Computational Economics最新文献

In this paper, we propose a new hybrid model based on a deep learning network to predict the prices of financial assets. The study addresses two key limitations in existing research: (1) the lack of standardized datasets, time scales, and evaluation metrics, and (2) the focus on prediction return. The proposed model employs a two-stage preprocessing approach utilizing Principal Component Analysis (PCA) for dimensionality reduction and de-noising, followed by Independent Component Analysis (ICA) for feature extraction. A Long Short-Term Memory (LSTM) network with five layers is fed with this preprocessed data to predict the price of the next day using a 5 day time horizon. To ensure comparability with existing literature, experiments employ an 18 year dataset of the Standard & Poor's 500 (S&P500) index and include over 40 technical indicators. Performance evaluation encompasses six metrics, highlighting the model's superiority in accuracy and return rates. Comparative analyses demonstrate the superiority of the proposed PCA-ICA-LSTM model over single-stage statistical methods and other deep learning architectures, achieving notable improvements in evaluation metrics. Evaluation against previous studies using similar datasets corroborates the model's superior performance. Moreover, extensions to the study include adjustments to dataset parameters to account for the COVID-19 pandemic, resulting in improved return rates surpassing traditional trading strategies. PCA-ICA-LSTM achieves a 220% higher return compared to the “hold and wait” strategy in the extended S&P500 dataset, along with a 260% higher return than its closest competitor in the comparison. Furthermore, it outperformed other models in additional case studies.

Graphical Abstract

This study focuses on the Black–Scholes American call option model as a moving boundary problem. Using a front-fixing approach, the model is derived as a fixed domain nonlinear parabolic problem, and the uniqueness of both the call option price and critical stock price is established. An iterative approach is established to numerically solve the problem, and the convergence of the iterative method is proved. For computational implementation, a finite difference scheme in conjunction with a second-order Runge–Kutta method is conducted. Finally, the numerical results for two test problems are reported in order to confirm our theoretical achievements.

We present a deep learning solution to address the challenges of simulating realistic synthetic first-price sealed-bid auction data. The complexities encountered in this type of auction data include high-cardinality discrete feature spaces and a multilevel structure arising from multiple bids associated with a single auction instance. Our methodology combines deep generative modeling (DGM) with an artificial learner that predicts the conditional bid distribution based on auction characteristics, contributing to advancements in simulation-based research. This approach lays the groundwork for creating realistic auction environments suitable for agent-based learning and modeling applications. Our contribution is twofold: we introduce a comprehensive methodology for simulating multilevel discrete auction data, and we underscore the potential of DGM as a powerful instrument for refining simulation techniques and fostering the development of economic models grounded in generative AI.

This study examines the influence of the sliding window in the LSTM model on its predictive performance in the stock market. The investigation encompasses three aspects: the impact of the stationarity of the original data, the effect of the time interval, and the influence of the input order of data. Additionally, a standard VAR model is established for a comparative benchmark. The experimental dataset comprises the daily stock index prices of the six major stock markets from the January 2010 to December 2019. The experimental results demonstrate that stationary input data enhances the predictive performance of the LSTM model. Furthermore, shorter time interval tends to yield improved outcomes, while the order of input data does not impact the performance of the LSTM. Although the predictive capability of the LSTM model may not consistently surpass that of the standard VAR model, which is different from the previous research, it serves to compensate for the conditional limitations associated with VAR model construction.

Since Kopel’s duopoly model was proposed about 3 decades ago, there are almost no analytical results on the equilibria and their stability in the asymmetric case. The first objective of our study is to fill this gap. This paper analyzes the asymmetric duopoly model of Kopel analytically by using several tools based on symbolic computations. We discuss the possibility of the existence of multiple positive equilibria and establish conditions for a given number of positive equilibria to exist. The possible positions of the equilibria in Kopel’s model are also explored. Furthermore, in the asymmetric model of Kopel, if the duopolists adopt the best response reactions or homogeneous adaptive expectations, we establish conditions for the local stability of equilibria for the first time. The occurrence of chaos in Kopel’s model seems to be supported by observations through numerical simulations, which, however, is challenging to prove rigorously. The second objective of this paper is to prove the existence of snapback repellers in Kopel’s map, which implies the existence of chaos in the sense of Li–Yorke according to Marotto’s theorem.

In an experimental study, we investigated the application of artificial neural networks (ANNs) and long-tail probability ranking in constructing investment portfolios to achieve superior returns compared to a benchmark. Our objective is to demonstrate that portfolio formation can be conceptualized as a classification problem by leveraging the inherent capabilities of ANNs to capture complex relationships and facilitate more informed decisions regarding portfolio composition. We conducted the experiment using lagged asset return information to predict stock returns, employing a pilot sample of 70 assets and a validation sample consisting of all companies belonging to the Standard & Poor's 500 (S&P 500) index. The study covers the period from 2018 to 2022, with 585,650 daily observations of active assets. The results indicate that the classification method proposed in this study, using the asymmetric probabilities of the Student´s (t) distribution, outperforms the market and traditional portfolios. Furthermore, the results suggest that the combined approach of ANN and security classification based on their asymmetric leptokurtic probabilities demonstrates superiority over portfolios that rely solely on security signal classification.

The complexity and ambiguity of financial and economic systems, along with frequent changes in the economic environment, have made it difficult to make precise predictions that are supported by theory-consistent explanations. Interpreting the prediction models used for forecasting important macroeconomic indicators is highly valuable for understanding relations among different factors, increasing trust towards the prediction models, and making predictions more actionable. In this study, we develop a fundamental-based model for the Canadian–U.S. dollar exchange rate within an interpretative framework. We propose a comprehensive approach using machine learning to predict the exchange rate and employ interpretability methods to accurately analyze the relationships among macroeconomic variables. Moreover, we implement an ablation study based on the output of the interpretations to improve the predictive accuracy of the models. Our empirical results show that crude oil, as Canada’s main commodity export, is the leading factor that determines the exchange rate dynamics with time-varying effects. The changes in the sign and magnitude of the contributions of crude oil to the exchange rate are consistent with significant events in the commodity and energy markets and the evolution of the crude oil trend in Canada. Gold and the TSX stock index are found to be the second and third most important variables that influence the exchange rate. Accordingly, this analysis provides trustworthy and practical insights for policymakers and economists and accurate knowledge about the predictive model’s decisions, which are supported by theoretical considerations.

Copper prices are commonly used as indicators of economic development due to the increased operational risks of copper trading companies caused by their fluctuations and the effect on the government's ability to formulate market regulation policies. However, due to the high volatility of copper prices and resulting database discrepancies, traditional models exhibit lower accuracy and limited applicability. In this study, an improved hybrid prediction model based on the Butterfly Optimization Algorithm (BOA) and the Least Squares Support Vector Machine (LSSVM) is proposed. Firstly, the BOA is introduced to optimize the hyperparameters of the LSSVM. Then principal component analysis (PCA) is applied to data preprocessing, and the correlations of principal components are used to analyze and select model variables. To compare the forecasting accuracy and generalization ability based on the dataset of copper prices, some models are applied to establish multiple copper-price forecast cases, short-term, medium-term, and long-term. The results indicate that the PCA-BOA-LSSVM model demonstrates the most significant improvement, particularly in long-term forecasting cases. The highest optimization rate for RMSE reach 55.61%. The evaluation metrics of RMSE and MAPE for each case do not exceed 0.5 and 0.1, respectively, while R² remains above 0.6. In conclusion, this study provides a high-precision model for short-term, medium-term, and long-term forecasts of copper prices and provides reliable theoretical support for government policy adjustment and market investment.

In this paper, two approaches for measuring the distance between stock returns and the network connectedness are presented that are based on the Pearson correlation coefficient dissimilarity and the generalized variance decomposition dissimilarity. Using these two procedures, the center of the network is determined. Also, hierarchical clustering methods are used to divide the dense networks into sparse trees, which provide us with information about how the companies of a financial market are related to each other. We implement the derived theoretical results to study the dynamic connectedness between the companies in the Swedish capital market by considering 28 companies included in the determination of the market index OMX30. The network structure of the market is constructed using different methods to determine the distance between the companies. We use hierarchical clustering methods to find the relation among the companies in each window. Next, we obtain a one-dimensional time series of the distances between the clustering trees that reflect the changes in the relationship between the companies in the market over time. The method from statistical process control, namely the Shewhart control chart, is applied to those time series to detect abnormal changes in the financial market.