virtually homogenous ultra-short sovereign fixed income (or “collateral”).

In our third and final paper, “Trading robots and financial markets trading solutions: the role of experimental economics”, Bianca Benedicto, Mara Madaleno and Anabela Botelho delve into the invaluable insights provided by experimental studies on financial markets, with a specific focus on automated trading strategies. They discover that, in most cases, trading robots outperform human investors in terms of profits, except when these robots operate at slower speeds or during periods of heightened market instability. They also sound a note of caution about the potential for faster, more advanced trading algorithms to exploit market manipulation strategies, thereby decreasing market efficiency. Interestingly, they point out a gap in the scientific literature: a lack of laboratory experiments evaluating cognitive and behavioral biases in human–robot interactions. Their study serves as a launch pad for this new and promising research direction. Through a meticulous review of the literature concerning the interaction between automated trading robots and human traders, Benedicto et al reach three key conclusions: by comparing human and robot performances using market conditions as a control variable, they determine that human traders generally underperform robots in terms of profit, barring significant market instability; they find that experimental approaches, unlike empirical field studies, allow for direct interaction between human investors and trading algorithms; and, via experimental approaches using the presence of trading robots and the speed of the trading algorithm as control variables, they conclude that trading algorithms can resort to market manipulation strategies, thus reducing market efficiency. Overall, Benedicto et al shed light on the intriguing interplay between humans and trading algorithms, hinting at future research opportunities to enhance market efficiency and fairness.

On behalf of the editorial board, I would like to thank you, our readers, for your continued support of and keen interest in the journal. We look forward to sharing with you the growing list of practical papers on a wide variety of topics on modern investment strategies that we continue to receive from both academics and practitioners.