Mohammad Ali Jafari; Sina Ghasemilo
Abstract
Predicting time series has always been one of the challenges in the financial markets. With the increase in the amount of data, the need to use modern tools instead of classical statistical and time series methods has become clear. In this paper, some deep learning algorithms such as Multilayer Perceptrons ...
Read More
Predicting time series has always been one of the challenges in the financial markets. With the increase in the amount of data, the need to use modern tools instead of classical statistical and time series methods has become clear. In this paper, some deep learning algorithms such as Multilayer Perceptrons (MLPs), Keras Classification, Temporal Fusion Transformer (TFT, developed by Google), Extreme Learning Machine Classification (ELMC) and Propagation Hierarical Learning Network (PHILNet) are used for trading on the foreign exchange market. The efficiency and accuracy of these algorithms are presented. In this order, the EUR/USD data is used as input for the above algorithms.
Reza Taleblou
Abstract
In this paper, we evaluate the performance of two machine learning architectures— Recurrent Neural Networks (RNN) and Transformer-based models—on four commodity-based company indices from the Tehran Stock Exchange. The Transformer-based models used in this study include AutoFormer, FEDformer, ...
Read More
In this paper, we evaluate the performance of two machine learning architectures— Recurrent Neural Networks (RNN) and Transformer-based models—on four commodity-based company indices from the Tehran Stock Exchange. The Transformer-based models used in this study include AutoFormer, FEDformer, Informer, and PatchTST, while the RNN-based models consist of GRU and LSTM. The dataset comprises daily observations collected from April 20, 2020, to November 20, 2024. To enhance the generalization power of the models and prevent overfitting, we employ two techniques: splitting the training and test samples, and applying regularization methods such as dropout. Hyperparameters for all models were selected using a visual method. Our results indicate that the PatchTST model outperforms other methods in terms of Root Mean Squared Error (RMSE) for both 1-day and 5-day (1-week) forecasting horizons. The FEDformer model also demonstrates promising performance, particularly for forecasting the MetalOre time series. In contrast, the AutoFormer model performs relatively poorly for longer forecasting horizons, while the GRU and LSTM models yield mixed results. These findings underscore the significant impact of model selection and forecasting horizon on the accuracy of time series forecasts, emphasizing the importance of careful model choice and hyperparameter tuning for achieving optimal performance.
Ali Bolfake; Seyed Nourollah Mousavi; Sima Mashayekhi
Abstract
This paper proposes a new approach to pricing European options using deep learning techniques under the Heston and Bates models of random fluctuations. The deep learning network is trained with eight input hyper-parameters and three hidden layers, and evaluated using mean squared error, correlation coefficient, ...
Read More
This paper proposes a new approach to pricing European options using deep learning techniques under the Heston and Bates models of random fluctuations. The deep learning network is trained with eight input hyper-parameters and three hidden layers, and evaluated using mean squared error, correlation coefficient, coefficient of determination, and computation time. The generation of data was accomplished through the use of Monte Carlo simulation, employing variance reduction techniques. The results demonstrate that deep learning is an accurate and efficient tool for option pricing, particularly under challenging pricing models like Heston and Bates, which lack a closed-form solution. These findings highlight the potential of deep learning as a valuable tool for option pricing in financial markets.
Maryam Moradi; Najme Neshat; Amir Mohammad Ahmadzade Semeskande
Abstract
Safe investment can be experienced by incorporating human experience and modern predicting science. Artificial Intelligence (AI) plays a vital role in reducing errors in this winning layout. This study aims at performance analysis of Deep Learning (DL) and Machine Learning (ML) methods in modellingand ...
Read More
Safe investment can be experienced by incorporating human experience and modern predicting science. Artificial Intelligence (AI) plays a vital role in reducing errors in this winning layout. This study aims at performance analysis of Deep Learning (DL) and Machine Learning (ML) methods in modellingand predicting the stock returns time series based on the return rate of previous periods and a set of exogenous variables. The data used includes the weekly data of the stock return index of 200 companies included in the Tehran Stock Exchange market from 2016 to 2021. Two Long Short-Term Memory (LSTM)and Deep Q-Network (DQN) models as DL processes and two Random Forest (RF) and Support Vector Machine (SVM) models as ML algorithms were selected. The results showed the superiority of DLalgorithms over ML, which can indicate the existence of strong dependence patterns in these time series, as well as relatively complex nonlinear relationships with uncertainty between the determinant variables. Meanwhile, LSTM with R-squared equals to 87 percent and the analysis of the results of five other evaluation models have shown the highest accuracy and the least error of prediction. On the other hand, the RF model results in the least prediction accuracy by including the highest amount of error.
