Non‐linear forecasts of stock returns

Kanas, Angelos

doi:10.1002/for.858

Cited by 48 publications

(23 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In another study of US listed company share prices and dividends, Kanas (2003) compares the forecasts generated by parametric standard and Markov regime switching models with the ones produced by a simultaneous (multivariate) nearest-neighbour (SNN) approach and ANNs and finds SNN to perform similarly to the others in terms of accuracy, but not as well as the Markov switching model in terms of forecast encompassing. In a recent study of short term stock price reactions to equity offering announcements, Bozos and Nikolopoulos (2011) compare forecasts by parametric, non-parametric models and expert judges and conclude that k-NN methods ranked high in terms of economic performance of the forecasts, despite their forecast accuracy being relatively lower.…”

Section: Economics and Financementioning

confidence: 99%

Forecasting supply chain sporadic demand with nearest neighbor approaches

Νικολόπουλος

Babaï

Bozos

2016

International Journal of Production Economics

View full text Add to dashboard Cite

One of the biggest challenges in Supply Chain Management (SCM) is to forecast sporadic demand. Our forecasting methods arsenal includes Croston s method, SBA and TSB as well as some more recent non-parametric advances, but none of these can identify and extrapolate patterns existing in data; this is essential as these patterns do appear quite often, driven by infrequent but nevertheless repetitive managerial practices. One could claim such patterns could be picked up by Artificial Intelligence approaches, however these do need large training datasets, unfortunately non-existent in industrial time series. Nearest Neighbors (NN) can however operate in these latter contexts, and pick up patterns even in short series. In this research we propose applying NN for supply chain data and we investigate the conditions under which these perform adequately through an extensive simulation. Furthermore, via an empirical investigation in automotive data we provide evidence that practitioners could benefit from employing supervised NN approaches. The contribution of this research is not in the development of a new theory, but in the proposition of a new conceptual framework that brings existing theory (i.e. NN) from Computer Science and Statistics and applies it successfully in an SCM setting.

show abstract

Section: Economics and Financementioning

confidence: 99%

Forecasting supply chain sporadic demand with nearest neighbor approaches

Νικολόπουλος

Babaï

Bozos

2016

International Journal of Production Economics

View full text Add to dashboard Cite

show abstract

“…Hutchinson et al (1994) showed that the learning networks could be used efficiently for pricing and hedging in securities markets. Studies such as Stengos (1997, 1998), Gencay and Liu (1997), Kanas (2003), Kanas and Yannopoulos (2001), Shively (2003) and Bildirici and Ersin (2009) applied ANN models to stock market return forecasting and financial analysis. The MLP model is evaluated as an important class of neural network models.…”

Section: Neural Network: An Overviewmentioning

confidence: 99%

Forecasting volatility in oil prices with a class of nonlinear volatility models: smooth transition RBF and MLP neural networks augmented GARCH approach

Bildirici

Ersin

2015

Pet. Sci.

View full text Add to dashboard Cite

In this study, the forecasting capabilities of a new class of nonlinear econometric models, namely, the LSTAR-LST-GARCH-RBF and MLP models are evaluated. The models are utilized to model and to forecast the daily returns of crude oil prices. Many financial time series are subjected to leptokurtic distribution, heavy tails, and nonlinear conditional volatility. This characteristic feature leads to deterioration in the forecast capabilities of traditional models such as the ARCH and GARCH models. According to the empirical findings, the oil prices and their daily returns could be classified as possessing nonlinearity in the conditional mean and conditional variance processes. Several model groups are evaluated: (i) the models proposed in the first group are the LSTAR-LST-GARCH models that are augmented with fractional integration and asymmetric power terms (FIGARCH, APGARCH, and FIAPGARCH); (ii) the models proposed in the second group are the LSTAR-LST-GARCH models further augmented with MLP and RBF type neural networks. The models are compared in terms of MSE, RMSE, and MAE criteria for in-sample and out-of-sample forecast capabilities. The results show that the LSTAR based and neural network augmented models provide important gains over the single-regime baseline GARCH models, followed by the LSTAR-LST-GARCH type models in terms of modeling and forecasting volatility in crude oil prices.

show abstract

“…Thus, some nonlinear, nonparametric alternative approaches are proposed and adopted to estimate the time series models, the prevailing representative among them is the Artificial Neural Network (ANN). Plentiful of studies on ANN denote that ANN approach outperforms traditional MLE in forecasting financial time series and, particularly, the recurrent ANN with richer dynamic structure could capture more characteristics of data in the generalization period than the feedforward one ( (Kuan, 1995), (Wu, 1995), (Tian, Juhola & Grönfors, 1997), (Lisi & Schiavo, 1999), (Ashok & Mitra, 2002), (Gaudart, Giusiano & Huiart, 2004), (Kamruzzaman & Sarker, 2004)), but some indicate mixed or opposite results ( (Adya & Collopy, 1998); (Kanas, 2003)). While the ANN is theoretically better in estimating nonlinear finite samples without invoking a probabilistic distribution, however, it has been criticized to be vulnerable to the over-fitting problem which usually leads to a local optimum and to the empirical risk minimization, same as the MLE 1 , the latter of which results in good fit and poor forecast out-of-sample.…”

Section: Introductionmentioning

confidence: 99%

Recurrent support vector regression for a non-linear ARMA model with applications to forecasting financial returns

2014

View full text Add to dashboard Cite

Motivated by the recurrent Neural Networks, this paper proposes a recurrent Support Vector Regression (SVR) procedure to forecast nonlinear ARMA model based simulated data and real data of financial returns. The forecasting ability of the recurrent SVR is compared with three competing methods, MLE, recurrent MLP and feedforward SVR. Theoretically, MLE and MLP only focus on fit in-sample, but SVR considers both fit and forecast out-of-sample which endows SVR with an excellent forecasting ability. This is confirmed by the evidence from the simulated and real data based on two forecasting accuracy evaluation metrics (NSME and sign). That is, for one-step-ahead forecasting, the recurrent SVR is consistently better than the MLE and the recurrent MLP in forecasting both the magnitude and turning points, and really improves the forecasting performance as opposed to the usual feedforward SVR.

show abstract

Non‐linear forecasts of stock returns

Cited by 48 publications

References 31 publications

Forecasting supply chain sporadic demand with nearest neighbor approaches

Forecasting supply chain sporadic demand with nearest neighbor approaches

Forecasting volatility in oil prices with a class of nonlinear volatility models: smooth transition RBF and MLP neural networks augmented GARCH approach

Recurrent support vector regression for a non-linear ARMA model with applications to forecasting financial returns

Contact Info

Product

Resources

About