SSRN Viewer

Value investing is the age-old investment strategy that involves buying securities that appear cheap relative to some fundamental anchor. For equity investors that anchor is typically a measure of intrinsic value linked to financial statement variables. Recently, there has been much written about the death of value investing. While undoubtedly many systematic approaches to value investing have suffered recently, we find the suggestion that value investing is dead to be premature. Both from a theoretical and empirical perspective, expectations of fundamental information have been and continue to be an important driver of security returns. We also address a series of critiques levelled at value investing and find them generally lacking in substance.

The merits of investing in private versus public equity have generated considerable debate, often fueled by concerns about data quality. In this paper, we use cash flow data derived from the holdings of almost 300 institutional investors to study over 1,800 North American buyout and venture capital funds. Average buyout fund returns for all vintage years but one before 2006 have exceeded those from public markets; averaging about 3% to 4% annually. Post-2005 vintage year returns have been roughly equal to those of public markets. We find similar performance results for a sample of almost 300 European buyout funds. Venture capital performance has varied substantially over time. North American venture funds from the 1990s substantially outperformed public equities; those from the early 2000s have underperformed; and recent vintage years have seen a modest rebound. The variation in venture performance is significantly linked to capital flows: performance is lower for funds started when there are large aggregate inflows of capital to the sector. We also examine the variation in performance of funds started in the same year. We find marked differences between venture and buyout leading to a much more pronounced impact of accessing high performing funds in venture investing.

We define “deep value” as episodes where the valuation spread between cheap and expensive securities is wide relative to its history. Examining deep value across global individual equities, equity index futures, currencies, and global bonds provides new evidence on competing theories for the value premium. Following these episodes, the value strategy has: (1) high average returns; (2) low market betas, but high betas to a global value factor; (3) deteriorating fundamentals; (4) negative news sentiment; (5) selling pressure; (6) increased limits to arbitrage; and (7) increased arbitrage activity. Lastly, we find that deep value episodes tend to cluster and a deep value trading strategy generates excess returns not explained by traditional risk factors.

Variable spending strategies can be situated on a continuum between two extremes: spending a constant amount from the portfolio each year without regard for the remaining portfolio balance, and spending a fixed percentage of the remaining portfolio balance. Variable spending strategies seek compromise between these extremes by avoiding too many spending cuts while also protecting against the risk that spending must subsequently fall to uncomfortably low levels. Two basic categories for variable spending rules explored include decision rule methods and actuarial methods. Ten strategies will be compared using a consistent set of portfolio return and fee assumptions, and using an XYZ formula to calibrate initial spending: the client willingly accepts an X% probability that spending falls below a threshold of $Y (in inflation-adjusted terms) by year Z of retirement. Presenting the distribution of spending and wealth outcomes for different strategies in which the initial spending rate is calibrated with the XYZ formula will allow for a more meaningful comparison of strategies. The article provides a framework for identifying appropriate spending strategies based on client preferences.

The safety of a 4% initial withdrawal strategy depends on asset return assumptions. Using historical averages to guide simulations for failure rates for retirees spending an inflation-adjusted 4% of retirement date assets over 30 years results in an estimated failure rate of about 6%. This modest projected failure rate rises sharply if real returns decline. As of January 2013, intermediate-term real interest rates are about 4% less than their historical average. Calibrating bond returns to the January 2013 real yields offered on 5-year TIPS, while maintaining the historical equity premium, causes the projected failure rate for retirement account withdrawals to jump to 57%. The 4% rule cannot be treated as a safe initial withdrawal rate in today’s low interest rate environment. Some planners may wish to assume that today’s low interest rates are an aberration and that higher real interest rates will return in the medium-term horizon. Although there is little evidence to support this assumption, we estimate how a reversion to historical real yields will impact failure rates. Because of sequence of returns risk, portfolio withdrawals can cause the events in early retirement to have a disproportionate effect on the sustainability of an income strategy. We simulate failure rates if today's bond rates return to their historical average after either 5 or 10 years and find that failure rates are much higher (18% and 32%, respectively for a 50% stock allocation) than many retirees may be willing to accept. The success of the 4% rule in the U.S. may be an historical anomaly, and clients may wish to consider their retirement income strategies more broadly than relying solely on systematic withdrawals from a volatile portfolio.

Deep reinforcement learning (DRL) has reached an unprecedent level on complex tasks like game solving (Go or StarCraft II), and autonomous driving. However, applications to real financial assets are still largely unexplored and it remains an open question whether DRL can reach super human level. In this ECML PKKDD demo, we showcase state-of-the-art DRL methods for selecting portfolios according to financial environment, with a final network concatenating three individual networks using layers of convolutions to reduce network's complexity. The multi entries of our network enables capturing dependencies from common financial indicators features like risk aversion, citigroup index surprise, portfolio specific features and previous portfolio allocations. Results on test set show this approach can overperform traditional portfolio optimization methods.

I define and analyze the agency costs of overvalued equity. They explain the dramatic increase in corporate scandals and value destruction in the last five years; costs that have totaled hundreds of billions of dollars. When a firm's equity becomes substantially overvalued it sets in motion a set of organizational forces that are extremely difficult to manage - forces that almost inevitably lead to destruction of part or all of the core value of the firm. WorldCom, Enron, Nortel, and eToys are only a few examples of what can happen when these forces go unmanaged. Because we currently have no simple solutions to the agency costs of overvalued equity this is a promising area for future research. The first step in managing these forces lies in understanding the incongruous proposition that managers should not let their stock price get too high. By too high I mean a level at which management will be unable to deliver the performance required to support the market's valuation. Once a firm's stock price becomes substantially overvalued managers who wish to eliminate it are faced with disappointing the capital markets. This value resetting (what I call the elimination of overvaluation) is not value destruction because the overvaluation would disappear anyway. The resulting stock price decline will generate substantial pain for shareholders, board members, managers and employees, and this makes it difficult for managers and boards to short circuit the forces leading to value destruction. And when boards and managers choose to defend the overvaluation they end up destroying part or all of the core value of the firm. WorldCom, Enron, Nortel, and eToys are only a few examples of what can happen if these forces go unmanaged. Control markets cannot solve the problem because you cannot buy up an overvalued firm, eliminate the overvaluation and make money. Equity-based compensation cannot solve the problem because it makes the problem worse, not better. While it is puzzling that short selling was unable to resolve the problem the evidence seems to be consistent with the Shleifer and Vishny (1997) arguments for the limits of arbitrage. It appears the solution to these problems lies in the board of directors and the governance system. But that is a problem because there is substantial evidence that weak governance systems have failed widely. It also appears that boards and audit committees would be well served by communicating with and carefully evaluating the information that could be provided by short sellers of the firm's securities.

Deep neural networks (DNNs) are powerful types of artificial neural networks (ANNs) that use several hidden layers. They have recently gained considerable attention in the speech transcription and image recognition community for their superior predictive properties including robustness to over fitting. However their application to algorithmic trading has not been previously researched, partly because of their computational complexity. This paper describes the application of DNNs to predicting financial market movement directions. In particular we describe the configuration and training approach and then demonstrate their application to back testing a simple trading strategy over 43 different Commodity and FX future mid-prices at 5-minute intervals. All results in this paper are generated using a C implementation on the Intel Xeon Phi co-processor which is 11.4x faster than the serial version and a Python strategy back testing environment both of which are available as open source code written by the authors.

Mutual funds represent one of the most popular investment instruments. Some institutions offer fund awards to recognize strong performing funds and fund groups that have shown excellent returns relative to their peers. Many fund companies also use awards won in their advertising and marketing material. This brings rise to the question: Do investors think award winning funds have a better brand image? Can awards increase investors’ purchase intention? The purpose of this study is to investigate the relationships and effects of brand image, perceived quality, perceived risk, perceived value, and purchase intention, as well as to examine the effects of demographic variables on these five dimensions. The research findings show significant relationships between brand image, perceived quality, perceived value, and purchase intention. In addition, some demographic variables may lead to significant differences in these five dimensions. Finally, the results from structural equation modeling show that there are positive and direct effects among brand image, perceived quality, perceived value, and purchase intention. Brand image indeed increases investors purchase intentions. The purchase intention is affected mainly by perceived quality, not by perceived risk.

Machine learning offers a set of powerful tools that holds considerable promise for investment management. As with most quantitative applications in finance, the danger of misapplying these techniques can lead to disappointment. One crucial limitation involves data availability. Many of machine learning’s early successes originated in the physical and biological sciences, in which truly vast amounts of data are available. Machine learning applications often require far more data than are available in finance, which is of particular concern in longer-horizon investing. Hence, choosing the right applications before applying the tools is important. In addition, capital markets reflect the actions of people, which may be influenced by others’ actions and by the findings of past research. In many ways, the challenges that affect machine learning are merely a continuation of the long-standing issues researchers have always faced in quantitative finance. While investors need to be cautious—indeed, more cautious than in past applications of quantitative methods—these new tools offer many potential applications in finance. In this article, the authors develop a research protocol that pertains both to the application of machine learning techniques and to quantitative finance in general.

Mean-Variance Optimization (MVO) as introduced by Markowitz (1952) is often presented as an elegant but impractical theory. MVO is "an unstable and error-maximizing" procedure (Michaud 1989), and "is nearly always beaten by simple 1/N portfolios" (DeMiguel, 2007). And to quote Ang (2014): "Mean-variance weights perform horribly… The optimal mean-variance portfolio is a complex function of estimated means, volatilities, and correlations of asset returns. There are many parameters to estimate. Optimized mean-variance portfolios can blow up when there are tiny errors in any of these inputs...". In our opinion, MVO is a great concept, but previous studies were doomed to fail because they allowed for short-sales, and applied poorly specified estimation horizons. For example, Ang used a 60 month formation period for estimation of means and variances, while Asness (2012) clearly demonstrated that prices mean-revert at this time scale, where the best assets in the past often become the worst assets in the future. In this paper we apply short lookback periods (maximum of 12 months) to estimate MVO parameters in order to best harvest the momentum factor. In addition, we will introduce common-sense constraints, such as long-only portfolio weights, to stabilize the optimization. We also introduce a public implementation of Markowitz's Critical Line Algorithm (CLA) programmed in R to handle the case when the number of assets is much larger than the number of lookback periods. We call our momentum-based, long-only MVO model Classical Asset Allocation (CAA) and compare its performance against the simple 1/N equal weighted portfolio using various global multi-asset universes over a century of data (Jan 1915-Dec 2014). At the risk of spoiling the ending, we demonstrate that CAA always beats the simple 1/N model by a wide margin.

We extend the analysis in Sloan (1996) to identify the source of information in accruals about earnings quality. Our results indicate that information in accruals about earnings quality is not limited to the current accruals analyzed by Sloan, but extends to non-current accruals. We also show that while information in accruals originates almost exclusively from asset accruals, liability accruals play a useful role in helping to isolate information in asset accruals about earnings quality. Finally, we show that information in accruals about earnings quality originates from both growth in the scale of operations and deterioration in the efficiency of asset usage. Overall, our results indicate that total accruals, defined as the difference between earnings and free cash flows, provide an intuitive, robust and parsimonious measure of earnings quality. Contrary to existing studies, our results also indicate that the information in accruals about earnings quality is not attributable to a single factor, such as 'discretionary' accruals or firm growth.

This is an advanced course in asset management taught at the University of Paris-Saclay. The 1400 slides cover the following topics: Lecture 1: Portfolio Optimization (232 pages), Risk Budgeting (269 pages), Lecture 3: Smart Beta, Factor Investing and Alternative Risk Premia (375 pages), Lecture 4: Equity Portfolio Optimization with ESG Scores (63 pages), Lecture 5: Climate Portfolio Construction: Portfolio Decarbonization & Net-Zero Investing (183 pages), Lecture 6: Equity and Bond Portfolio Optimization with Green Preferences (73 pages), and Lecture 7: Machine Learning in Asset Management (196 pages).

The essence of structured finance activities is the pooling of economic assets (e.g. loans, bonds, mortgages) and subsequent issuance of a prioritized capital structure of claims, known as tranches, against these collateral pools. As a result of the prioritization scheme used in structuring claims, many of the manufactured tranches are far safer than the average asset in the underlying pool. We examine how the process of securitization allowed trillions of dollars of risky assets to be transformed into securities that were widely considered to be safe, and argue that two key features of the structured finance machinery fueled its spectacular growth. At the core of the recent financial market crisis has been the discovery that these securities are actually far riskier than originally advertised.

This study evaluates the effectiveness of specialised large language models (LLMs) developed for accounting and finance. Empirical analysis demonstrates that these domain-specific models, despite being nearly 50 times smaller, consistently outperform state-of-the-art general-purpose LLMs in return prediction. By pre-training the models on year-specific financial datasets from 2007 to 2023, the study also mitigates look-ahead bias, a common limitation of general-purpose LLMs. The findings highlight the critical importance of addressing look-ahead bias to ensure reliable results. Extensive robustness checks further validate the superior performance of these models.

There is ongoing debate as to whether the practice known as “payment for order flow” (PFOF) leads to price improvement for retail investors. In this paper, I use a randomized controlled trial to attempt to answer this question and measure the extent of price improvement. The trial involves trading random stocks at random times across random brokers and comparing execution quality across direct market access and PFOF-based brokers. Consistent with the national best bid and offer (NBBO) not representing the prevailing market conditions, I find that orders executed via direct market access receive significant price improvement relative to the NBBO. Using direct orders as the counterfactual, I find considerable heterogeneity in the extent of price improvement provided by PFOF-based brokers. Brokers deriving comparatively more revenue from PFOF (e.g., Robinhood) provide negligible price improvement, whereas brokers deriving comparatively less revenue from PFOF (e.g., TDA) provide economically and statistically significant price improvement. Collectively, the evidence is consistent with the notion that PFOF benefits retail investors, and that agency problems can prevent brokers from passing these benefits along to their customers.

Our goal is to develop insights about the max squared Sharpe ratio for model factors as a metric for ranking asset-pricing models. We consider nested and non-nested models. The nested models are the CAPM, the three-factor model of Fama and French (1993), the five-factor extension in Fama and French (2015), and a six-factor model that adds a momentum factor. The non-nested models examine three issues about factor choice in the six-factor model: (i) cash profitability versus operating profitability as the variable used to construct profitability factors, (ii) long-short spread factors versus excess return factors, and (iii) factors that use small or big stocks versus factors that use both.

Spanish Abstract: El beneficio de una empresa es un dato arbitrario supuestas determinadas hipótesis de contabilización de gastos e ingresos. Por el contrario, el cash flow o flujo (dinero que va de la caja de la empresa al bolsillo de alguien: accionistas, propietarios de deuda,…) es una medida objetiva, una cifra única no sometida a un criterio particular. El Cash flow para las acciones (CFac) es el dinero que sale de la caja y llega al bolsillo de los accionistas. El FCF (free cash flow) es el CFac de la empresa en el caso de que ésta no tuviera deuda. English Abstract: We use three different definitions of cash flow: equity cash flow (ECF), free cash flow (FCF) and capital cash flow (CCF). A company's net income is a quite arbitrary figure obtained after assuming certain accounting hypotheses regarding expenses and revenues (one of several that can be obtained, depending on the criteria applied). However, the ex-post cash flow is an objective measure, a single figure that is not subject to any personal criterion.

We develop a new measure of impact elasticity, defined as a firm's change in environmental impact due to a change in its cost of capital. We show empirically that a reduction in financing costs for firms that are already green leads to small improvements in impact at best. In contrast, increasing financing costs for brown firms leads to large negative changes in firm impact. Thus, sustainable investing that directs capital away from brown firms and toward green firms may be counterproductive, in that it makes brown firms more brown without making green firms more green. We further show that brown firms face very weak financial incentives to become more green. Due to a mistaken focus on percentage reductions in emissions, the sustainable investing movement primarily rewards green firms for economically trivial reductions in their already low levels of emissions.

In this paper, we review twelve valuations of Terra performed by Spanish and non-Spanish bank analysts and brokers.Of the twelve valuations, only one used cash flow discounting. Another valuation was based on multiples, but also used cash flow discounting to perform a reverse valuation. All others used several multiples. Only one valuation report recommended to sell.Terra started trading on the stock market in November 1999. The placement price was 13 euros per share (11.81 for retailers). In February 2000, its price stood at 139.75 euros. Between November 1999 and February 2000, Terra provided a return of 975% for its shareholders. However, by December 2000, the share price had plummeted to 11.6 euros, 8.3% of its February high. The average annual volatility of the Terra share was almost 100%.If you can't find a rational explanation for a share to continue rising, you can be sure that it will fall. To become a millionaire, you must sell your shares at the right time. A website is not necessarily a business. Selling below cost gets you lots of customers, but not much money.

This paper contains the statistics of the Equity Premium or Market Risk Premium (MRP) used in 2011 for 56 countries. We got answers for 85 countries, but we only report the results for 56 countries with more than 6 answers. Most previous surveys have been interested in the Expected MRP, but this survey asks about the Required MRP. The paper also contains the references used to justify the MRP, comments from persons that do not use MRP, and comments from persons that do use MRP.

Traditional approaches to valuing equities have largely focused on the Dividend Discount Model. It may be hard to reliably estimate dividend processes in small samples and market participants focus primarily on earnings and other accounting information in analyzing stocks. For these reasons we try to value stocks using earnings and book value. Building on the seminal work of Miller and Modigliani (1961) and Ohlson (1990, 1995) we develop a Generalized Earnings Model of stock valuation which uses earnings and book values. This is a general no-arbitrage model which uses stochastic pricing kernels. The model can be implemented by assuming the driving variables follow affine processes which allows tractable calculations. We apply the model to several individual stocks.

The paper compares the forecasting ability of the most popular volatility forecasting models and develops an alternative. The comparison of existing models focuses on four issues: 1) the relative weighting of recent versus older observations, 2) the estimation criteria, 3) the trade-off in terms of out-of-sample forecasting error between simple and complex models, and 4) the emphasis placed on large shocks. Like previous studies, we find that financial markets have longer memories than reflected in GARCH(1,1) model estimates but find this has little impact on out-of-sample forecasting ability. While more complex models which allow a more flexible weighting pattern than the exponential model forecast better on an in-sample basis, due to the additional estimation error introduced by additional parameters, they forecast poorly out-of-sample. With the exception of GARCH models, we find that models based on absolute return deviations generally forecast volatility better than otherwise equivalent models based on squared return deviations. Among the most popular time series models, we find that GARCH(1,1) generally yields better forecasts than the historical standard deviation and exponentially weighted moving average models though between GARCH and EGARCH there is no clear favorite. However, in terms of forecast accuracy, all are dominated by a new, simple, non-linear least squares model, based on historical absolute return deviations, that we develop and test here.

For much of the last century, value investors considered themselves to be the winners in the investment world, a result they attributed to their patience, maturity and good sense. That view, at least on the surface, was backed up by evidence that “value” stocks, defined as stocks that trade at low multiples of earnings and book value, earned higher returns than “growth” stocks, defined loosely as companies that trade at high multiples of earnings or book value. It was reinforced by the mythology of great value investors, with Warren Buffett and Charlie Munger taking center stage, as deep thinkers, with profound insights on how markets work. In the last two decades, value investing lost its edge, and a debate has revolved around whether this is a temporary phase, and the result of an unusual macro environment, or a reflection of a permanent change in economies and markets. In this paper, we argue that value investing, at least as practiced today, has become rigid and ritualistic, and that while some of its failures can be attributed to external factors, many can be traced back to practices and rules of thumb that have outlived their usefulness. We argue that if value investing is to be successful in the future, it needs to develop a more dynamic view of value and a greater willingness to live with and invest in the face of uncertainty.

In this paper, we propose a stop-loss strategy to limit the downside risk of the well-known momentum strategy. At a stop-level of 10%, we find, with data from January 1926 to December 2013, that the maximum monthly losses of the equal- and value-weighted momentum strategies go down from -49.79% to -11.36% and from -64.97% to -23.28%, while the Sharpe ratios are more than doubled at the same time. We also provide a general equilibrium model of stop-loss traders and non-stop traders and show that the market price differs from the price in the case of no stop-loss traders by a barrier option.