Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userForecasting High-Dimensional Covariance Matrices of Asset Returns with Hybrid GARCH-LSTMs
178
0
0.0
(
0
)
Authors
Lucien Boulet
Ask ChatGPT about the research
No Arabic abstract
Several academics have studied the ability of hybrid models mixing univariate Generalized Autoregressive Conditional Heteroskedasticity (GARCH) models and neural networks to deliver better volatility predictions than purely econometric models. Despite presenting very promising results, the generalization of such models to the multivariate case has yet to be studied. Moreover, very few papers have examined the ability of neural networks to predict the covariance matrix of asset returns, and all use a rather small number of assets, thus not addressing what is known as the curse of dimensionality. The goal of this paper is to investigate the ability of hybrid models, mixing GARCH processes and neural networks, to forecast covariance matrices of asset returns. To do so, we propose a new model, based on multivariate GARCHs that decompose volatility and correlation predictions. The volatilities are here forecast using hybrid neural networks while correlations follow a traditional econometric process. After implementing the models in a minimum variance portfolio framework, our results are as follows. First, the addition of GARCH parameters as inputs is beneficial to the model proposed. Second, the use of one-hot-encoding to help the neural network differentiate between each stock improves the performance. Third, the new model proposed is very promising as it not only outperforms the equally weighted portfolio, but also by a significant margin its econometric counterpart that uses univariate GARCHs to predict the volatilities.
rate research
Read More
This paper presents performance analysis of hybrid model comprise of concordance and Genetic Programming (GP) to forecast financial market with some existing models. This scheme can be used for in depth analysis of stock market. Different measures of concordances such as Kendalls Tau, Ginis Mean Difference, Spearmans Rho, and weak interpretation of concordance are used to search for the pattern in past that look similar to present. Genetic Programming is then used to match the past trend to present trend as close as possible. Then Genetic Program estimates what will happen next based on what had happened next. The concept is validated using financial time series data (S&P 500 and NASDAQ indices) as sample data sets. The forecasted result is then compared with standard ARIMA model and other model to analyse its performance.
The market events of 2007-2009 have reinvigorated the search for realistic return models that capture greater likelihoods of extreme movements. In this paper we model the medium-term log-return dynamics in a market with both fundamental and technical traders. This is based on a Poisson trade arrival model with variable size orders. With simplifications we are led to a hybrid SDE mixing both arithmetic and geometric Brownian motions, whose solution is given by a class of integrals of exponentials of one Brownian motion against another, in forms considered by Yor and collaborators. The reduction of the hybrid SDE to a single Brownian motion leads to an SDE of the form considered by Nagahara, which is a type of Pearson diffusion, or equivalently a hyperbolic OU SDE. Various dynamics and equilibria are possible depending on the balance of trades. Under mean-reverting circumstances we arrive naturally at an equilibrium fat-tailed return distribution with a Student or Pearson Type IV form. Under less restrictive assumptions richer dynamics are possible, including bimodal structures. The phenomenon of variance explosion is identified that gives rise to much larger price movements that might have a priori been expected, so that $25sigma$ events are significantly more probable. We exhibit simple example solutions of the Fokker-Planck equation that shows how such variance explosion can hide beneath a standard Gaussian facade. These are elementary members of an extended class of distributions with a rich and varied structure, capable of describing a wide range of market behaviours. Several approaches to the density function are possible, and an example of the computation of a hyperbolic VaR is given. The model also suggests generalizations of the Bougerol identity.
We define a random-matrix ensemble given by the infinite-time covariance matrices of Ornstein-Uhlenbeck processes at different temperatures coupled by a Gaussian symmetric matrix. The spectral properties of this ensemble are shown to be in qualitative agreement with some stylized facts of financial markets. Through the presented model formulas are given for the analysis of heterogeneous time-series. Furthermore evidence for a localization transition in eigenvectors related to small and large eigenvalues in cross-correlations analysis of this model is found and a simple explanation of localization phenomena in financial time-series is provided. Finally we identify both in our model and in real financial data an inverted-bell effect in correlation between localized components and their local temperature: high and low temperature/volatility components are the most localized ones.
For a GJR-GARCH specification with a generic innovation distribution we derive analytic expressions for the first four conditional moments of the forward and aggregated returns and variances. Moment for the most commonly used GARCH models are stated as special cases. We also the limits of these moments as the time horizon increases, establishing regularity conditions for the moments of aggregated returns to converge to normal moments. Our empirical study yields excellent approximate predictive distributions from these analytic moments, thus precluding the need for time-consuming simulations.
In recent years, hyperparameter optimization (HPO) has become an increasingly important issue in the field of machine learning for the development of more accurate forecasting models. In this study, we explore the potential of HPO in modeling stock returns using a deep neural network (DNN). The potential of this approach was evaluated using technical indicators and fundamentals examined based on the effect the regularization of dropouts and batch normalization for all input data. We found that the model using technical indicators and dropout regularization significantly outperforms three other models, showing a positive predictability of 0.53% in-sample and 1.11% out-of-sample, thereby indicating the possibility of beating the historical average. We also demonstrate the stability of the model in terms of the changes in its feature importance over time.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
