Abstract:

After reviewing some differences between traditional statistics and data science, we present a modular machine learning framework for model validation which blends the two paradigms. Model validation is set up as a sequence of procedures, in which the output from one procedure serves as the input to another procedure within a single validation framework. An econometric model is used in the first module to classify data in an economically intuitive way. Proceeding modules apply data science techniques to evaluate the predictive characteristics of the model components. We apply the framework to the fundamental law of active management, a well-known formal characterization of portfolio managers’ alpha generation process. In contrast to standard applications of the law, in which it has been used to evaluate a manager’s existing active management process, we recast the law within his framework as a means to test investment signals for potential use, individually or collectively, in a manager’s investment process. To illustrate how this application works, we provide an example using the well-known Fama–French factors as test signals.

Bio:

Joseph Simonian is the founder and CIO, Autonomous Investment Technologies, LLC and co-editor of the Journal of Financial Data Science. Previously Joe held the roles of Senior Investment Strategist at Acadian Asset Management and Director of Quantitative Research at Natixis Investment Managers, where he led the quantitative research and portfolio strategy for the Portfolio Research and Consulting Group. He was also a member of the investment oversight committee. Prior to working at Natixis, Joseph was the Principal Research Analyst at Global Institutional Solutions. He was also the Vice President of Portfolio Management at J.P. Morgan and PIMCO. Joseph gained his PhD from University of California, Santa Barbara, MA from Columbia University, New York, and BA from University of California, Los Angeles. Joseph is a noted contributor to leading finance journals and is also a prominent speaker at investment events worldwide. He is currently the co-editor of the Journal of Financial Data Science and Advisory Board member for the Journal of Portfolio Management and the Financial Data Professional Institute.

Abstract:

The proliferation of blockchain-based cryptocurrencies, which essentially use public accounting ledgers, has two opposing effects related to the global financial system's transparency. On the one hand, governments, market regulators, and financial institutions make significant efforts to curtail the financing of illicit activities that cryptocurrencies may be able to circumvent. On the other hand, funds and transfers that used to be known only by the involved parties are now transparent to anyone with knowledge of the economics of blockchain. This study empirically examines whether it is possible for outsiders to identify evidence of financing international terrorist attacks on public blockchain systems. We do so by utilizing empirical strategies based on finance, forensic accounting, and machine learning algorithms on millions of cryptocurrencies' transfers. We provide evidence that blockchain-based cryptocurrencies are used to finance terrorist attacks. However, the blockchain ledger underlying transparency also enables outsiders to identify the fund trails and predict terrorist attacks.

Bio:

Daniel Rabetti is a financial economist currently pursuing a P.hD in Business at the Coller School of Management, Tel Aviv University. Daniel research interest lies in the intersection of financial innovation and disclosure - as means to mitigate asymmetric information in alternative financing markets. He has been working on several academic research in the field of economics of blockchain, industrial organization, and disclosure in unregulated markets. Daniel work has been featured in prominent conferences in accounting, finance, computer science, blockchain, and economics, and his work is currently on the path for publication in distinguished peer-reviewed academic journals.