Past Returns Research Articles

On an adaptive Black–Litterman investment strategy using conditional fundamentalist information: A Brazilian case study

We propose an investment strategy based on the Black–Litterman model with conditional information. We present how observed price-earnings ratio and past returns can be used to determine 1-step ahead returns, considering investors with different risk profiles. The provided approach updates the conditional probability distribution of asset returns and mitigates asset allocation instability due to estimation errors. Our case study using Brazilian data shows the resulting optimal portfolios outperform traditional mean-variance portfolios even in an emerging market with one of the highest nominal interest rates.

Extrapolative Beliefs in the Cross-Section: What Can We Learn from the Crowds?

Using novel data from a crowdsourcing platform for ranking stocks, we investigate how investors form expectations about stock returns over the next week. We find that investors extrapolate from stocks' recent past returns, with more weight on more recent returns, especially when recent returns are negative, salient, or from a dispersed cross-section. Such extrapolative beliefs are stronger among nonprofessionals and large stocks. Moreover, consensus rankings negatively predict returns over the next week, more so among stocks with low institutional ownership and a high degree of extrapolation. A trading strategy that sorts stocks on investor beliefs generates an economically significant profit.

Linear models for the impact of order flow on prices. I. History dependent impact models

Market impact is a key concept in the study of financial markets and several models have been proposed in the literature so far. The propagator model posits that the price at high frequency time scales is a linear combination of the signs of the past executed market orders, weighted by a so-called propagator function. This model needs to be extended since prices are a priori influenced not only by the past order flow, but also by the past realization of returns themselves. In this paper, we propose a two-event framework, where price-changing and non price-changing events are considered separately. We show that two-event propagator models provide a remarkable improvement of the description of the market impact, especially for large tick stocks, where the events of price changes are very rare and very informative. Specifically the extended approach captures the excess anti-correlation between past returns and subsequent order flow which is missing in one-event models.

Linear models for the impact of order flow on prices. II. The Mixture Transition Distribution model

Modelling the impact of the order flow on asset prices is of primary importance to understand the behaviour of financial markets. Part I of this paper reported the remarkable improvements in the description of the price dynamics which can be obtained when one incorporates the impact of past returns on the future order flow. However, impact models presented in Part I consider the order flow as an exogenous process, only characterised by its two-point correlations. This assumption seriously limits the forecasting ability of the model. Here we attempt to model directly the stream of discrete events with a so-called Mixture Transition Distribution (MTD) framework, introduced originally by Raftery [J. R. Stat. Soc. Ser. B, 1985, 528–539]. We distinguish between price-changing and non price-changing events and combine them with the order sign in order to reduce the order flow dynamics to the dynamics of a four-state discrete random variable. The MTD represents a parsimonious approximation of a full high-order Markov chain. The new approach captures with adequate realism the conditional correlation functions between signed events for both small and large tick stocks and signature plots. From a methodological point of view, constraining the MTD within the class of ergodic Markov models, and exploiting the buy–sell symmetry of the data, we propose a weak restriction on the transition matrices which solves the problem of identifiability of mixture models. In spite of the large number of parameters, this translates into a feasible and robust estimation procedure. Out-of-sample analyses demonstrate that the model does not overfit the data.

TRADING RULES ON A SMALL STOCK MARKET

In this article, the results of an extensive study of the weak form efficiency of the Iceland stock market are presented. This study almost covers the market’s entire history, with the research starting at the beginning of 1993 and ending in July 2017. Four trading rules based on 70-day moving averages were constructed and compared with the passive investment strategy of buying the market index. All of these trading rules provided significantly better returns than the passive strategy, even when considering trading costs. This result indicates that the Icelandic stock market did not show weak form efficiency, and past returns predicted future returns during the period examined.

Detailed study of a moving average trading rule

We present a detailed study of the performance of a trading rule that uses moving averages of past returns to predict future returns on stock indexes. Our main goal is to link performance and the stochastic process of the traded asset. Our study reports short-, medium- and long-term effects by looking at the Sharpe ratio (SR). We calculate the Sharpe ratio of our trading rule as a function of the probability distribution function of the underlying traded asset and compare it with data. We show that if the performance is mainly due to presence of autocorrelation in the returns of the traded assets, the SR as a function of the portfolio formation period (look-back) is very different from performance due to the drift (average return). The SR shows that for look-back periods of a few months the investor is more likely to tap into autocorrelation. However, for look-back larger than few months, the drift of the asset becomes progressively more important. Finally, our empirical work reports a new long-term effect, namely oscillation of the SR and proposes a non-stationary model to account for such oscillations.

Mutual fund performance attribution and market timing using portfolio holdings

We propose a novel performance attribution model for equity fund portfolios. The model analyses investment decisions based on portfolio holdings and measures the value added from different sources of performance such as past return strategies, security selection, market timing and passive timing. The model was tested for a sample of mutual funds. Empirical results show that security selection is the main contributor to fund performance regardless of the sample period considered or the asset pricing model used. The evidence of timing ability is mixed with low significance. Nevertheless there are noticeable differences between the timing ability of the best and worst performing funds, especially in crisis periods. Analysing the relationship between mutual fund performance (and its different components) and fund characteristics, we find that top funds are significantly smaller and more concentrated than other funds. Finally, we also examine the persistence in the performance and in its components finding evidence of positive persistence in past return strategies and picking skills although this persistence is not shown in the overall performance.

Does overreaction still exist in Thailand?

Overreaction of stock prices is identified when investors give more credence to new information than its intrinsic value, thus driving stock prices away from the level at which they should be. This study aims to examine the overreaction hypothesis in Thailand in recent years, with the intention of suggesting investment strategies based on the results obtained. The final sample of 438 companies was collected from the period 1990 to 2016 and is categorised into two portfolios: loser portfolios, with the lowest past returns; and winner portfolios, with the highest past returns. Both equally-weighted and value-weighted methods are used to examine these two portfolios. The results show evidence of stock price overreaction on the Thai stock market, particularly during periods involving interesting situations, such as the Asian financial crisis of 1997, political chaos in 2005, and the global financial crisis in 2008–2009. Moreover, the contrarian strategy is preferred when investing in Thailand, as the loser portfolios reveal a reversed performance in the following period. However, when the value-weighted method is applied, evidence of overreaction is stronger. This indicates that larger stocks appear to overreact more in comparison to smaller ones. Thus, the size effect should be an interesting point to consider prior to making investment decisions in Thailand.

Beating Momentum: Using Neural Networks to Uncover Additional Information in Past Returns

One way of thinking about the popular momentum return factor is as a simple algorithm that predicts what percentile a stock’s returns will fall into next month using the previous eleven months of return data. We want to determine if there is more information in those past returns that could be useful in predicting next month’s relative return. To explore this question we are constructing a neural network and training it to predict what percentile a stock’s returns will fall into next month using the past eleven months worth of returns. We feel that a neural network is a good choice here because they are capable of learning non-linear patterns. After training the neural network using CRSP returns data from 1926 to 2006 our network is able to generate a portfolio that beats the momentum strategy from 2007 to 2016.

Initial Fund Size Anomaly and Its Implication for Returns-to-Scale: Evidence From Japanese Mutual Funds

The relationship between the size and return of mutual funds has been a central interest from practical viewpoints. Using the substantial heterogeneity associated with the size of Japanese mutual funds over the early stage of their life cycle, which we call as fund size anomaly, we examine how the returns-to-scale property of mutual funds depend on the initial size of the funds and its size dynamics. Toward this end, first, we document the systematic pattern of the timing for launching fund and find that the fund flows are related to the introduction of new funds. An asset management firm is more likely to launch a fund both when the recently introduced funds attract large inflow and the incumbent funds suffer from large outflow. Second, regarding the reaction of fund flow to the past return, we confirm that fund inflow follows higher past returns, which is consistent with the findings in the extant studies. Third, as the most important results, the returns-to-scale property crucially depends on the initial size and the dynamics of mutual funds. While the funds starting with small size show the widely observed decreasing returns-to-scale, it is not the case for the funds with large initial size and experiencing the reduction in its size. These results jointly suggest that large fund size is driven by the motivations of asset management firms, which results in the biased behavior such as the employment of closet-index strategy.

Aggregate Investor Confidence in the Stock Market

ABSTRACTOverconfidence is one of the most robust findings in the field of behavioral finance, and is associated with excessive trading and risk taking among market participants. Assessment of the level of confidence of individuals in their abilities and skills is well documented. However, the literature lacks an aggregate measure of investor confidence, with this required to test its implications on a macro level. The author introduces a simple measure of aggregate investor confidence by adopting a formal model of overconfidence. The applications of the measure suggest that, in aggregate, higher trading activity occurs when investor confidence soars, particularly for smaller stocks. Subsequently, the effect partially reverses, implying a correction to an initial overreaction. The newly introduced investor confidence index possesses better ability to predict trading activity than past returns, as used in prior studies. Additionally, investors tend to have a higher risk appetite when confident, as shown by increased investment in small stocks with higher risk.

Market Efficiency in Malaysia: An Empirical Study of Random Walk Hypothesis of Kuala Lumpur Stock Market (Composite Index) Bursa Malaysia

As long as financial markets are concerned, for many years’ economists, statisticians and financial analyst have been interested in developing and testing models of stock price behaviour and their forecast. This study examines whether the Kuala Lumpur Stock Market, Malaysia is efficient if the Stock Returns follow a random walk. The study employs daily closing prices of Kuala Lumpur Stock Exchange – Bursa Malaysia Composite Index for a time period of 28 Apr 1998 to 30 Dec 2014. The existence of random walk for Bursa Malaysia Index has been examined through autocorrelation, Q-statistics and the run test and finds that the Kuala Lumpur Stock Market was not efficient in the weak form during the testing period. The results suggest that the stock prices in Malaysia do not reflect all the information in the past stock prices and abnormal returns can be achieved by investors through exploiting the market inefficiency.

Market Efficiency in India: An Empirical Study of Random Walk Hypothesis of Indian Stock Market NSE Midcap

As long as financial markets are concerned, for many years’ economists, statisticians and financial analyst have been interested in developing and testing models of stock price behaviour and their forecast. This study examines whether the Indian stock market is efficient if the stock returns follow a random walk. The study employs daily closing prices of NSE Midcap 50 Index for a time period of 15 Sept 2010-28 Nov 2014. The existence of random walk for NSE Midcap Index has been examined through autocorrelation, Q-statistics and the run test and finds that the Indian stock market was not efficient in the weak form during the testing period. The results suggest that the stock prices in India do not reflect all the information in the past stock prices and abnormal returns can be achieved by investors through exploiting the market inefficiency.

Investor Base and Stock Return Anomalies

After controlling for market capitalization, the predictability of future stock returns associated with each of the earnings-to-price ratio, the book-to-market ratio, the past return, the total vol ...

Shrunk volatility value-at-risk: an application on US balanced portfolios

We test the naive model to forecast ex ante value-at-risk (VaR) using a shrinkage estimator between realized volatility estimated on past return time series and implied volatility quoted on the market. Implied volatility is often indicated as the operator’s expectation about future risk, while the historical volatility straightforwardly represents the realized risk prior to the estimation point, which by definition is backward looking. Therefore, the VaR prediction strategy uses information both on the expected future risk and on the past estimated risk. We examine Cesarone and Colucci’s 2016 model – Shrunk Volatility VaR (ShVolVaR) – and generalize it, assuming that returns are conditionally Student t distributed (the authors assume that returns are normally distributed, but normality is a particular case of Student t distribution with degrees of freedom that tend to infinity). The ShVolVaR results are compared with those of six benchmark industry VaR models. The performance of all VaR models is validated using both statistical accuracy and efficiency evaluation tests on thirty-nine equally spaced, balanced portfolios composed by US equity and bonds over an out-of-sample period that covers different crises. We evaluate model performances on four VaR confidence levels (95%, 99%, 99.5% and 99.9%). We also validate the models under loss function backtests; our results confirm the efficacy of implied volatility indexes as inputs for a VaR model, combined with realized volatilities. Further, we confirm Cesarone and Colucci’s 2016 conclusion in almost all balanced portfolios. In our empirical analysis, we find that a VaR model that correctly estimates VaR for all ε and all the balanced portfolios considered does not exist. However, for each confidence level, the ShVolVaR model, with appropriate values for the parameters α and v, can achieve the same results as the common VaR models that are widely used in the finance industry.

Toward a Factor Structure in Crypto Asset Returns

A new market for crypto assets has emerged from active trading of major cryptocurrencies and the listing of many ICO tokens. We conduct an early investigation into potential factor structures in the expected returns of crypto assets. We find that crypto assets with large market capitalization, low volatility, and high past returns tend to outperform in the following month. These are suggestive evidences for an emerging factor structure, even though crypto asset returns are still largely dominated by idiosyncratic noises. Our findings could help investors make better decisions in the nascent crypto asset market.

What Accounts for the Differences in Rent-Price Ratio and Turnover Rate? A Search-and-Matching Approach

We build an on-the-house search model and show that two frequently used metrics for the housing market, the rent-to-price ratio (or rental yield) and the turnover rate, are jointly determined in equilibrium. We, therefore, estimate a simultaneous equations system on matched sale-rental pairs, as a housing unit cannot be owner-occupied and renter-occupied at the same time. We confirm that a higher turnover rate is associated with a lower rental yield. We also identify a form of “dichotomy” in empirical determinants of rental yields and turnover rates at the estate level: the demographic structure and past returns affect an estate’s turnover rate, while popularity, human capital, mortgage burdens, and long-run rent growth determine its rental yield. The robustness of our results is established through a series of tests. Our popularity index for 130 estates in Hong Kong, estate rankings, and the brand premia of major real estate developers may carry independent interests.

Do Individual Investors Consciously Speculate on Reversals? Evidence from Leveraged Warrant Trading

Using proprietary data on bank-issued knock-out warrants, we find that individual investors, on aggregate, bet on price reversals. In a simple model we demonstrate that a mechanical channel due to market and product characteristics may account for investors' betting on reversals, even if investors' purchasing and selling decisions are independent of past returns. Our empirical results suggest that the mechanical channel can explain almost one half of the associationbetween past returns and individual investors' order flow, while the rest can be attributed to the disposition effect, that is, investors' higher propensity to sell assets from their portfolios that have appreciated.

Forecasting ETFs with Machine Learning Algorithms

In this article, the authors apply cutting-edge machine learning algorithms to one of the oldest challenges in finance: predicting returns. For the sake of simplicity, they focus on predicting the direction (either up or down) of several liquid exchange-traded funds (ETFs) and do not attempt to predict the magnitude of price changes. The ETFs serve as asset class proxies. The authors employ approximately five years (from January 2011 to January 2016) of historical, daily data obtained through Yahoo Finance. Using supervised learning classification algorithms, readily available from Python’s Scikit-Learn, they employ three powerful techniques: (1) deep neural networks, (2) random forests, and (3) support vector machines (linear and radial basis function). They document the performance of the three algorithms across four information sets: past returns, past volume, dummies for days/months, and a combination of all three. They use a gain criterion to compare classifiers’ performance. First, they find that these algorithms work well over one- to three-month horizons. Short-horizon predictability, over days, is extremely difficult, and thus the results support the short-term random walk hypothesis. Second, they document the importance of cross-sectional and intertemporal volume as a powerful information set. Third, they show that many features are needed for predictability because each feature makes very small contributions toward predictability. The authors conclude that ETF returns can be predicted with machine learning algorithms, but practitioners should incorporate prior knowledge of markets and intuition on asset class behavior. <b>TOPICS:</b>Exchange-traded funds and applications, big data/machine learning, performance measurement

The Cross-Section of Risk and Return

In the finance literature, a common practice is to create characteristic portfolios by sorting on characteristics associated with average returns. We show that the resulting portfolios are likely to capture not only the priced risk associated with the characteristic, but also unpriced risk. We develop a procedure to remove this unpriced risk using covariance information estimated from past returns. We apply our methodology to the five Fama and French (2015) characteristic portfolios. The squared Sharpe ratio of the optimal combination of the resulting characteristic efficient portfolios is 2.16, compared with 1.16 for the original characteristic portfolios.