Generating asset-specific trading signals based on the financial conditions of the assets is one of the challenging problems in automated trading. Various asset trading rules are proposed ...experimentally based on different technical analysis techniques. However, these kind of trading strategies are profitable, extracting new asset-specific trading rules from vast historical data to increase total return and decrease the risk of portfolios is difficult for human experts. Recently, various deep reinforcement learning (DRL) methods are employed to learn the new trading rules for each asset. In this paper, a novel DRL model with various feature extraction modules is proposed. The effect of different input representations on the performance of the models is investigated and the performance of DRL-based models in different markets and asset situations is studied. The proposed model in this work outperformed the other state-of-the-art models in learning single asset-specific trading rules and obtained almost 12.4% more profit over the best state-of-the-art model on the Dow Jones Index in the same time period.
•A model to learn asset-specific trading rules using DRL technique is proposed.•Performance of DRL models in learning trading rules on single assets is studied.•The effects of different feature extraction models and input types are studied.•DRL models trained on OHLC data dominate the candlestick trading rules.
•Robust power trading strategies for waste-to-energy plants are studied.•A two-stage distributionally robust optimization model is developed.•A solution method based on linear decision rule ...techniques is designed.•A case study with real-world data is conducted.
Waste-to-energy (WTE) plants are operated worldwide to address the management of municipal solid waste. Against this background, an increasing number of WTE plants serve as combined heat and power (CHP) producers that supply heat to the heating systems in local districts and trade electricity in the regional power markets. This paper studies a short-term operation planning problem of determining effective power trading strategies for WTE CHP plants that participate in day-ahead markets. A two-stage distributionally robust optimization (DRO) model is developed with the consideration of uncertain electricity prices, waste supply, and district heating demand. These different kinds of uncertainty are captured by an ambiguity set that contains a collection of possible probability distributions of the uncertain parameters. The two-stage DRO model seeks to ascertain a power trading strategy that maximizes the expected profit of a WTE CHP plant on a regular operating day under the worst-case distribution in the ambiguity set. As the DRO model is intractable, a solution method based on linear decision rule techniques is designed to reformulate the model as a tractable robust linear program. To test the applicability of the DRO model, a case study with real-world data is conducted. The computational results show that the two-stage DRO model can facilitate a WTE CHP plant in obtaining economical and robust power trading strategies for regular operating days in a day-ahead market. Furthermore, the impacts of the parameters in the ambiguity set on deriving robust power trading strategies for WTE CHP plants are investigated.
With the continuous reform of the electricity market in the whole world, internal trading models in parks are generally studied in terms of load peaking, load flexibility and energy management. In ...this paper, the regional differential trading method is adopted for industrial users with different energy needs, and the interests of all parties within the industrial park are considered. Spot transactions involving surplus energy on both the supply side and the demand side are considered, and the weight allocation of transactions in each park is coordinated. Firstly, the way of circulating waste heat and power within the industrial park is analyzed from the composition of the trading system of the industrial park. The demand side is transported heat and electricity while the energy storage device is considered. The contact line is modeled on the basis of the energy transmission line characteristics. Secondly, the industrial park is disaggregated regionally using cascade analysis within the target area to decide load regulation mode between systems and modules. Thirdly, regional differences in trading are adopted for industrial users with different energy needs while the interests of all parties within the park are considered. Both sides of the supply side and the demand side participate in spot trading of surplus energy. The allocation of weights is coordinated for trading in the various industrial parks. Finally, The DICOPT solver optimizes operating costs and takes into account energy abandonment penalties. In the waste heat power transaction in the industrial park, the remaining power can be saved by 9.78 %, and the waste heat can be saved by 4.35 %. Numerical results from instances indicate that this method surpasses traditional trading approaches, aligning with the rigorous standards of scientific research.
•The DICOPT solver is used to optimize the energy operating costs of industrial parks.•By considering the penalty of waste electricity and waste heat, the economy of industrial park can be improved.•The strategy of bilateral thermal power cooperation trading in industrial park is put forward.
We use a unique set of prices from the German EPEX market and take a closer look at the fine structure of intraday markets forelectricity, with their continuous trading for individual load periods up ...to 30 min before delivery. We apply the least absolute shrinkage and selection operator (LASSO) in order to gain statistically sound insights on variable selection and provide recommendations for very short-term electricity price forecasting.
The paper proposes a trading strategy for investing in the cryptocurrency market that uses instant market entries based on additional sources of information in the form of a developed dataset. The ...task of predicting the moment of entering the market is formulated as the task of classifying the trend in the value of cryptocurrencies. To solve it, ensemble models and deep neural networks were used in the present paper, which made it possible to obtain a forecast with high accuracy. Computer analysis of various investment strategies has shown a significant advantage of the proposed investment model over traditional machine learning methods.
We define an extreme loss event as a daily return at the left tail of negative two standard deviations of all daily returns for a specific stock. Prior studies focus on the relationship between ...extreme losses and specific anticipated announcements. Our study identifies the extreme loss events after they are randomly realized, and examines the return patterns of the equities in question on stochastic event setups. We investigate the daily returns of 2,651 stocks traded in the U.S. equity markets and identified 217,990 extreme loss events from the 1950s to early 2019. Our findings show that after an extreme loss, an asset realizes, on average, a daily return of 0.8459% on the first day, and 1.8099% cumulatively in the following 5-day window. We attribute the fast recovery to the investors’ overreaction. This suggests an extreme loss reversal trading strategy. Our confirmation suggests that behavioral bias may not be corrected or eliminated through arbitrage.
In the rapid promotion of China’s electricity spot market, a large number of electricity retailers and large consumers participate in power trading, of which medium- and long-term power trading ...accounts for a large proportion. In the electricity spot market, the previous medium- and long-term transactions need to be closely combined with the current spot market transaction settlement rules. This paper analyzes the trading strategy of large retailers in the power market. In order to effectively reduce the total electricity cost, it is necessary to optimize the medium- and long-term transactions based on three aspects: electricity quantity and benchmark price decisions of medium- and long-term contracts, the daily electricity decomposition method in the day-ahead (DA) market, and the daily load curve decomposition strategy. According to load history characteristics that are extracted by the X12 method, daily electricity is decomposed from the medium- and long-term electricity quantity in the DA market. This paper introduces three methods of decomposing the daily load curve and proves that the particle swarm algorithm is the best method for effectively minimizing the cost in the DA market. Through analyzing the total electricity cost change pattern, we prove that the basic component of decision making is the relative relationship between the electricity price of medium- and long-term contracts and the equivalent kWh price of medium- and long-term electricity in the DA market, which is determined by the decomposition daily curve method. If the equivalent kilowatt-hour price obtained by the decomposition method in the DA market is greater than the electricity price of medium- and long-term contracts, the larger the electrical energy of medium- and long-term contracts, the lower the costs. Based on the above principles, electricity retailers can carry out planning for medium- and long-term transactions, as well as the decomposition and declaration of the daily electricity quantities and daily load curves.
With the deepening financialization of crude oil, an increasing number of investors are involved in trading crude oil futures. Still, because various factors influence crude oil futures price ...fluctuations, they present complex characteristics. For this reason, a trading strategy plays a crucial role in investment. In this paper, we construct an ensemble trading strategy combining time-frequency feature extraction of crude oil volume-price series with deep reinforcement learning. We decompose the original oil volume-price series by using the optimized variational mode decomposition (OVMD). Additionally, taking into account the long memory characteristics of the crude oil time series, the Sharpe ratio is used to filter the appropriate agents corresponding to the market volatility. Moreover, the best features of the three agents are effectively combined and used in the next phase of trading, thus robustly adapting to different market volatility situations. Finally, we tested our strategy using Brent crude oil futures. Compared with other benchmark models, the proposed OVMD(V–P)-DRL-Ensemble strategy is suitable for the highly volatile crude oil market and performs well in terms of the corresponding performance evaluation metrics, showing excellent return performance and stability.
•The strategy aims to exploit time-varying market efficiencies in the crude oil market for higher yields.•The strategy uses time-frequency domain extraction of oil volume-price data and a deep reinforcement learning algorithm.•The strategy is built by combining DRL-agents with optimal returns under varying market performance risk constraints.
Energy storage use right (ESUR) is a novel concept to make more people share the energy storage (ES) and give full play to its values. However, the integrated bidding, clearing and pricing method of ...ESURs is never reported nowadays. In this paper, we design some new ESURs, including the discrete power capacity right (DPCR) and the continuous energy capacity right (CECR), to meet the different needs of subscribers for sharing the ES. The ESUR trade is formulated as a bilevel problem, in which the upper-level problem represents the bidding strategies of subscribers, and the lower-level problem represents the market clearing. A pricing strategy is proposed to make the trading effective and fair by ensuring the optimalities of problems in both two levels. Using the Karush–Kuhn–Tucker (KKT) conditions and the binary expansion technique, the bilevel model is transferred into a linear problem to facilitate solving the model. In addition, it can be theoretically proved that the profit of each participant is non-negative. Numerical experiments are provided to demonstrate the effectiveness of the proposed method. Compared with the trading of DPCR, the participants’ overall profit is increased by 23.6%–145.91% owing to the additional consideration of the trading of CECR.
•Energy storage use rights designed to share the energy storage.•An integrated bidding, clearing and pricing method of energy storage use rights.•Conversion and linearization of the bilevel model.•Proof of the non-negative profit of each participant.