Search

As the leading component of the financial market, the price formation mechanism of gold futures has been attracting extra attention of financiers and scholars. However, the data of gold futures price belongs to time series, and its forecast is very challenging owing to its chaotic, noisy, and non-stationary characteristics in data. A new forecast model named SGRU-AM, based on the special gated recurrent unit (SGRU) and attention mechanism (AM) is proposed in this paper to tackle these challenges. SGRU is the rectified model of gated recurrent unit (GRU) though executing the 1-tanh function on the reset gate output of the basic GRU to transform the value range of the reset gate value and adjusting the memory ratio between the current moment and the previous moment. Firstly, SGRU has the advantage of capturing long-distance information and can forecast the closing price of gold futures in the next trading day. Then, AM is introduced to adjust the SGRU time dimension’s feature expression, so that the model can obtain more comprehensive feature information, learn the importance of current local sequence features and improve the forecast accuracy. Taking China’s gold futures as an example, the gold futures data of the Shanghai Futures Exchange are selected from January 9, 2008, to May 31, 2021. Compared with the baseline methods, the experimental results show that SGRU-AM has the best performance among all baseline models in forecast efficiency and forecast accuracy.

The foreign exchange market (Forex) is the world’s largest market for trading foreign money, with a trading volume of over 5.1 trillion dollars per day. It is known to be very complicated and volatile. Technical analysis is the observation of past market movements with the aim of predicting future prices and dealing with the effects of market movements. A trading system is based on technical indicators derived from technical analysis. In our work, a complete trading system with a combination of trading rules on Forex time series data is developed and made available to the scientific community. The system is implemented in two phases: In the first phase, each trading rule, both the AI-based rule and the trading rules from the technical indicators, is tested for selection; in the second phase, profitable rules are selected among the qualified rules and combined. Training data is used in the training phase of the trading system. The proposed trading system was extensively trained and tested on historical data from 2010 to 2021. To determine the effectiveness of the proposed method, we also conducted experiments with datasets and methodologies used in recent work by Hernandez-Aguila et al., 2021 and by Munkhdalai et al., 2019. Our method outperforms all other methodologies for almost all Forex markets, with an average percentage gain of 20.2%. A particular focus was on training our AI-based rule with two different architectures: the first is a widely used convolutional network for image classification, i.e. ResNet50; the second is an attention-based network Vision Transformer (ViT). The results provide a clear answer to the main question that guided our research and which is the title of this paper.

Stock price forecasting is a relevant and challenging problem that has attracted a lot of interest from engineers and scientists. In this paper we apply two techniques for stock price and price intervals forecasting. Both techniques, derived from previous works by the authors, are based on the use of local data extracted from a database. These data are those that correspond to similar market states to the current one. The first technique uses these local data to compute a price forecast by finding an optimal combination of past states that equals the current state. The price forecast is then obtained by combining the past actual prices associated to the past market states. The second technique can be used to forecast prices but its main use is to forecast price intervals that will contain the real future price with a guaranteed probability. This is accomplished by building a probability distribution for the forecasted price and then setting the intervals by a choice of desired percentiles. Thus, this technique can be used in financial risk management. Both techniques are purely data driven and do not need a theoretical description or model of the price trend being forecasted. The proposed techniques adapt very easily to market changes because they use only the subset of the database that it is closer to the current state. Furthermore, the database can be updated as new data is available. Finally, both approaches are highly parallelizable, thus making possible to manage large data sets. As a case study, the proposed approaches have been applied to the $k$ -step forecasting of the Dow Jones Industrial Average index. The results have been validated in relation with some baseline approaches, such as martingale and neural network predictors and quantile regression for the interval forecasting.

Research on crude oil price forecasting has attracted tremendous attention from scholars and policymakers due to its significant effect on the global economy. Besides supply and demand, crude oil prices are largely influenced by various factors, such as economic development, financial markets, conflicts, wars, and political events. Most previous research treats crude oil price forecasting as a time series or econometric variable prediction problem. Although recently there have been researches considering the effects of real-time news events, most of these works mainly use raw news headlines or topic models to extract text features without profoundly exploring the event information. In this study, a novel crude oil price forecasting framework, AGESL, is proposed to deal with this problem. In our approach, an open domain event extraction algorithm is utilized to extract underlying related events, and a text sentiment analysis algorithm is used to extract sentiment from massive news. Then a deep neural network integrating the news event features, sentimental features, and historical price features is built to predict future crude oil prices. Empirical experiments are performed on West Texas Intermediate (WTI) crude oil price data, and the results show that our approach obtains superior performance compared with several benchmark methods., Comment: 14 pages, 5 figures, 5 tables

Scenario forecasting methods have been widely studied in recent years to cope with the wind power uncertainty problem. The main difficulty of this problem is to accurately and comprehensively reflect the time-series characteristics and spatial-temporal correlation of wind power generation. In this paper, the marginal distribution model and the dependence structure are combined to describe these complex characteristics. On this basis, a scenario generation method for multiple wind farms is proposed. For the marginal distribution model, the autoregressive integrated moving average-generalized autoregressive conditional heteroskedasticity-t (ARIMA-GARCH-t) model is proposed to capture the time-series characteristics of wind power generation. For the dependence structure, a time-varying regular vine mixed Copula (TRVMC) model is established to capture the spatial-temporal correlation of multiple wind farms. Based on the data from 8 wind farms in Northwest China, sufficient scenarios are generated. The effectiveness of the scenarios is evaluated in 3 aspects. The results show that the generated scenarios have similar fluctuation characteristics, autocorrelation, and crosscorrelation with the actual wind power sequences.

The Hurst index is widely used to describe the long memory process of time series in economics, finance, and other fields. The setting for Hurst index estimation has not been thoroughly investigated by current literature. Many scholars choose the estimation settings subjectively without comparing the results of different settings or just choose the settings providing the desirable results. In this research, the impact of various settings such as correlation processing, trend processing, scale range selection, subsequence division, fitting method, and tail-cut method on the Hurst index estimation models, such as R/S and DFA, are evaluated through the Monte Carlo simulation. The results show that the settings have significant impact on the accuracy of the Hurst index estimation. The detailed optimal six setting for Hurst index estimation is summarized for future researchers. In addition, the results show that the comprehensive performance of the DFA estimation is better than the R/S-type estimation. On the basis of the optimal setting, the Hurst indexes of China’s stock market are estimated. The daily volatility in the Chinese stock market show long memory by the Hurst index. Our contribution is to identify the optimal model setting for the Hurst index estimation by comparing all the six optional settings in every step of estimation. Our stepwise analysis for model effectiveness may shed light on challenges for various approaches to market trend analysis.

Stock market volatility has a significant impact on many economic and financial activities in the world. Forecasting stock price movement plays an important role in setting an investment strategy or determining the right timing for trading. However, stock price movements are noisy, nonlinear, and chaotic. It is difficult to forecast stock trends for improving return on investment. Here, we proposed a novel improved particle swarm optimization (IPSO) and long-short term memory (LSTM) hybrid model for stock price forecasting. An adaptive mutation factor was used as a parameter for model optimization to avoid premature convergence to a local optimum. Furthermore, we presented a nonlinear approach to improve the inertia weight of the particle swarm optimization and then used the IPSO model to optimize the hyperparameters of LSTM. The experimental results showed that the proposed model outperformed the other related baseline models: support-vector regression, LSTM and PSO-LSTM on the Australian stock market index. These results indicated the proposed model possesses high reliability and good forecasting capability.

Goal: The United States (US) is currently one of the countries hardest-hit by the novel SARS-CoV-19 virus. One key difficulty in managing the outbreak at the national level is that due to the US’ diversity, geographic spread, and economic inequality, the COVID-19 pandemic in the US acts more as a series of diverse regional outbreaks rather than a synchronized homogeneous one. Method: In order to determine how to assess regional risk related to COVID-19, a two-phase modeling approach is developed while considering demographic and economic criteria. First, an unsupervised clustering technique, specifically $k$-means, is employed to group US counties based on demographic and economic similarities. Then, time series forecasting of each cluster of counties is developed to assess the short-run viral transmissibility risk. Results: To this end, we test ARIMA and Seasonal Trend Random Walk forecasts to determine which is more appropriate for modeling the spread and lethality of COVID-19. From our analysis, we then utilize the superior ARIMA models to forecast future COVID-19 trends in the clusters, and present the areas in the US which have the highest COVID-19 related risk heading into the winter of 2020. Conclusion: Including sub-national socioeconomic characteristics to data-driven COVID-19 infection and fatality forecasts may play a key role in assessing the risk associated with changes in infection patterns at the national level.

To investigate whether and how e-government affects corruption, this study collects longitudinal data (2006-2015) from 29 Chinese provinces for empirical analysis by adopting Feasible Generalized Least Squares (FGLS) and Panel Threshold Regression (PTR) model. The results of analysis show that e-government has a restraining effect on corruption and presents a significant threshold characteristics. Specifically, the restraining effect of e-government on corruption shows a dynamic non-linear characteristic, with a trend of first rising and then declining. Besides, the restraining effect is more obvious in provinces with lower e-government level.

Today, there are constant changes in terms of securities in stock markets. In these stock market investments, investors use fundamental analysis tools and indicators very widely. In this way, it is possible to have some knowledge of the situations experienced in the markets and to make a profit. In this study, manipulations on Bitcoin are discussed. Popular machine and statistical forecasting methods have been used to detect these manipulations and the road maps to be followed in order to be detected in the most successful way have been shared. Social media sentiments, which were thought to have an effect on manipulations during the studies, were also evaluated with the most advanced text analysis methods and evaluated together with these price changes. The allegations that the prediction methods carried out before the crisis were more successful were investigated. The Covid-19 pandemic was evaluated as a period of global crisis and the studies that might be relevant were examined. It would not be wrong to say that the actors that make big gains in the stock markets are the ones that determine the direction of the stock market. The manipulation periods of the market actors to be successful in the virtual money markets have been tried to be verified by various estimation methods. These estimations can achieve up to F1 score of 93% success according to our experimental result. Besides, it is stated that accounts with the highest volume of transactions in the periods, when anomalies were detected, were labeled as potential manipulators.

This study aims to explore whether the intensity of internet searches, according to the Google Trends search volume index (SVI), is a predictor of changes in real estate prices. The motivation of this study is the possibility to extend the understanding of the extra predictive power of Google search engine query volume of future housing price change (shift direction) by (i) the introduction of a research approach that combines the advantages of the complementary use of cross-correlation analysis and machine learning classification algorithms; (ii) applying the multi-class HPI values classifier which allows predicting the housing price increase, decrease or relative stability; (iii) exploiting the SVI that relates to interests in both ‘real estate’ and ‘credit to buy real estate’; (iv) evaluation of the introduced approach in the context of the Polish real estate market. The main theoretical contribution of our work is a confirmation that the freely available information regarding Google user searches can provide an in-depth insight into enriching the generally accepted statistics on supply and demand in the real estate market. From the practical perspective, this research confirms that SVI can be associated as a sole determinant to anticipate the housing price change with time-lag sufficient for making decisions regarding the purchase (sale) of individual property or the real estate market control. Such findings can be also helpful for researchers who intend to use Google Trends data as an extra variable from demand side to improve the prediction accuracy if it is included in the model which is based on the existing housing prices determinants.

Detecting electricity price spikes is crucial as it helps market participants to gain confidence and formulate appropriate strategy to maximize their benefits. In this paper, a multi-feature based approach with the incorporation of variable thresholds is developed to detect electricity price spikes in the national electricity market of Australia. The variable thresholds, which are determined using a weighted sliding window average and an adjusted standard deviation, help to segregate spikes from normal price variations. Also, significant features are extracted from the market after thoroughly analyzing the underlying causes resulting into the price spikes. These features are employed as inputs to a support vector machine to classify electricity prices as spikes or non-spikes. A case study is conducted using a dataset acquired from the state of New South Wales, Australia. The results show that the proposed method can successfully detect the price spikes with high accuracy and confidence.

Load forecasts are fundamental inputs for the reliable and resilient operation of a power system. Globally, researchers endeavor to improve the accuracy of their forecast models. However, lack of studies detailing standardized model development procedures remains a major issue. In this regard, this study advances the knowledge of the systematic development of short-term load forecast (STLF) models for electric power utilities. The proposed model has been developed by using hourly load (time series) of five years of an electric power utility in Pakistan. Following the investigation of previously developed load forecast models, this study addresses the challenges of STLF by utilizing multiple linear regression, bootstrap aggregated decision trees, and artificial neural networks (ANNs) as mutually competitive forecasting techniques. The study also highlights both rudimentary and advanced elements of data extraction, synthetic weather station development, and the use of elastic nets for feature space development to upscale its reproducibility at global level. Simulations showed the superior forecasting prowess of ANNs over other techniques in terms of mean absolute percentage error (MAPE), root mean squared error (RMSE) and R2 score. Furthermore, an empirical approach has been taken to underline the effects of data recency, climatic events, power cuts, human activities, and public holidays on the model’s overall performance. Further analysis of the results showed how climatic variations, causing floods and heavy rainfalls, could prove detrimental for a utility’s ability to forecast its load demand in future.

In this work, we present a methodology to identify COVID-19 spreaders using the analysis of the relationship between socio-cultural and economic characteristics with the number of infections and deaths caused by the COVID-19 virus in different countries. For this, we analyze the information of each country using the complex networks approach, specifically by analyzing the spreaders countries based on the separator set in 5-layer multiplex networks. The results show that, we obtain a classification of the countries based on their numerical values in socioeconomics, population, Gross Domestic Product (GDP), health and air connections; where, in the spreader set there are those countries that have high, medium or low values in the different characteristics; however, the aspect that all the countries belonging to the separator set share is a high value in air connections.

With the upgrading and adjustment of China's industrial structure, the load of the electronic information industry is increasingly important. The safety of production and electricity is a major concern in the electronic information industry. This paper adopts the recurrence interval analysis method to investigate the risk prediction of extreme load fluctuations and proposes a suitable risk threshold setting method for the electronic information industry. Initially, the probability distribution model of recurrence intervals is established by adding the stretched exponential distribution function to analyze the recurrence law of load fluctuations. Then, this paper introduces the conditional probability density function to verify the short-term memory effect on the fluctuations. The risk predication model is further established to estimate the risk probability of the recurrence of extreme load fluctuations. At last, a risk threshold is set for identifying abnormal operating conditions of electronic enterprises based on the risk probability. The results indicate that for the typical electronic enterprise, the risk probability of recurrence of extreme load fluctuations is greater within 20 minutes after the last extreme load fluctuation occurred. It is very likely that abnormal operating conditions will occur within the duration. Based on the analysis results, managers of electronic enterprises can better arrange safety inspections.

Multiple Machine Learning solutions in Industry exist where interpretability is required. In retail, this is especially important when dealing with cold-start forecasting of promotional sales. In the planning phase of these promotions, retailers produce sales predictions that are scrutinised by both forecasting experts and managers. In this paper, we combine the predictive benefits of Gradient Boosted Decision Trees regressors and the interpretability of contrastive explanations. These are implicitly generated by the manner we shape data. Our method presents the cold-start forecasts in relation to the observed promotional sales of other products, which we call neighbours . They are selected based on a measure of closeness to the predicted promotion, which is derived from the variable importance calculated during the training of the regressors. With this information, the expert reviewer adjusts the cold-start prediction by simply varying the contribution of the neighbours. To validate our results we test our method on a surrogate model, as well as on real-market data. The results on the surrogate model demonstrate that our method is able to accurately identify the features that contribute to sales and then select the closest neighbours to produce a contrastive explanation. The results on real-market data also show that the proposed method performs at a similar level to widespread methods such as conventional CatBoost , NGBoost or AutoGluon , and has the advantage of providing interpretability.

Owing to the uncertainty and volatility of wind energy, forecasted wind power scenarios with proper spatio-temporal correlations are needed in various decision-making problems involving power systems. In this study, forecasted scenarios are generated from an estimated multi-variate distribution of multiple regional wind farms. According to the theory of copulas, marginal distributions and the dependence structure of multi-variate distribution are modeled through the proposed distance-weighted kernel density estimation method and the regular vine (R-vine) copula, respectively. Owing to the flexibility of decomposing correlations of high dimensions into different types of pair-copulas, the R-vine copula provides more accurate results in describing the complicated dependence of wind power. In the case of 26 wind farms located in East China, high-quality forecasted scenarios as well as the corresponding probabilistic forecasting and point forecasting results are obtained using the proposed method, and the results are evaluated using a comprehensive verification framework.

The global horizontal irradiance data recorded at the earth's horizontal surface is a mixture of deterministic (extra-terrestrially) and stochastic components due to the ever-changing atmospheric conditions. The Box-Jenkins short memory stochastic models and their hybrid versions have been used successfully to forecast solar irradiance data. However, these models lack robustness and faulter for distant horizon forecasting such as more than 2 hours in the hourly case. Using a quantile regression model as both a robust benchmark and robustification model, we remedy this drawback by using long memory models and their hybrid versions in 24 hour ahead irradiance forecasting at three sites in South Africa (RVD, UPR and SUN). Results show that the inherent long memory is anti-persistent in all models but one that shows persistence. The forecasts from these long memory-based models are generally within the prediction interval of the benchmark model forecasts for two sites. Based on the RMSE, rRMSE, MAE, rMAE and the pinball loss function at the central quantile level and two extreme quantile levels, forecasts obtained through robustification via forecasts combinations with the quantile regression yield the best performance. The predictive ability of the best model before forecasts combination against the quantile regression model combined (robustified) model based Diebold-Mariano test rejected the null hypothesis of the same accuracy. As a follow up to distinguish which has a superior predictive accuracy, “murphy diagrams” were used. For the RVD and UPR sites, the model based on forecast combination has the best predictive accuracy while the quantile regression based benchmark model has the best accuracy for the SUN data.

To quantify travel demand, it is necessary to understand the travelers’ mode choice behavior. The Logit model is widely used in travel mode choice because of its closed form. Nevertheless, the variance of the utility function is unchanged in Logit-based models, indicating that the perceived error of the traveler on the option is fixed as the utility changes, which is inconsistent with the actual situation. While in Weibit-based models, travelers’ perception error of options grows with the increase of the utility. Moreover, the relative difference is captured, and the asymmetric property exists, which is different from Logit-based models. This paper contributes to the literature by comparing the performances of the Logit-based and Weibit-based models. In this article, six discrete choice models for travel mode choice are discussed based on data of Swiss metro, which includes multinomial Logit model, multinomial Weibit model, and derived models. The Weibit-based models outperform the Logit-based models, considering with the adjusted likelihood ratio index of all models in this paper.

Recurrent neural networks have received vast amount of attention in time series prediction due to their flexibility in capturing dependencies on various scales. However, as in most of the classical forecasting methods, its accuracy is strongly tied to the degree of signal complexity. Specifically, stock market prices are commonly classified to be non-linear, non-stationary and chaotic signals, since they exhibit erratic behavior that conducts a poor performance in the long short-term memory (LSTM). In this paper, we propose a methodology to improve the predictability of financial time series by using the complete ensemble empirical mode decomposition with adaptive noise and the intrinsic sample entropy (SampEn). We evaluated the integrated model by applying it to S&P 500 index stocks for the period between January 2018 and April 2020 and for each time series of stock closing prices, an LSTM model was trained to forecast the next closing price. The experimental results represent a dependency between the decomposed signal entropy and the performance of forecast accuracy. This suggests that in those cases where the short-term complexity in financial time series is smaller compared to the series energy, the forecasting capabilities are significantly improved after the removal of decomposed highest frequency. Furthermore, our results show an improvement in forecasting the direction of the stock price by 31% using the classical LSTM architecture.

This paper selects the daily data of the exchange rates of Chinese Yuan (CNY) over the currencies of 14 countries along the Belt and Road, Shanghai composite index and Shenzhen composite index to study the influence of the Belt and Road Initiative on the linkages between exchange rates and Chinese stock index based on the flow-oriented model and the stock-oriented model. To reflect the fluctuations in daily data and reduce the central bank’s interference with the exchange rate, two fuzzy techniques are used to process data, that is, the centroid based measure and the integral based measure. Then we judge the relationship between exchange rate and stock index through the Pearson correlation coefficient and the Granger causality test. Besides, we further compare the results and their differences by the classic crisp method and our two fuzzy techniques, which enable us to judge their correlation more accurately, and provide a reference for a wider application of the proposed fuzzy methods. We find that there is a correlation between exchange rate and stock index under certain conditions, and the Belt and Road initiative strengthens the relationship between the Chinese foreign exchange market and the stock market, more importantly, the fuzzy techniques are effective to judge this relation.

The motivation of this study is built from the previous research to find a way to enhance the forecast of advanced and emerging market currency volatilities. Given the exchange rate’s nonlinear and time-varying characteristics, we introduce the neural networks (NN) approach to enhance the Markov Switching Beta-Exponential Generalized Autoregressive Conditional Heteroscedasticity (MS-Beta-t-EGARCH) model. Our hybrid model synthesizes these two approaches’ advantages to predict exchange rate volatility. We validate the performance of our proposed model by comparing it with various traditional volatility forecasting models. In-sample and out-of-sample volatility forecasts are considered to achieve our comparison. The empirical results suggest that our hybrid NN-MS Beta-t-EGARCH outperforms the other models for both emerging and advanced market currencies.

Reliable poverty data are critical for regional economic analysis and policy making, especially considering that economic inequality and sustainable development are widespread social concerns. This article proposes a multitask learning model combining deep residual neural networks and feature pyramid networks to estimate poverty level from multiple sources including the night-time light data, Landsat 8 imagery, and spectral index data. We first train the multitask learning model using the multisource data in Chongqing, China and then estimate the representative economic indicators in the study area. The model is evaluated with the Pearson correlation coefficient of the actual and estimated economic indicators. The result shows that the proposed model outperforms other models with the Pearson correlation coefficient up to 0.87 in the annual estimates of economic indicators between 2013 and 2017. As all the data used in this article are publicly available, the proposed model can be used to estimate the economic indicators in other regions as well.

Coronavirus Disease 2019 (COVID-19) has spread the world resulting in detrimental effects on human health, lives, societies, and economies. The state authorities mostly take non-pharmacological actions against the outbreak since there are no confirmed vaccines or treatments yet. In this paper, we developed Suspicious-Infected-Death with Non-Pharmacological policies (SpID-N) model to analyze the properties of the COVID-19 casualties and also estimate the future behavior of the outbreak. We can state the key contributions of the paper with three folds. Firstly, we propose the SpID-N model covering the higher-order internal dynamics which cause the peaks in the casualties. Secondly, we parametrize the non-pharmacological policies such as the curfews on people with chronic disease, people age over 65, people age under 20, restrictions on the weekends and holidays, and closure of the schools and universities. Thirdly, we explicitly incorporate the internal and coupled dynamics of the model with these multi-dimensional non-pharmacological policies. The corresponding higher-order and strongly coupled model has utterly unknown parameters and we construct a batch type Least Square (LS) based optimization algorithm to learn these unknown parameters from the available data. The parametric model and the predicted future casualties are analyzed extensively.

In this paper, we show a simple but novel approach in an attempt to improve value-at-risk forecasts. We use mutually dependent covariate returns to create exogenous break variables and jointly use the variables to augment GARCH models to account for time-variations and breaks in the unconditional volatility processes simultaneously. A study of hypothetical mutual dependencies between volatility and the covariates is first carried out to investigate the levels of the shared mutual information among the variables before using the augmented models to forecast 1% and 5% value-at-risks. The results provide evidence of some substantial exchange of information between volatility and the lagged exogenous covariates. In addition, the results show that the estimated augmented models have lower volatility persistence, reduced information leakages, and improved explanatory powers. Furthermore, there is evidence that our approach leads to fewer violations, improved 1% value-at-risk forecasts, and optimal daily capital requirements for all the models. There is, however evidence of relative superiority of the majority of the models for the 5% value-at-risks forecasts from our approach, although they have relatively higher failure rates. Based on these results, we recommend the incorporation of our approach to existing risk modeling frameworks. It is believed that such models may lead to fewer bank failures, expose banks to optimal market risks, and assist them in computing optimal regulatory capital requirements and minimize penalties from regulators.

The oil price is influenced not only by the fundamentals of supply and demand but also by unpredictable political conflicts, climate emergencies, and investor intentions, which cause enormous short-term fluctuations in the oil price. The proposition of the Google index-driven decomposition ensemble model to forecast crude oil price risk uses big data technology and a time series decomposition method. First, by constructing an index of investor attention for the market and emergencies combined with a bivariate empirical mode decomposition, we analyze the impact of investor attention on oil price fluctuations. Second, we establish a vector autoregression model, and the impulse responses define the impact of emergencies on the crude oil price. Finally, with the help of machine learning and historical simulation methods, the risk of crude oil price shocks from unexpected events is predicted. Empirical research demonstrates that concerns related to the oil market and emergencies that appear in Google search data are closely related to changes in oil prices. Based on the Google index, our model’s prediction of crude oil prices is more accurate than other models, and the prediction of value-at-risk is closer to the theoretical value than the historical simulation with the ARMA forecasts method. Considering the impact of emergencies in the prediction of crude oil price risk can help provide technical guidance for investors and risk managers and avoid economic risks caused by climate disasters or political conflicts.

Over the past few years, with the advent of blockchain technology, there has been a massive increase in the usage of Cryptocurrencies. However, Cryptocurrencies are not seen as an investment opportunity due to the market's erratic behavior and high price volatility. Most of the solutions reported in the literature for price forecasting of Cryptocurrencies may not be applicable for real-time price prediction due to their deterministic nature. Motivated by the aforementioned issues, we propose a stochastic neural network model for Cryptocurrency price prediction. The proposed approach is based on the random walk theory, which is widely used in financial markets for modeling stock prices. The proposed model induces layer-wise randomness into the observed feature activations of neural networks to simulate market volatility. Moreover, a technique to learn the pattern of the reaction of the market is also included in the prediction model. We trained the Multi-Layer Perceptron (MLP) and Long Short-Term Memory (LSTM) models for Bitcoin, Ethereum, and Litecoin. The results show that the proposed model is superior in comparison to the deterministic models.

The COVID-19 virus that began in late December 2019 continues to spread rapidly in many countries around the world. Due to its contagious and fast-spreading nature, it causes great harm to countries economically, medically, socially and in all other areas. Therefore, it is imperative to predict the evolution and spread of the epidemic. By understanding the trend of developing confirmed cases in an area, governments can control the epidemic by launching appropriate plans and instructions.Many scientists have tried to predict the number of cases using traditional mathematical techniques; however, the common traditional mathematical differential equations have limitations in estimating cases numbers in time series data and even have major errors in estimation. To solve this problem, we propose an improved method for predicting validated states based on the LSTM (long-term memory) neural network.Since the traditional prediction models predict the number of cumulative cases only, so they expect that the rate of infections will always rise and they cannot predict when the spread of the virus will decrease or end, so our model is built on short-term memory that predicts the number of daily cases but not the number of cumulative cases (LSTM).

Automobiles in the late 2000s are huge role players in real life. People try to understand the basic features of automobiles when they decide to buy them. More specifically, when people decide to buy an automobile, they look for the specifications of the automobile like fuel type, engine power, engine capacity, etc. to find the most suitable automobile for the needs and budget. The aim and main responsibility of this study are to make an exploratory data analysis on a specific dataset that consists of automobile specifications within a range of years, to find the relationships between these features and the correlations with the price feature, and to build an automobile price prediction machine learning model with the comparison of real prices and predicted prices at the end of the study. The study is carried on with the dataset that is collected from a Turkey’s automobile seller website arabam.com with a web scraper written in Python. Linear Regression technique is used during the model evaluation phase. It is found that the predicted prices are matched with the real prices with a nearly 75% success rate. Some limitations are applied to both dataset and for the price feature to obtain better results for the Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) metrics.

In this paper, a form of adaptive weak-form market efficiency of the Romanian stock market is examined via a state-space approach. A general methodology for estimating evolving efficiency is proposed within the extended Kalman filter associated with the customized state-space models in this work. The suggested approach for measuring a time-varying degree of market efficiency is based on time-dependent autoregressive (AR) models where a conclusion regarding the evolving efficiency is deduced by tracking the changes in regression coefficients. Motivated by previous studies of the Bucharest stock exchange, we explore a structural time-series model that contains a stochastic trend and a white noise component. Monthly data covers the period from September 1997 to December 2020, which includes the period of the 2008–2009 global financial crisis and the recent COVID-19 recession. Our major findings reveal that the Romanian stock market is weak-form efficient at the end of our empirical study and has not been greatly affected by the first wave of the COVID-19 pandemic. However, it suffered severely during the 2008–2009 crisis, which is reflected in a lingering recovery process.

Data freshness, measured by Age of information (AoI), is becoming an increasingly significant metric for data valuation. However, most existing data trading markets ignore the impact of such a metric. In this paper, we study a fresh data market, where users with heterogeneous valuations for AoI stochastically arrive over time. The platform decides data sampling (which affects the AoI) and pricing policies (to the users), to maximize its profit. We consider three types of pricing policies with increasing flexibility, i.e., a uniform pricing policy, a dual pricing policy, and a dynamic pricing policy. The joint data sampling and pricing optimization is a non-smooth mixed integer programming problem, which is challenging to solve. Despite the difficulty, we derive the closed-form solutions of the optimal data sampling policies and pricing policies for all three cases. Our analysis yields several interesting practical insights. First, the optimal data prices decrease in the unit sampling cost and increase in the users’ arrival rate. Second, for all three pricing policies, the equal-spacing data sampling policy is optimal. Third, numerical results show that the optimal dual pricing policy significantly outperforms the optimal uniform pricing policy. Specifically, the optimal dual pricing policy produces up to 280% of the profit that is achieved by the optimal uniform pricing policy.

In this paper we propose a Quantile Regression Deep Neural Network capable of forecasting multiple quantiles in one model using a combined quantile loss function, and apply it to probabilistically forecast the prices of 8 European Day Ahead Markets. We show that the proposed loss function significantly reduces the quantile crossing problem to (near) 0% in all markets considered, while in some cases simultaneously increasing forecasting performance based on classical point forecast metrics applied to the expected value of the probabilistic forecast. The models are optimized using an automated approach with an elaborate feature- and hyperparameter search space, leading to good model performance in all considered markets.