SSRN Viewer

These slides are part of the Data & Ai reskilling seminar in Banking hosted by Dauphine and Ecoles des Mines PSL. They provide tips & tricks for classification using logistic regression, decision trees and neural networks

These slides are part of the Data & Ai reskilling seminar in Banking hosted by Dauphine and Ecoles des Mines PSL. They provide tips & tricks for similarities with unsupervised learning

Virtually all journal articles in the factor investing literature make associational claims, in denial of the causal content of factor models. Authors do not identify the causal graph consistent with the observed phenomenon, they justify their chosen model specification in terms of correlations, and they do not propose experiments for falsifying causal mechanisms. Absent a causal theory, their findings are likely false, due to rampant backtest overfitting and incorrect specification choices.I differentiate between type-A and type-B spurious claims, and explain how both types prevent factor investing from advancing beyond its current phenomenological stage. This monograph analyzes the current state of causal confusion in the factor investing literature, and proposes solutions with the potential to transform factor investing into a truly scientific discipline.

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. Results can be validated using the Python code in the Appendix.

These slides are part of the Data & Ai reskilling seminar in Banking hosted by Dauphine and Ecoles des Mines PSL. They discuss how to interpret machine learning models with Shapley values

Linear and compounded returns are at times used interchangeably: this practice has dangerous repercussions on risk and portfolio management.

Cryptocurrencies are among the largest unregulated markets in the world. We find that approximately one-quarter of bitcoin users are involved in illegal activity. We estimate that around $76 billion of illegal activity per year involves bitcoin (46% of bitcoin transactions), which is close to the scale of the US and European markets for illegal drugs. The illegal share of bitcoin activity declines with mainstream interest in bitcoin and with the emergence of more opaque cryptocurrencies. The techniques developed in this paper have applications in cryptocurrency surveillance. Our findings suggest that cryptocurrencies are transforming the black markets by enabling “black e-commerce.”

Berkshire Hathaway has realized a Sharpe ratio of 0.79 with significant alpha to traditional risk factors. However, the alpha becomes insignificant when controlling for exposures to Betting-Against-Beta and Quality-Minus-Junk factors. Further, we estimate that Buffett’s leverage is about 1.7-to-1 on average. Therefore, Buffett’s returns appear to be neither luck nor magic, but, rather, reward for leveraging cheap, safe, quality stocks. Decomposing Berkshires’ portfolio into ownership in publicly traded stocks versus wholly-owned private companies, we find that the former performs the best, suggesting that Buffett’s returns are more due to stock selection than to his effect on management.

Financial statements of major money-center commercial banks increasingly include reference to a credit valuation adjustment (CVA), debit (or debt) valuation adjustment (DVA), and funding valuation adjustment (FVA). This article explains the concepts behind CVA, DVA, and FVA using examples of interest rate swap valuation. A binomial forward rate tree model is used to get the value of the swap assuming no default. The CVA (the credit risk of the counterparty) and the DVA (the credit risk of the entity itself) depend on assumptions about the probability of default, the recovery rate and the expected exposure, which depends of projected values and settlement payments for the swap. The FVA arises when an uncollateralized swap is hedged with a collateralized or centrally cleared contract. In this version of the paper, two methods to calculate FVA are shown, both using the same assumptions about the credit risk parameters for the bank.

The efficient markets theory reached the height of its dominance in academic circles around the 1970s. Faith in this theory was eroded by a succession of discoveries of anomalies, many in the 1980s, and of evidence of excess volatility of returns. Finance literature in this decade and after suggests a more nuanced view of the value of the efficient markets theory, and, starting in the 1990s, a blossoming of research on behavioral finance. Some important developments in the 1990s and recently include feedback theories, models of the interaction of smart money with ordinary investors, and evidence on obstacles to smart money.

All too often, the concept of risk-neutral probabilities in mathematical finance is poorly explained, and misleading statements are made. The aim of this paper is to provide an intuitive understanding of risk-neutral probabilities, and to explain in an easily accessible manner how they can be used for arbitrage-free asset pricing. The paper is meant as a stepping-stone to further reading for the beginning graduate student in finance.

There exist two separate branches of finance that require advanced quantitative techniques: the "Q" area of derivatives pricing, whose task is to "extrapolate the present"; and the "P" area of quantitative risk and portfolio management, whose task is to "model the future." We briefly trace the history of these two branches of quantitative finance, highlighting their different goals and challenges. Then we provide an overview of their areas of intersection: the notion of risk premium; the stochastic processes used, often under different names and assumptions in the Q and in the P world; the numerical methods utilized to simulate those processes; hedging; and statistical arbitrage.

We propose a unified, fully general methodology to analyze and act on diversification in any environment, including long-short trades in highly correlated markets with complex derivatives. First, we build the diversification distribution, i.e. the distribution of the uncorrelated bets in the portfolio that are consistent with the portfolio constraints. Next, we summarize the wealth of information provided by the diversification distribution into one single diversification index, the effective number of bets, based on the entropy of the diversification distribution. Then, we introduce the mean-diversification efficient frontier, a diversification approach to portfolio optimization. Finally, we describe how to perform mean-diversification optimization in practice in the presence of transaction and market impact costs, by only trading a few optimally chosen securities. Fully documented code illustrating our approach can be downloaded from MATLAB Central File Exchange.

We survey 392 CFOs about the cost of capital, capital budgeting, and capital structure. Large firms rely heavily on net present value techniques and the capital asset pricing model, while small firms are relatively likely to use the payback criterion. Older executives without an MBA are more likely to rely on payback than are younger executives with an MBA. Surprisingly, most companies use a single company-wide discount rate to evaluate a project in a new industry and country. In addition to market risk, firms also frequently adjust cash flows or discount rates for interest rate risk, exchange rate risk, business cycle risk, and inflation risk. Few firms adjust discount rates or cash flows for book-to-market, distress, or momentum risks. A majority of large firms have a tight or somewhat tight target debt ratio, in contrast to only one-third of small firms. Executives rely heavily on informal rules when choosing capital structure. The most important factors affecting debt policy are maintaining financial flexibility and having a good credit rating. When issuing equity, respondents are concerned about earnings per share dilution and recent stock price appreciation. We find some support for the pecking-order and trade-off capital structure hypotheses but little evidence that executives are concerned about asset substitution, asymmetric information, transactions costs, free cash flows, or personal taxes. If CFOs behave according to these deeper hypotheses, they apparently do so unknowingly.

Peer review is fundamental to the efficacy of the scientific process. We draw from our experience both as editors, authors and association representatives to provide a set of guidelines for referees in preparing their reports and cover letters to journal editors. While our document is directed to anyone asked to review a paper, our suggestions are especially relevant for new members of the profession when preparing their first reports. The Appendix to this document contains a Checklist. The Checklist is also available here. Also see our companion paper, How to Write an Effective Referee Report and Improve the Scientific Review Process, which includes comments by former editors of economics and finance journals.

During the week of August 6, 2007, a number of quantitative long/short equity hedge funds experienced unprecedented losses. Based on TASS hedge-fund data and simulations of a specific long/short equity strategy, we hypothesize that the losses were initiated by the rapid unwind of one or more sizable quantitative equity market-neutral portfolios. Given the speed and price impact with which this occurred, it was likely the result of a forced liquidation by a multi-strategy fund or proprietary-trading desk, possibly due to a margin call or a risk reduction. These initial losses then put pressure on a broader set of long/short and long-only equity portfolios, causing further losses by triggering stop/loss and de-leveraging policies. A significant rebound of these strategies occurred on August 10th, which is also consistent with the unwind hypothesis. This dislocation was apparently caused by forces outside the long/short equity sector - in a completely unrelated set of markets and instruments - suggesting that systemic risk in the hedge-fund industry may have increased in recent years.

Economic Value Added (EVA®) has attracted considerable attention as an alternative to traditional accounting earnings for use in both valuation and incentive compensation. With a host of consultants now marketing related metrics, numerous claims have been made – most based on anecdotal evidence or in-house studies. This paper summarizes independent evidence regarding EVA’s alleged advantages. We begin by reviewing the theory that links the underlying concept of residual income to shareholder value. Second, we discuss how Stern Stewart modifies residual income to produce its proprietary EVA metric and show how median EVA compares with residual income, net income and operating cash flows over the period 1988-97. Third, we examine the claim that EVA is more closely associated with stock returns and firm value than is net income. The evidence indicates that EVA does not dominate net income in associations with stock returns and firm values. Fourth, we examine a second claim that compensation plans based on residual income motivate managers to take actions consistent with increasing shareholder value. Here, the independent evidence suggests that managers do respond to residual income-based incentives. Finally, we discuss how a metric such as EVA can be useful for internal incentive purposes even if it conveys little news to market participants regarding the firm’s valuation.

In this paper we examine global tactical asset allocation (GTAA) strategies across a broad range of asset classes. Contrary to market timing for single asset classes and tactical allocation across similar assets, this topic has received little attention in the existing literature. Our main finding is that momentum and value strategies applied to GTAA across twelve asset classes deliver statistically and economically significant abnormal returns. For a long top-quartile and short bottom-quartile portfolio based on a combination of momentum and value signals we find a return exceeding 9% per annum over the 1986-2007 period. Performance is stable over time, also present in an out-of-sample period and sufficiently high to overcome transaction costs in practice. The return cannot be explained by implicit beta exposures or the Fama French and Carhart hedge factors. We argue that financial markets may be macro inefficient due to insufficient 'smart money' being available to arbitrage mispricing effects away.

The Kelly criterion is a money management principle that beats any other approach in many respects. In particular, it maximizes the expected growth rate and the median of the terminal wealth. However, until recently application of the Kelly criterion to multivariate portfolios has seen little analysis. We briefly introduce the Kelly criterion and then present its multivariate version based only on the first and the second moments of the asset excess returns. Additionally, we provide a simple numerical algorithm to manage virtually arbitrarily large portfolios according to so-called fractional Kelly strategies.

We study the influence of financial innovation by fintech brokerages on individual investors’ trading and stock prices. Using data from Robinhood, we find that Robinhood investors engage in more attention-induced trading than other retail investors. For example, Robinhood outages disproportionately reduce trading in high-attention stocks. While this evidence is consistent with Robinhood attracting relatively inexperienced investors, we show that it can also be partially driven by the app’s unique features. Consistent with models of attention-induced trading, intense buying by Robinhood users forecast negative returns. Average 20-day abnormal returns are -4.7% for the top stocks purchased each day.

In corporate finance and valuation, we start off with the presumption that the riskfree rate is given and easy to obtain and focus the bulk of our attention on estimating the risk parameters of individuals firms and risk premiums. But is the riskfree rate that simple to obtain? Both academics and practitioners have long used government security rates as riskfree rates, though there have been differences on whether to use short term or long-term rates. In this paper, we not only provide a framework for deciding whether to use short or long term rates in analysis but also a roadmap for what to do when there is no government bond rate available or when there is default risk in the government bond. We look at common errors that creep into valuations as a consequence of getting the riskfree rate wrong and suggest a way in which we can preserve consistency in both valuation and capital budgeting.

This report describes compound return outcomes for the 29,078 publicly-listed common stocks contained in the CRSP database from December 1925 to December 2023. The majority (51.6%) of these stocks had negative cumulative returns. However, the investment performance of some stocks was remarkable. Seventeen stocks delivered cumulative returns greater than five million percent (or $50,000 per dollar initially invested), with the highest cumulative return of 265 million percent (or $2.65 million per dollar initially invested) accruing to long-term investors in Altria Group. Annualized compound returns to these top performers were relatively modest, averaging 13.47% across the top seventeen stocks, thereby affirming the importance of "time in the market." The highest annualized compound return for any stock with at least 20 years of return data was 33.38%, earned by Nvidia shareholders.

The purpose of this report is to review the evidence on the profitability of technical analysis. The empirical literature is categorized into two groups, "early" and "modern" studies, according to the characteristics of testing procedures. Early studies indicated that technical trading strategies were profitable in foreign exchange markets and futures markets, but not in stock markets before the 1980s. Modern studies indicated that technical trading strategies consistently generated economic profits in a variety of speculative markets at least until the early 1990s. Among a total of 92 modern studies, 58 studies found positive results regarding technical trading strategies, while 24 studies obtained negative results. Ten studies indicated mixed results. Despite the positive evidence on the profitability of technical trading strategies, it appears that most empirical studies are subject to various problems in their testing procedures, e.g., data snooping, ex post selection of trading rules or search technologies, and difficulties in estimation of risk and transaction costs. Future research must address these deficiencies in testing in order to provide conclusive evidence on the profitability of technical trading strategies.

Machine learning (ML) is changing virtually every aspect of our lives. Today ML algorithms accomplish tasks that until recently only expert humans could perform. As it relates to finance, this is the most exciting time to adopt a disruptive technology that will transform how everyone invests for generations.

We propose a new latent factor conditional asset pricing model. Like Kelly, Pruitt, and Su (KPS, 2019), our model allows for latent factors and factor exposures that depend on covariates such as asset characteristics. But, unlike the linearity assumption of KPS, we model factor exposures as a flexible nonlinear function of covariates. Our model retrofits the workhorse unsupervised dimension reduction device from the machine learning literature—autoencoder neural networks—to incorporate information from covariates along with returns themselves. This delivers estimates of nonlinear conditional exposures and the associated latent factors. Furthermore, our machine learning framework imposes the economic restriction of no-arbitrage. Our autoencoder asset pricing model delivers out-of-sample pricing errors that are far smaller (and generally insignificant) compared to other leading factor models.