Application of Enhanced Hidden Markov Model in Stock Price Prediction

2020 ◽  
Vol 3 (1) ◽  
pp. 70-78
Author(s):  
Donata D. Acula ◽  
Teofilo De Guzman
Keyword(s):  
Neural Networks ◽  
Markov Model ◽  
Hidden Markov Model ◽  
Stock Price ◽  
Hidden Markov ◽  
Moving Average ◽  
High Price ◽  
Percentage Error ◽  
Autoregressive Integrated Moving Average ◽  
Stock Price Prediction

The main focus of this research is the enhancement of the Hidden Markov Model by using some features of Neural Networks and the forecasted values of predictors by Seasonal Autoregressive Integrated Moving Average. The enhanced method was used to predict the close price of stocks whose predictors are open price, high price, low price, and volume of Apple and Nokia data. The performance of the method was measured using the Mean Absolute Percentage Error of the predicted price. The result was compared against the actual close price by using the paired T-test. The testing of the hypothesis showed that the Enhanced Hidden Markov Model obtained more than 94% accuracy rate. It also shows that in Apple data, the predicted close price of the Enhanced Hidden Markov Model is significantly better than the predicted close price of Neural Networks. Using Nokia data, the test claims that there is no difference between the performance of Enhanced Hidden Markov Model and Neural Network in prediction. 

Point and Figure Portfolio Optimization using Hidden Markov Models and Its Application on the Bumi Resources Tbk Shares

2021 ◽  
Vol 3 (1) ◽  
pp. 44-52
Author(s):  
Kastolan Kastolan ◽  
Berlian Setiawaty ◽  
N. K. Kutha Ardana
Keyword(s):  
Markov Model ◽  
Hidden Markov Model ◽  
Portfolio Optimization ◽  
Expectation Maximization ◽  
Stock Prices ◽  
Stock Price ◽  
Hidden Markov ◽  
Expectation Maximization Algorithm ◽  
Percentage Error ◽  
Time Sampling

AbstractThe problem of portfolio optimization is to select a trading strategy which maximizes the expected terminal wealth. Since the stocks are traded at discrete random times in a real-world market, we are interested in a time sampling method. The sampling of stock price is obtained from the process of time sampling which is used in a point and figure chart. Point and figure (PF) chart displays the up and down movements of unbalanced stock prices. The basic idea is to describe essential movements of the unbalanced stock prices using a hidden Markov model. The model parameters are transition probability matrices. They are estimated using maximum likelihood method and expectation maximization algorithm. The estimation procedure involves change of measure. The model is then applied to the stock price of Bumi Resources Tbk. collected on a daily basis. The estimated parameters are used to calculate the optimal portfolio using a recursive algorithm. The results show that the discrete hidden Markov model can be applied to describe essential movements of the stock price. The best result gives 93.63% accuracy of the estimate of observation sequence with mean absolute percentage error (MAPE) 3.63%. The numerical calculation shows that the optimal logarithmic PF-portfolio increases the wealth.Keywords: point and figure portfolio; optimization portfolio; discrete hidden Markov model; expectation maximization algorithm; stock price of Bumi Resources Tbk. AbstrakMasalah pengoptimalan portofolio adalah pemilihan strategi perdagangan yang dapat memaksimalkan kekayaan terminal yang diharapkan. Karena di pasar dunia nyata, saham diperdagangkan pada waktu acak yang berbeda, sehingga kami tertarik pada metode pengambilan sampel waktu. Proses pengambilan sampel waktu diperoleh sampling harga saham yang digunakan dalam diagram point and figure (PF-chart). Grafik point and figure hanya menampilkan pergerakan naik atau turun harga saham yang tidak seimbang. Ide dasarnya adalah untuk mendeskripsikan pergerakan esensial dari harga saham yang tidak seimbang menggunakan model hidden Markov. Parameter dari model ini adalah matriks probabilitas transisi. Parameter diestimasi menggunakan metode maximum likelihood dan algoritma expectation maximization. Prosedur estimasi melibatkan perubahan ukuran. Model ini kemudian diaplikasikan pada harga saham Bumi Resources Tbk. dari tanggal 2 Januari 2007 sampai dengan 31 Januari 2011. Hasil estimasi parameter tersebut digunakan untuk menghitung portofolio optimal menggunakan algoritma rekursif. Hasil penelitian ini menunjukkan bahwa model hidden Markov diskrit dapat diterapkan untuk menggambarkan pergerakan esensial dari harga saham. Model terbaik memberikan akurasi 93.63% dari estimasi deretan observasi dengan mean absolute percentage error (MAPE) 3,63% dan 5 faktor penyebab kejadian. Perhitungan numerik menunjukkan bahwa logaritma portofolio-PF yang optimal dapat meningkatkan kekayaan.Kata kunci: portofolio point and figure; optimalisasi portofolio; model hidden Markov diskrit; algoritma expectation maximization; harga saham PT Bumi Resources.

Moving average filtering with deconvolution (MAD) for hidden Markov model with filtering and correlated noise

2019 ◽  
Vol 48 (4) ◽  
pp. 383-393
Author(s):  
Ibrahim M. Almanjahie ◽  
Ramzan Nazim Khan ◽  
Robin K. Milne ◽  
Takeshi Nomura ◽  
Boris Martinac
Keyword(s):  
Markov Model ◽  
Hidden Markov Model ◽  
Hidden Markov ◽  
Moving Average ◽  
Correlated Noise

Hidden Markov model interpretations of neural networks

2000 ◽  
Vol 23 (4) ◽  
pp. 494-495
Author(s):  
Ingmar Visser
Keyword(s):  
Neural Networks ◽  
Markov Model ◽  
Hidden Markov Model ◽  
Hidden Markov Models ◽  
Markov Models ◽  
Hidden Markov ◽  
Distributed Representations ◽  
Hidden Layer

Page's manifesto makes a case for localist representations in neural networks, one of the advantages being ease of interpretation. However, even localist networks can be hard to interpret, especially when at some hidden layer of the network distributed representations are employed, as is often the case. Hidden Markov models can be used to provide useful interpretable representations.

scholarly journals Development of Vietnamese Speech Synthesis System using Deep Neural Networks

2019 ◽  
Vol 34 (4) ◽  
pp. 349-363 ◽  
Author(s):  
Thinh Van Nguyen ◽  
Bao Quoc Nguyen ◽  
Kinh Huy Phan ◽  
Hai Van Do
Keyword(s):  
Neural Networks ◽  
Markov Model ◽  
Hidden Markov Model ◽  
Speech Synthesis ◽  
Deep Neural Networks ◽  
Hidden Markov ◽  
Experimental Results ◽  
Training Data ◽  
The Internet ◽  
Synthesis System

In this paper, we present our first Vietnamese speech synthesis system based on deep neural networks. To improve the training data collected from the Internet, a cleaning method is proposed. The experimental results indicate that by using deeper architectures we can achieve better performance for the TTS than using shallow architectures such as hidden Markov model. We also present the effect of using different amounts of data to train the TTS systems. In the VLSP TTS challenge 2018, our proposed DNN-based speech synthesis system won the first place in all three subjects including naturalness, intelligibility, and MOS.

scholarly journals Comparison of Stock Price Prediction Models using Pre-trained Neural Networks

2021 ◽  
Vol 3 (2) ◽  
pp. 122-134
Author(s):  
C Anand
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Stock Price ◽  
Linear Models ◽  
Moving Average ◽  
Stock Price Prediction ◽  
Price Prediction ◽  
Data Prediction ◽  
Comparison Results ◽  
Auto Regressive

Several intelligent data mining approaches, including neural networks, have been widely employed by academics during the last decade. In today's rapidly evolving economy, stock market data prediction and analysis play a significant role. Several non-linear models like neural network, generalized autoregressive conditional heteroskedasticity (GARCH) and autoregressive conditional heteroscedasticity (ARCH) as well as linear models like Auto-Regressive Integrated Moving Average (ARIMA), Moving Average (MA) and Auto Regressive (AR) may be used for stock forecasting. The deep learning architectures inclusive of Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Recurrent Neural Networks (RNN), Multilayer Perceptron (MLP) and Support Vector Machine (SVM) are used in this paper for stock price prediction of an organization by using the previously available stock prices. The National Stock Exchange (NSE) of India dataset is used for training the model with day-wise closing price. Data prediction is performed for a few sample companies selected on a random basis. Based on the comparison results, it is evident that the existing models are outperformed by CNN. The network can also perform stock predictions for other stock markets despite being trained with single market data as a common inner dynamics that has been shared between certain stock markets. When compared to the existing linear models, the neural network model outperforms them in a significant manner, which can be observed from the comparison results.

scholarly journals Comparison of Financial Models for Stock Price Prediction

2020 ◽  
Vol 13 (8) ◽  
pp. 181
Author(s):  
Mohammad Rafiqul Islam ◽  
Nguyet Nguyen
Keyword(s):  
Neural Network ◽  
Stochastic Process ◽  
Predictive Models ◽  
Stock Price ◽  
Financial Models ◽  
Moving Average ◽  
Autoregressive Integrated Moving Average ◽  
Stock Price Prediction ◽  
Price Prediction ◽  
The Neural Network

Time series analysis of daily stock data and building predictive models are complicated. This paper presents a comparative study for stock price prediction using three different methods, namely autoregressive integrated moving average, artificial neural network, and stochastic process-geometric Brownian motion. Each of the methods is used to build predictive models using historical stock data collected from Yahoo Finance. Finally, output from each of the models is compared to the actual stock price. Empirical results show that the conventional statistical model and the stochastic model provide better approximation for next-day stock price prediction compared to the neural network model.

Sign in / Sign up

Export Citation Format

Share Document