Options as a predictor of common stock price changes

Since rather novel techniques such as neural nets allow investigation of nonlinear model specification previously untested, it may be that traditional models of price formation underperform through misspecification rather than market efficiency. This paper explores whether a multilayer backpropagation model offers exploitable profit opportunities for some limited period. Using an intradaytransaction dataset obtained from the European Options Exchange (Amsterdam), We attempted to predict the return on Philips. Two neural nets are contrasted to ordinary linear regression analysis on the basis of three benchmarks (MSE, and net realized returns). An adaptively trained 33-14-1 architecture scored best on all criteria and yielded an annualized 11% return following a simple one-period trading strategy.     

Ordinal Pairwise Partitioning (OPP) Approach to Neural Networks Training in Bond rating

Statistical classification methods such as multivariate discriminant analysis have been widely used in bond rating classification in spite of the limitations of the methodology. Recently, neural networks have emerged as new methods for business classification. This approach to neural networks training is to categorize a new instance as one of the predefined bond classes. Such a conventional approach has limitations in dealing with the ordinal nature of bond rating. In addition, most of the prior studies have used sample data which are evenly divided among the classes. However, the natural population in real application is usually unevenly divided among the classes. Under such circumstances, it is hard to achieve good predictive performance. As the number of classes to be recognized increases, the predictive performance decreases. In this article, to increase the predictive performance in real-world bond rating, we propose the ordinal pairwise partitioning (OPP) approach to backpropagation neural networks training. The main idea of the OPP approach is to partition the data set in the ordinal and pairwise manner according to the output classes. Then, each backpropagation neural networks model is trained by using each partitioned data set and is separately used for classification. Experimental results show that the predictive performance of the proposed OPP approach can be significantly enhanced, when compared to the conventional neural networks modeling approach as well as multivariate discriminant analysis. The OPP approach has two computation methods, and we discuss under which circumstances one method performs better than the other. We also show the generalizability of the OPP approach. © 1997 by John Wiley & Sons, Ltd.     

An Empirical Examination of U.K. International Unit Trust Performance

We examine the performance of U.K. unit trusts with international equity objectives between January 1985 and December 2000 using four international factor models. The international version of the Carhart (1997) model performs the best in explaining the cross-section of international stock returns. There is little evidence of superior performance by international trusts relative to the global models. We also find that the choice between a local and global version of the Carhart model has a significant impact on the relation between the investment sector of the trust and performance.JEL classification: G10, G12     

Simulating and Forecasting Utility Stock Returns: Arbitrage Pricing Theory vs. Capital Asset Pricing Model

Just when the capital asset pricing model (CAPM) has become accepted by public utility regulators as a method for estimating a utility's screening rate, academic criticism of the model's theoretical and empirical shortcomings has led to empirical testing of the alternative arbitrage pricing theory (APT). This paper expands on recent APT-CAPM performance comparisons by simulating returns of public utility stocks using versions of both models, as was done by Bower, Bower, and Logue in a 1984 paper. In addition, the models are used for ex-post forecasting of returns in a subsequent time period. The Litzenberger-Ramaswamy method is used to correct for errors-in-variables in the CAPM cross-sectional equation. This allows for estimating the security market line using firm betas. The same methodology is used in the APT stages. Three different criteria—the Theil inequality, the sources of mean square error, and Chen's estimated weights of expected return-are used to compare CAPM and APT simulation and forecasting of the equity screening rates. Tested on a sample of 128 public utility companies, results show that neither model is clearly dominant. There is a tendency for reversal of performance. The model that is superior for simulating returns tends to be inferior for forecasting them, and vice-versa.     

A horse race of models and estimation methods for predicting bankruptcy

We use a comprehensive set of performance metrics to analyze the improvement in the classification power and prediction accuracy of various bankruptcy prediction models after adding governance variables and/or varying the estimation method used. In a sample covering bankruptcies of U.S. public firms in the period 2000 to 2015, we find that the addition of governance variables significantly improves the performance of all bankruptcy prediction models. We also find that the additional explanatory power provided by governance measures improves the further the firm is from bankruptcy, which suggests that governance variables may provide earlier and more accurate warning of the firm's bankruptcy potential. Our findings show that the performance of any bankruptcy prediction model is significantly affected by the estimation method used. We find that regardless of the bankruptcy model, hazard analysis provides the best classification and out-of-sample forecast accuracy among the parametric methods. Furthermore, non-parametric methods such as neural networks, data envelopment analysis or classification and regression trees appear to provide comparable and sometimes superior classification accuracy to hazard analysis. Lastly, we use the dynamic panel generalized methods of moments model to address concerns raised in prior studies about the susceptibility of similar studies to endogeneity issues and find that our findings continue to hold.     

Predicting a corporate credit analyst's risk estimate by logistic and linear models

The purpose is to predict corporate credit analyst's risk estimate by the weighted logistic (binary response) and linear regression (20-class risk estimate) analyses. The data comprise filed register information from Finska (Suomen Asiakastieto Oy) including 35 variables from 3200 companies. The coefficient of concordance was 95% and the rate of multiple determination 75% for the logistic and linear models, respectively. In a binary classification the differences in performance between the models were insignificant provided that the linear model is rotated. Both of the models give a classification accuracy of 90% in the estimation sample and 96% in the test sample.     

A neural network approach to the prediction of going concern status

The assessment of a firm's going concern status is not an easy task. To assist auditors, going concern prediction models based on statistical methods such as multiple discriminant analysis and logit/probit analysis have been explored with some success. This study attempts to look at a different and more recent approach—neural networks. In particular, a neural network model of the feedforward, backpropagation type was constructed to predict a firm's going concern status from six financial ratios, using a data set containing 165 non-going concerns and 165 matched going concerns. On an evenly distributed hold-out sample, the trained network model correctly predicted all 30 test cases. The results suggest that neural networks can be a promising avenue of research and application in the going concern area.     

Predicting Corporate Failure Using a Neural Network Approach

This paper investigates the performance of Artificial Neural Networks for the classification and subsequent prediction of business entities into failed and non-failed classes. Two techniques, back-propagation and Optimal Estimation Theory (OET), are used to train the neural networks to predict bankruptcy filings. The data are drawn from Compustat data tapes representing a cross-section of industries. The results obtained with the neural networks are compared with other well-known bankruptcy prediction techniques such as discriminant analysis, probit and logit, as well as against benchmarks provided by directly applying the bankruptcy prediction models developed by Altman (1968) and Ohlson (1980) to our data set. We control the degree of ‘disproportionate sampling’ by creating ‘training’ and ‘testing’ populations with proportions of bankrupt firms ranging from 1% to 50%. For each population, we apply each technique 50 times to determine stable accuracy rates in terms of Type I, Type II and Total Error. We show that the performance of various classification techniques, in terms of their classification errors, depends on the proportions of bankrupt firms in the training and testing data sets, the variables used in the models, and assumptions about the relative costs of Type I and Type II errors. The neural network solutions do not achieve the ‘magical’ results that literature in this field often promises, although there are notable 'pockets' of superior performance by the neural networks, depending on particular combinations of proportions of bankrupt firms in training and testing data sets and assumptions about the relative costs of Type I and Type II errors. However, since we tested only one architecture for the neural network, it will be necessary to investigate potential improvements in neural network performance through systematic changes in neural network architecture.     

Stock market return predictability: A combination forecast perspective

Based on traditional macroeconomic variables, this paper mainly investigates the predictability of these variables for stock market return. The empirical results show the mean combination forecast model can achieve superior out-of-sample performance than the other forecasting models for forecasting the stock market returns. In addition, the performances of the mean combination forecast model are also robust during different forecasting windows, different market conditions, and multi-step-ahead forecasts. Importantly, the mean combination forecast consistently generates higher CER gains than other models considering different investors' risk aversion coefficients and trading costs. This paper tries to provide more evidence of combination forecast model to predict stock market returns.     

Modelling small‐business credit scoring by using logistic regression,neural networks and decision trees

Previous research on credit scoring that used statistical and intelligent methods was mostly focused on commercial and consumer lending. The main purpose of this paper is to extract important features for credit scoring in small‐business lending on a dataset with specific transitional economic conditions using a relatively small dataset. To do this, we compare the accuracy of the best models extracted by different methodologies, such as logistic regression, neural networks (NNs), and CART decision trees. Four different NN algorithms are tested, including backpropagation, radial basis function network, probabilistic and learning vector quantization, by using the forward nonlinear variable selection strategy. Although the test of differences in proportion and McNemar's test do not show a statistically significant difference in the models tested, the probabilistic NN model produces the highest hit rate and the lowest type I error. According to the measures of association, the best NN model also shows the highest degree of association with the data, and it yields the lowest total relative cost of misclassification for all scenarios examined. The best model extracts a set of important features for small‐business credit scoring for the observed sample, emphasizing credit programme characteristics, as well as entrepreneur's personal and business characteristics as the most important ones. Copyright © 2005 John Wiley & Sons, Ltd.     

Risk,Mispricing, and Value Investing

We evaluate the stock return performance of a modified version of the book-to-market strategy and its implications for market efficiency. If the previously documented superior stock return of the book-to-market strategy represents mispricing, its performance should be improved by excluding fairly valued firms with extreme book-to-market ratios. To attain this, we classify stocks as value or glamour on book-to-market ratios and accounting accruals jointly. This joint classification is likely to exclude stocks with extreme book-to-market ratios due to mismeasured accounting book values reflecting limitations underlying the accounting system. Using both 12-month buy-and-hold returns and earnings announcement returns, our results show that this joint classification generates substantially higher portfolio returns in the post-portfolio-formation year than the book-to-market classification alone with no evidence of increased risk. In addition, this superior stock return performance is more pronounced among firms held primarily by small (unsophisticated) investors and followed less closely by market participants (stock price <$10). Finally, and most importantly, financial analysts are overly optimistic (pessimistic) about earnings of glamour (value) stock, and for a subset of firms identified as overvalued by our strategy, the earnings announcement raw return, as well as abnormal return, is negative. These last results are particularly important because it is hard to envision a model consistent with rational investors holding risky stocks with predictable negative raw returns for a long period of time rather than holding fT-bills and with financial analysts systematically overestimating the earnings of these stocks while underestimating earnings of stocks that outperform the stock market.     

Bankruptcy prediction of financially stressed firms: an examination of the predictive accuracy of artificial neural networks

This is an extension of prior studies that have used artificial neural networks to predict bankruptcy. The incremental contribution of this study is threefold. First, we use only financially stressed firms in our control sample. This enables the models to more closely approximate the actual decision processes of auditors and other interested parties. Second, we develop a more parsimonious model using qualitative ‘bad news’ variables that prior research indicates measure financial distress. Past research has focused on the ‘usefulness’ of accounting numbers and therefore often ignored non‐accounting variables that may contribute to the classification accuracy of the distress prediction models. In addition, rather than use multiple financial ratios, we include a single variable of financial distress using the Zmijewski distress score that incorporates ratios measuring profitability, liquidity, and solvency. Finally, we develop and test a genetic algorithm neural network model. We examine its predictive ability to that of a backpropagation neural network and a model using multiple discriminant analysis. The results indicate that the misclassification cost of the genetic algorithm‐based neural network was the lowest among the models. Copyright © 2001 John Wiley & Sons, Ltd.     

Cross‐sectional Tests of Conditional Asset Pricing Models: Evidence from the German Stock Market

We study the performance of conditional asset pricing models and multifactor models in explaining the German cross‐section of stock returns. We focus on several variables, which (according to previous research) are associated with market expectations on future market excess returns or business cycle conditions. Our results suggest that the empirical performance of the Capital Asset Pricing Model (CAPM) can be improved when allowing for time‐varying parameters of the stochastic discount factor. A conditional CAPM using the term spread explains the returns on our size and book‐to‐market sorted portfolios about as well as the Fama‐French three‐factor model and performs best in terms of the Hansen‐Jagannathan distance. Structural break tests do not necessarily indicate parameter instability of conditional model specifications. Another major finding of the paper is that the Fama‐French model – despite its generally good cross‐sectional performance – is subject to model instability. Unconditional models, however, do a better job than conditional ones at capturing time‐series predictability of the test portfolio returns.     

Predicting Bond Ratings Using Neural Networks: A Comparison with Logistic Regression

Bond rating agencies examine the financial outlook of a company and the characteristics of a bond issue and assign a rating that indicates an independent assessment of the degree of default risk associated with the firm’s bonds. Predicting this bond rating has been of interest to potential investors as well as to the firm. Prior research in this area has primarily relied upon traditional statistical methods to develop models with reasonably good prediction accuracy. This article utilizes a neural network approach to modeling the bond rating process in an attempt to increase the overall prediction accuracy of the models. A comparison is made to a more traditional logistic regression approach to classification prediction. The results indicate that the neural networks-based model performs significantly better than the logistic regression model for classifying a holdout sample of newly issued bonds in the 1990–92 period. A potential drawback to a neural network approach is a tendency to overfit the data which could negatively affect the model’s generalizability. This study carefully controls for overfitting and obtains significant improvement in bond rating prediction compared to the logistic regression approach. © 1997 by John Wiley & Sons, Ltd.     

The hedging effect of green bonds on carbon market risk

This paper explores effective hedging instruments for carbon market risk. Examining the relationship between the carbon futures returns and the returns of four major market indices, i.e., the VIX index, the commodity index, the energy index and the green bond index, we find that the connectedness between the carbon futures returns and the green bond index returns is the highest and this connectedness is extremely pronounced during the market's volatile period. Further, we develop and evaluate hedging strategies based on three dynamic hedge ratio models (DCC-APGARCH, DCC-T-GARCH, and DCC-GJR-GARCH models) and the constant hedge ratio model (OLS model). Empirical results show that among the four market indices the green bond index is the best hedge for carbon futures and performs well even in the crisis period. The paper also provides evidence that the dynamic hedge ratio models are superior to the OLS model in the volatile period as more sophisticated models can capture the dynamic correlation and volatility spillover between the carbon futures and market index returns.     

NONLINEAR FORECASTING OF THE GOLD MINER SPREAD: AN APPLICATION OF CORRELATION FILTERS

This paper models and forecasts the Gold Miner Spread from 23 May 2006 to 30 June 2011. The Gold Miner Spread acts as a suitable performance indicator for the relationship between physical gold and US gold equity. The contribution of this investigation is twofold. First, the accuracy of each model is evaluated from a statistical perspective. Second, various forecasting methodologies are then applied to trade the spread. Trading models include an ARMA (12,12) model, a cointegration model, a multilayer perceptron neural network (NN), a particle swarm optimization radial basis function NN and a genetic programming algorithm (GPA). Results obtained from an out‐of‐sample trading simulation validate the in‐sample back test as the GPA model produced the highest risk‐adjusted returns. Correlation filters are also applied to enhance performance and, as a consequence, volatility is reduced by 5%, on average, while returns are improved between 2.54% and 8.11% across five of the six models. Copyright © 2013 John Wiley & Sons, Ltd.     

Bankruptcy prediction: Application of the Taylor's expansion in logistic regression

The purpose of the present study is to test whether Taylor's series expansion can be used to solve the problem associated with the functional form of bankruptcy prediction models. To avoid the problems associated with the normality of variables, the logistic model to describe the insolvency risk is applied. Taylor's expansion is then used to approximate the exponent of the logistic function, or the logit. The cash to total assets, cash flow to total assets, and shareholder's equity to total assets ratios operationalize the factors affecting the insolvency risk. The usefulness of Taylor's model in bankruptcy prediction is evaluated applying the logistic regression model to the data from the Compustat database. The classification accuracy in the test data for the first and second years before bankruptcy show that the classification accuracy of a simple financial ratio model can be increased using the second-order and interaction terms of these ratios. However, in the third year, for the test data, Taylor's expansion is not able to increase the classification accuracy when compared with the first-order model.     

Predicting SME loan delinquencies during recession using accounting data and SME characteristics: The case of Greece

The objective of this paper is the comparison of various credit‐scoring models (i.e. binomial logistic regression, decision tree, multilayer perceptron neural network, radial basis function, and support vector machine) in evaluating the risk of small and micro enterprises' (SMEs') loan delinquencies based on accounting data and applicants' specific attributes. Exploiting a representative large data set of SMEs' loans granted by a large Greek commercial bank in the expansion period, we track the evolution of SMEs' delinquencies over the recession period August 2010 to July 2012. This time frame encompasses a period of manageable levels of delays (early recession period: August 2011–July 2012) and a period when delays were increased to a very high degree (deep recession period: August 2011–July 2012). Comparison of the employed credit‐scoring models during the early recession period shows that the multilayer perceptron neural network produces the highest predicting capacity, followed by the support vector machine model. As the crisis deepens, the support vector machine model presents the highest predicting accuracy, followed by the decision tree and then the multilayer perceptron model. Generally, the predictive performance of all credit‐scoring models seems to be substantially reduced as the recession escalates. Our paper has important implications for the proper financing of SMEs given their importance for the European economy.     

Stock returns,dividend yield,and book-to-market ratio

A dividend yield model has been widely used in previous research that relates stock market valuations to cash flow fundamentals. Given controversies about using dividends as a proxy for cash flows, a loglinear book-to-market model has recently been proposed. However, these models rely on the assumption that dividend yield and book-to-market ratio are both stationary, and empirical evidence for this is, at best, mixed. We develop a new model, the loglinear cointegration model, that explains future profitability and excess stock returns in terms of a linear combination of log book-to-market ratio and log dividend yield. The loglinear cointegration model performs better than the log dividend yield model and the log book-to-market model in terms of cross-equation restriction tests and forecasting performance comparisons. The superior performance of the loglinear cointegration model suggests that the linear combination may be a better indicator of intrinsic fundamentals than the dividend yield or the book-to-market ratio separately.     

An examination of alternative CAPM-based models in UK stock returns

We evaluate the performance of unconditional and conditional versions of seven stochastic discount factor models in UK stock returns between January 1975 and December 2001. We find that the conditional four-moment capital asset pricing model (CAPM) has the best performance among the models we consider in terms of the lowest [Hansen, L.P., Jagannathan, R., 1997. Assessing specification errors in stochastic discount factor models. Journal of Finance 52, 591–607] distance measure and explaining the time-series predictability of industry portfolio excess returns. Conditional models also do a better job than unconditional models. However we find that the superior performance of the conditional four-moment CAPM, and conditional models in general, arises in part due to overfitting the data.