Sebastian Ceria (Axioma)
The fundamental premise of quantitative portfolio construction is to balance the risk return tradeoff wherein an alpha factor model is assigned the task of predicting returns and a factor risk model is entrusted the duty to controlling risk. Naturally, the interaction between the alpha and risk factors plays an important role determining the composition of the resulting portfolios prompting the investment community to ask the question: Do risk models eat alpha? In this talk we analyze the mechanics of portfolio construction based on Markowitz MVO framework, and analyze client data to conclude with an unambiguous answer, namely, misaligned risk models eat alpha while custom risk models, that do include all the factors that are part of the alpha factor model, do not. We give empirical results to corroborate our findings.
Paul Glasserman (Columbia University, Graduate School of Business)
Portfolio selection is vulnerable to the error-amplifying effects of combining optimization with statistical estimation and model error. For dynamic portfolio control, sources of model error include the evolution of market factors and the influence of these factors on asset returns. We develop portfolio control rules that are robust to this type of uncertainty, applying a stochastic notion of robustness to uncertainty in model dynamics. In this stochastic formulation, robustness reflects uncertainty about the probability law generating market data, and not just uncertainty about model parameters. The model incorporates transaction costs and leads to simple and tractable optimal robust controls for multiple assets. We illustrate the performance of the controls on commodity futures data and find that robustness improves performance in out-of- sample tests. (This is joint work with Xingbo Xu.)
Garud Iyengar (Columbia University, Department of IE & OR)
We propose an iterative algorithm to eciently solve the portfolio selection problem with multiple spectral risk constraints. Since the conditional value at risk (CVaR) is a special case of the spectral risk function, our algorithm solves portfolio selection problems with multiple CVaR constraints. In each step, the algorithm solves a very simple separable convex quadratic program. The algorithm extends to the case where the objective is a utility function with mean return and a weighted combination of a set of spectral risk constraints, or maximum of a set of spectral risk functions. We report numerical results that show that our proposed algorithm is very efficient, and is at least two orders of magnitude faster than the state of the art general purpose solver for all practical instances. (This is joint work with Carlos Abad.)
Anthony Lazanas (Barclays Capital)
Title: Investing with Risk Premia Factors: From Theory to Practice
Portfolio construction practitioners in find it increasingly difficult to navigate an ever-expanding set of sources of returns and portfolio construction methods. In this presentation we discuss factor-based investing, analyze potential sources of returns and place portfolio construction methods in a unified framework. While maximization of long-term Sharpe or Sortino ratio is the target of many proposed methods, in practice the actual utility function of most portfolio managers is different, and this can lead to sub-optimal allocations. We present an example of a tradable diversified portfolio and discuss practical issues of its implementation.
Andrew Lo (MIT, Sloan School of Management)
Title: Can Financial Engineering Cure Cancer?
Biomedical innovation has become riskier, more expensive and more difficult to finance with traditional sources such as private and public equity. In this talk, Professor Lo will describe a new financial structure in which a large number of biomedical programs at various stages of development are funded by a single entity to substantially reduce the portfolio’s risk. The portfolio entity can finance its activities by issuing debt, a critical advantage because a much larger pool of capital is available for investment in debt versus equity. By employing financial engineering techniques such as securitization, it can raise even greater amounts of more-patient capital. In a simulation using historical data for new molecular entities in oncology from 1990 to 2011, he and his co-authors find that megafunds of $5–15 billion may yield average investment returns of 8.9–11.4% for equity holders and 5–8% for ‘research-backed obligation’ holders, which are lower than typical venture-capital hurdle rates but attractive to pension funds, insurance companies and other large institutional investors.
Marcos López de Prado (Tudor Investment Corporation)
We evaluate the probability that an estimated Sharpe ratio exceeds a given threshold in presence of non-Normal returns. We show that this new uncertainty-adjusted investment skill metric (called Probabilistic Sharpe ratio, or PSR) has a number of important applications: First, it allows us to establish the track record length needed for rejecting the hypothesis that a measured Sharpe ratio is below a certain threshold with a given confidence level. Second, it models the trade-off between track record length and undesirable statistical features (e.g., negative skewness with positive excess kurtosis). Third, it explains why track records with those undesirable traits would benefit from reporting performance with the highest sampling frequency such that the IID assumption is not violated. Fourth, it permits the computation of what we call the Sharpe ratio Efficient Frontier (SEF), which lets us optimize a portfolio under non-Normal, leveraged returns while incorporating the uncertainty derived from track record length.
Attilio Meucci (Kepos Capital)
We introduce a new framework to integrate liquidity risk, funding risk and market risk, which goes beyond the simple bid-ask spread overlay to a VaR number. In our approach, we overlay a whole distribution of liquidity uncertainty to each future market-risk scenario. Then we allow for the liquidity uncertainty to vary scenario by scenario, depending on what liquidation policy or funding policy is implemented in that scenario. The result is one easy-to-interpret and easy-to-implement formula for the total liquidity-plus-market-risk P&L distribution. Using this formula we can stress-test different market risk P&L distributions and different scenario-dependent liquidation policies and funding policies; compute total risk and decompose it into a novel liquidity-plus-market risk formula; and define a liquidity score as a monetary measure of portfolio liquidity. Our approach relies on three pillars: first, the literature on optimal execution, to model liquidity risk as a function of the actual trading involved; second, an analytical conditional convolution, to blend market risk and liquidity/funding risk; third the Fully Flexible Probabilities framework, to model and stress-test market risk even in highly non-normal portfolios with complex derivatives. Our approach can be implemented efficiently with portfolios of thousand of securities. The code for the case study is available for download.