Alternative Data Solutions Cookbook
Alternative Data which are non-traditional datasets used by both fundamental and quantitative institutional investors have become the hot topic on Wall St. As the funds dig deeper to enhance the investment process, today’s alternative data analysts face a new crop of technical challenges. We will discuss these challenges and existing tools and techniques to solve them. These include identifier mapping, stable panel creation, dataset evaluation and sensitive information extraction. Finally, we will briefly discuss alternative data in the context of the regulatory environment.
Can Swing Pricing Prevent Mutual Fund Runs and Fire Sales?
We develop a model of the feedback between mutual fund outflows and asset illiquidity. Alert investors anticipate the price impact of other investors’ redemptions and exit first at favorable prices. This first-mover advantage may lead to fund failure through a cycle of falling prices and increasing redemptions. We show that (i) the first-mover advantage introduces a nonlinear dependence between an exogenous market shock and the aggregate impact of redemptions on the asset price; (ii) there is a critical magnitude of the shock beyond which a run brings down the fund; (iii) properly designed swing pricing transfers liquidation costs from the fund to redeeming investors and, by removing the first-mover advantage, it reduces these costs and avoids fire sales. This is joint work with Agostino Capponi and Marko Weber.
How the Sharpe Ratio Died, and Came Back to Life
Marcos López de Prado
Selection bias under multiple backtesting makes it impossible to assess the probability that a strategy is false (Bailey et al. ). This has two implications:
1) “Most claimed research findings in empirical Finance are likely false” (Harvey et al. )
2) Most quantitative firms invest in false positives
Selection bias explains the high rate of failure among quantitative hedge funds: They do not have the technology to distinguish between a true strategy and a false strategy.
The goal of this presentation is to introduce such technology, so that academic journals, regulators and investors may discard false strategies with confidence.
The Omega Parameter: Multi-stage Black-Litterman Optimization with Statistical Forecasts
We investigate the theoretical underpinnings of the Black-Litterman model in order to better discern how the model may be generalized and used in applications beyond those for which it was originally intended. In particular, we represent the model as a special case of a class of statistical models known as Bayesian networks, and gain new insights on how to choose the elusive “omega parameter” which is meant to quantify the forecaster’s uncertainty. We also discuss how the model lends itself to a world in which machine learning and data science are of increasing importance.
Opportunities for Computational Value Added in Sustainable Finance
Sustainable finance, comprising the integration of environmental, social and governance information in security selection, impact investing, conservation finance, climate finance, the pricing of ecosystem services and related fields represent a growing segment of the financial intermediation process. The use of quantitative or computational techniques in these fields have been limited while the variety and complexity of salient and feasible data sources is increasing rapidly. The presentation provides a map of the sustainable finance eco-system and then outlines some pressing questions in these fields that represent promising avenues for computationally-inclined researchers and practitioners to address.
Risk Management of Large Option Portfolios via Monte Carlo Simulation with Applications to Central Clearing
We present statistical work done in the context of a consulting contract with The Options Clearing Corporation, regarding risk management of Clearing Participants’ option portfolios. The use of Big Data techniques to reduce the dimension of the market (500K to 1 million option contracts cleared daily) is emphasized throughout. We also describe a (related) compact software (ICEBERG) which can be used by market participants to estimate their option book risk profile (VaR, ES), accessing statistical risk-scenarios through the cloud.