Sanjiv Das - Academia.edu (original) (raw)

Papers by Sanjiv Das

International Portfolio Choice Returns on international equities are characterized by jumps; more... more International Portfolio Choice Returns on international equities are characterized by jumps; moreover, these jumps tend to occur at the same time across countries leading to systemic risk. We capture these stylized facts using a multivariate system of jump-diffusion processes where the arrival of jumps is simultaneous across assets. We then determine an investor’s optimal unconditional portfolio for this model of returns. Systemic risk has two effects: it reduces the gains from diversification and also penalizes investors for holding levered positions. We find that the loss resulting from

We develop a market-wide illiquidity risk factor based on run lengths and find that it is priced ... more We develop a market-wide illiquidity risk factor based on run lengths and find that it is priced using standard asset-pricing specifications. Our theoretical frame-work of equity returns derives the result that average run lengths of individual stocks proxy for illiquidity, and are related to common measures of liquidity such as trading volume and trade price-impact. This relationship holds irrespective of the sampling frequency in the computation of run lengths. Thus, liquidity can be quantified by ex-amining a stock’s run length signature, providing a statistical mechanics link across illiquidity metrics. Tests using daily equity return data for all stocks over the period 1962-2005 find that run lengths are decreasing in turnover, and increasing with bid-ask spreads, and price-impact. Illiquidity is shown to be a risk factor/characteristic in explaining equity returns.

Fast and Curious: VC Drift Rapid changes in investment behavior offers a VC an opportunity to lea... more Fast and Curious: VC Drift Rapid changes in investment behavior offers a VC an opportunity to learn but carries potential costs such as dilution of skills. These benefits and costs may have ramifications not just for a specific investment but importantly for a VC’s entire portfolio. To capture these ideas, we first locate each VC financing round in one of twenty styles, and develop a measure of change in a VC’s investment styles (“style drift”) at the level of her portfolio. We find that drift is more likely among younger VCs, those without a focus on early stage investments and those who experience the pressure to invest their funds. Style drift is associated with poorer performance among seasoned VCs who are likely to have already developed expertise through past investments, and for VCs who drift in a correlated fashion (herd) with other VC firms. Finally, the more recent investments in the VC’s portfolio are more adversely affected when the VC drifts than other investments in th...

Journal of Investment Strategies, 2017

Journal of Risk and Financial Management, 2021

This paper considers investors who are looking to maximize their probability of remaining solvent... more This paper considers investors who are looking to maximize their probability of remaining solvent throughout their lifetime by using an algorithm that aims to optimize their investment allocation strategy and optimize their tax strategy for withdrawal allocations between tax deferred accounts (TDAs), Roth accounts, and taxable stock and bond accounts. This optimization works with stochastic investment returns and stochastic mortality, extending and combining different investment and tax-efficiency paradigms. We find that optimizing the investment strategy has a much larger impact on the investor remaining solvent than optimizing the tax strategy. This result is key to effectively optimizing both strategies simultaneously. This optimized investment strategy soundly beats a standard target date fund strategy, and the novel optimized tax strategy displays optimal desired properties suggested by non-stochastic tax optimization research.

ArXiv, 2021

With the ever-increasing complexity of neural language models, practitioners have turned to metho... more With the ever-increasing complexity of neural language models, practitioners have turned to methods for understanding the predictions of these models. One of the most welladopted approaches for model interpretability is feature-based interpretability, i.e., ranking the features in terms of their impact on model predictions. Several prior studies have focused on assessing the fidelity of feature-based interpretability methods, i.e., measuring the impact of dropping the top-ranked features on the model output. However, relatively little work has been conducted on quantifying the robustness of interpretations. In this work, we assess the robustness of interpretations of neural text classifiers, specifically, those based on pretrained Transformer encoders, using two randomization tests. The first compares the interpretations of two models that are identical except for their initializations. The second measures whether the interpretations differ between a model with trained parameters an...

In this article we propose a new approach for implementing option pricing models in finance. Fina... more In this article we propose a new approach for implementing option pricing models in finance. Financial engineers typically prototype such models in an interactive language (such as Matlab) and then use a compiled language such as C/C++ for production systems. Code is therefore written twice. In this article we show that the Python programming language and the Cython compiler allows prototyping in a Matlab-like manner, followed by direct generation of optimized C code with very minor code modifications. The approach is able to call upon powerful scientific libraries, uses only open source tools, and is free of any licensing costs. We provide examples where Cython speeds up a prototype version by over 500 times. These performance gains in conjunction with vast savings in programmer time make the approach very promising.

The financial crisis of 2008 highlighted the absence of metrics for measuring, decomposing, manag... more The financial crisis of 2008 highlighted the absence of metrics for measuring, decomposing, managing, and predicting systemic risk. Systemic risk is interpreted as a risk that has (a) large impact, (b) is widespread, i.e., affects a large number of entities or institutions, and (c) has a ripple effect that endangers the existence of the financial system. Whereas there is now a wide-ranging literature on systemic risk in the US, there is little work on other financial systems, especially not in countries very different from the US. In this project, we undertake a large-scale empirical examination of systemic risk among major financial institutions in a large sample of 23 emerging markets. We present a novel systemic risk score for each financial system by region. This score is a per-bank, size-weighted, and network-weighted credit risk measure that may be compared across geographical regions, and across time. It is also additively decomposable and attributable to each financial insti...

The Journal of Finance and Data Science, 2021

The Journal of Financial Data Science, 2021

Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, 2021

SSRN Electronic Journal, 2020

Journal of Economic Dynamics and Control, 2018

SSRN Electronic Journal, 2019

We develop a dynamic programming methodology that seeks to maximize investor outcomes over multip... more We develop a dynamic programming methodology that seeks to maximize investor outcomes over multiple, potentially competing goals (such as upgrading a home, paying college tuition, or maintaining an income stream in retirement), even when financial resources are limited. Unlike Monte Carlo approaches currently in wide use in the wealth management industry, our approach uses investor preferences to dynamically make the optimal determination for fulfilling or not fulfilling each goal and for selecting the investor’s investment portfolio. This can be computed quickly, even for numerous investor goals spread over different or concurrent time periods, where each goal may allow for partial fulfillment or be all-or-nothing. The probabilities of attaining each (full or partial) goal under the optimal scenario are also computed, so the investor can ensure the algorithm accurately reflects their preference for the relative importance of each of their goals. These portfolio prescriptions are consistent with Prospect Theory.

Journal of Financial Services Research, 2018

SSRN Electronic Journal, 2018

We report a dynamic programming algorithm which, given a set of efficient (or even inefficient) p... more We report a dynamic programming algorithm which, given a set of efficient (or even inefficient) portfolios, constructs an optimal portfolio trading strategy that maximizes the probability of attaining an investor’s specified target wealth at the end of a designated time horizon. Our algorithm also accommodates periodic infusions or withdrawals of cash with no degradation to the dynamic portfolio’s performance or runtime. We explore the sensitivity of the terminal wealth distribution to restricting the segment of the efficient frontier available to the investor. Since our algorithm’s optimal strategy can be on the efficient frontier and is driven by an investor’s wealth and goals, it soundly beats the performance of target date funds in attaining investors’ goals. These optimal goals-based wealth management strategies are useful for independent financial advisors to implement behavioral-based FinTech offerings and for robo-advisors.

The Journal of Financial Data Science, 2019

In this article, the authors propose a theory-driven framework for monitoring system-wide risk by... more In this article, the authors propose a theory-driven framework for monitoring system-wide risk by extending data science methods widely deployed in social networks. Their approach extends the one-firm Merton credit risk model to a generalized stochastic network-based framework across all financial institutions, comprising a novel approach to measuring systemic risk over time. The authors identify four desired properties for any systemic risk measure. They also develop measures for the risks created by each individual institution and a measure for risk created by each pairwise connection between institutions. Four specific implementation models are then explored, and brief empirical examples illustrate the ease of implementation of these four models and show general consistency among their results.

We examine the use of deep learning (neural networks) to predict the movement of the S&P 500 Inde... more We examine the use of deep learning (neural networks) to predict the movement of the S&P 500 Index using past returns of all the stocks in the index. Our analysis finds that the future direction of the S&P 500 index can be weakly predicted by the prior movements of the underlying stocks in the index. Decomposition of the prediction error indicates that most of the lack of predictability comes from randomness and only a little from nonstationarity. We believe this is the first test of S&P500 market efficiency that uses a very large information set, and it extends the domain of weak-form market efficiency tests.