Futures combination weight optimization and transaction strategy method based on electricity price prediction

Abstract

The invention relates to a futures combination weight optimization and transaction strategy method based on electricity price prediction, and the method comprises the following steps: S1, calculating the weights corresponding to two futures under the optimal condition, and obtaining a futures combination under the optimal weight; s2, taking the price data of the futures combination and other related data as the input of an electricity price prediction model, and outputting the predicted electricity price by the electricity price prediction model; s3, giving a required warehouse opening contract number and a warehouse opening strategy based on the predicted electricity price and correlation; s4, building a generalized autoregression condition heterovariance model to carry out volatility modeling, giving an Alpha yield rate, a Beta yield rate and an Omega yield rate of the futures combination, and obtaining a row right price based on the three yield rates; and S5, based on a BSM model, calculating a Delta value of each day, and adjusting a strategy and a final income. Compared with the prior art, the method has the advantages of improving the accuracy of electricity price data and volatility prediction and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of power data prediction, and in particular to a futures portfolio weight optimization and trading strategy method based on electricity price prediction. Background Art

[0002] In the electricity market, futures trading has become an important means of managing price risks and optimizing asset allocation. The forecast of electricity prices has a key impact on the weight optimization of futures portfolios and the formulation of trading strategies. Traditional futures portfolio optimization methods mostly rely on historical data and simple statistical models, which often find it difficult to provide accurate predictions and optimizations when faced with the complexity of electricity price fluctuations. The development of electricity price forecasting models is crucial to improving the accuracy of electricity price forecasts. Modern technology combines multiple data sources, such as historical electricity prices, weather information, load demand, and fuel prices, to improve forecasting results. Although technological advances have made electricity price forecasts more and more accurate, the electricity market still faces challenges, including the improvement of forecast accuracy and the impact of market volatility. These challenges have prompted the continuous optimization of electricity price forecasting models to adapt to the dynamic changes and uncertainties of the market. Summary of the invention

[0003] The purpose of the present invention is to improve the accuracy of the prediction of electricity price data and its volatility and to provide a futures portfolio weight optimization and trading strategy method based on electricity price prediction.

[0004] The purpose of the present invention can be achieved by the following technical solutions:

[0005] A futures portfolio weight optimization and trading strategy method based on electricity price forecasting, the method comprising the following steps:

[0006] S1. Obtain commodity futures data related to monthly transaction electricity prices, determine the correlation between the commodity futures data and the monthly transaction electricity prices, determine two futures commodities based on the correlation, construct a VaR model, and use the Monte Carlo simulation method to calculate the optimal VaR value of the commodity, obtain the weights corresponding to the two futures in the optimal case, and obtain the futures combination under the optimal weight;

[0007] S2, using the futures data of the futures combination and other relevant data as inputs to the electricity price prediction model, and the electricity price prediction model outputs the predicted electricity price;

[0008] S3. Provide the required number of opening contracts and opening strategy based on the predicted electricity price and correlation;

[0009] S4. Build a generalized autoregressive conditional heteroskedasticity model to model volatility, give the Alpha yield, Beta yield and Omega yield of the futures portfolio, and get the exercise price based on the three yields;

[0010] S5. Based on the BSM model, calculate the daily Delta value and adjust the strategy and final profit.

[0011] Furthermore, the specific steps of S1 are:

[0012] Calculate the correlation between commodity futures data and monthly trading electricity prices, determine two futures as a futures combination based on the correlation, use the Monte Carlo simulation method to generate random sample paths for the futures combination, simulate future price trends, calculate the corresponding loss value for each path, and calculate the VaR at a specific confidence level based on the loss value distribution of all paths to determine the optimal weight of the futures combination.

[0013] Furthermore, the electricity price prediction model is an LSTM model, in which the state update process of the memory unit is:

[0014] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0015] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0016] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0017]

[0018] h t =o t *tanh(C t )

[0019] where x t is the current input data, h t-1 is the hidden state of the previous moment, C t is the current memory cell state, f t 、i t , o t are the activation values ​​of the forget gate, input gate, and output gate, respectively, W f , W i , W o , W C and b f , b i , b o , bC are weight and bias parameters.

[0020] Furthermore, the required number of open contracts and the opening strategy are as follows:

[0021] Determine the relationship between the predicted electricity price and the electricity price of last month, set the opening strategy of the futures portfolio based on the correlation, and obtain the required number of opening contracts based on the correlation.

[0022] Furthermore, the specific steps of S4 are:

[0023] Build a generalized autoregressive conditional heteroskedasticity model GARCH to obtain the residual value and volatility of the futures portfolio, calculate the Alpha yield, Beta yield and Omega yield of the hedged commodity portfolio based on the residual value and volatility, and combine the three yields to obtain the exercise price K;

[0024] The generalized autoregressive conditional heteroskedasticity model GARCH is:

[0025]

[0026] Among them, σ t represents the volatility at time t, ∈ t-1 is the residual of the previous moment, α0, α1 and β1 are model parameters;

[0027] The exercise price K is specifically:

[0028]

[0029] Among them, α represents Alpha rate of return, β represents Beta rate of return, ω represents Omega rate of return, S represents price, P represents m is the market benchmark value, e rT represents the impact of the risk-free interest rate during the maturity period, and σ is the volatility.

[0030] Furthermore, the Alpha rate of return is:

[0031] α=r i -(β×r m +∈)

[0032] Among them, r i is the actual rate of return of the portfolio, r m is the market benchmark rate of return, β is the market risk coefficient, and ∈ represents the residual.

[0033] Furthermore, the Beta rate of return is:

[0034]

[0035] Among them, Cov represents covariance, VaR represents VaR model, r i is the actual rate of return of the portfolio, r m The market benchmark rate of return.

[0036] Furthermore, the Omega rate of return is:

[0037]

[0038] Among them, F(r) is the cumulative distribution function of the rate of return, and threshold is the predetermined rate of return.

[0039] Furthermore, the specific steps of S5 are:

[0040] The Delta value is calculated based on the strike price, and the Delta calculation is performed in the BSM model. The number of positions and the weight of the futures portfolio are adjusted according to the change in the Delta value.

[0041] Furthermore, the adjustment of the number of positions and the weight of the futures portfolio according to the change in the Delta value is specifically as follows:

[0042] When the Delta value decreases, the number of positions is increased by buying futures, which consumes cash; when the Delta value increases, the number of positions is reduced by selling futures, which earns cash;

[0043] The Delta value is:

[0044] Δ=N(d1)

[0045]

[0046] Where Δ represents the Delta value, N(·) is the cumulative distribution function of the standard normal distribution, S is the futures price, K is the strike price, r is the risk-free interest rate, σ is the volatility, and T is the expiration time.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The present invention combines the VaR model, LSTM model, GARCH model and BSM model, determines the weight of the commodity portfolio based on the VaR model, and gives a risk prediction within a certain confidence interval, then predicts the electricity price through the LSTM model, applies the GARCH model to model and analyze the volatility of various commodity futures portfolios, and finally adjusts the strategy through the BSM model. The combination of the above models can accurately predict the electricity price, and effectively utilize the market volatility characteristics in combination with the predicted volatility to achieve investment strategy optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a flow chart of the present invention;

[0050] Figure 2 It is the result of calculating VaR based on Monte Carlo simulation;

[0051] Figure 3 This is a comparison chart between the monthly transaction electricity price predicted by LSTM and the actual electricity price;

[0052] Figure 4 It is based on the electricity purchase income after futures operations. DETAILED DESCRIPTION

[0053] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0054] The present invention comprehensively applies the expertise of financial engineering and electrical engineering, processes, analyzes and predicts the monthly transaction electricity price, and uses the electricity price fluctuations in the electricity sales business to find the associated futures value. The method is applicable to risk management and investment strategy optimization in the electricity market, energy trading market and related financial markets. The present invention proposes a futures portfolio weight optimization and trading strategy method based on electricity price prediction. The flowchart of the method is as follows: Figure 1 As shown, the method comprises the following steps:

[0055] S1. Obtain commodity futures data related to monthly transaction electricity prices, determine the correlation between the commodity futures data and the monthly transaction electricity prices, determine two futures commodities based on the correlation, construct a VaR model, and use the Monte Carlo simulation method to calculate the optimal VaR value of the commodity, obtain the weights corresponding to the two futures in the optimal case, and obtain the futures combination under the optimal weight;

[0056] S2, using the futures data of the futures combination and other relevant data as inputs to the electricity price prediction model, and the electricity price prediction model outputs the predicted electricity price;

[0057] S3. Provide the required number of opening contracts and opening strategy based on the predicted electricity price and correlation;

[0058] S4. Build a generalized autoregressive conditional heteroskedasticity model to model volatility, give the Alpha yield, Beta yield and Omega yield of the futures portfolio, and get the exercise price based on the three yields;

[0059] S5. Based on the BSM model, calculate the daily Delta value and adjust the strategy and final profit.

[0060] In S1, commodity futures data related to monthly trading electricity prices are collected, the correlation between their prices and electricity prices is calculated, and based on this, a VaR model is constructed to determine the commodity portfolio ratio and give a confidence interval.

[0061] Commodity futures include, but are not limited to, thermal coal futures, coking coal futures, petroleum gas futures or other commodity futures that are closely correlated with monthly trading electricity prices.

[0062] Determining the combination of commodity futures includes the following steps: selecting a variety of commodity futures that are closely correlated with the monthly trading electricity price, such as thermal coal futures, coking coal futures and petroleum gas futures; calculating the correlation coefficient between each commodity futures price and the monthly trading electricity price; based on the correlation coefficient, constructing a VaR model for a multi-commodity portfolio to quantify the risk of purchasing futures; determining the weight of the commodity portfolio based on the VaR model, and providing a risk forecast within a certain confidence interval.

[0063] The specific steps of S1 are:

[0064] Collect commodity futures data related to monthly trading electricity prices, calculate the correlation between their prices and monthly trading electricity prices, and build a VaR model based on this. The VaR model is used to quantify futures risks and maximize returns by optimizing the weights of commodity portfolios.

[0065] VaR=μ-zσ (1)

[0066] Where μ is the mean of commodity futures prices, σ is the standard deviation, and z is the quantile of the standard normal distribution at a given confidence level.

[0067] The formula for calculating the mean μ and standard deviation σ is:

[0068]

[0069] Then use the Monte Carlo simulation method to generate a large number of random sample paths, simulate future price trends, calculate the corresponding loss value for each path, and calculate the VaR at a specific confidence level based on the loss value distribution of all paths. Use the following formula to generate future price paths:

[0070]

[0071] Among them, S t is the commodity price at time t, S0 is the initial price, Δt is the time interval, and Z is a random variable with standard normal distribution.

[0072] Simulate each path, calculate the corresponding loss value L, sort the loss values ​​of all simulated paths, and find the VaR at a specific confidence level. For example, at a 95% confidence level, VaR is the 95th percentile loss value.

[0073] In S2, the long short-term memory (LSTM) model is used to predict future electricity prices. The input data includes futures data with high correlation with monthly transaction electricity prices, bilateral negotiation data, and load and other data closely related to electricity prices.

[0074] The use of the LSTM model includes the following: screening and processing the input data to ensure its accuracy and relevance, setting model parameters including learning rate, number of hidden layers and number of nodes, etc.

[0075] The specific steps of S2 are:

[0076] The long short-term memory (LSTM) model is used to predict future electricity prices. The input data includes futures data that are highly correlated with monthly transaction electricity prices, bilateral negotiation data, and load and other data that are closely related to electricity prices.

[0077] The LSTM model predicts time series data through memory cells and three gating mechanisms (input gate, forget gate, output gate). The state update formula of the memory cell is:

[0078] f t =σ(W f ·[h t-1 ,x t ]+b f ) (5)

[0079] i t =σ(W i ·[h t-1 ,x t ]+b i ) (6)

[0080] o t =σ(W o ·[h t-1 ,x t ]+b o ) (7)

[0081]

[0082] h t =o t *tanh(C t ) (10)

[0083] where x t is the current input data, h t-1 is the hidden state of the previous moment, C t is the current memory cell state, f t 、i t , o tare the activation values ​​of the forget gate, input gate, and output gate, respectively, W f , W i , W o , W C and b f , b i , b o , b C are weight and bias parameters.

[0084] In S3, specific operation strategies are given according to the results of electricity price forecast, including determining the number of futures opening positions and trading strategies.

[0085] In S4, a generalized autoregressive conditional heteroskedasticity (GARCH) model is built to model volatility and give the Alpha yield, Beta yield and Omega yield of the selected products.

[0086] The construction of the GARCH model includes the following: selecting an appropriate GARCH model form; estimating model parameters using historical data; predicting future volatility; and applying the GARCH model to model and analyze the volatility of various commodity futures portfolios.

[0087] The specific steps of S4 are:

[0088] The GARCH model is used to model the volatility of futures prices. Select an appropriate GARCH model form, use historical data to estimate the model parameters, and predict future volatility. The basic formula of the GARCH model is:

[0089]

[0090] where ∈ t-1 is the residual of the previous moment, and α0, α1 and β1 are model parameters.

[0091] The calculated residual value and volatility are used to further calculate the Alpha return, Beta return and Omega return.

[0092] In S5, according to the volatility analysis results of the GARCH model, delta calculation is performed in the BSM (Black-Scholes-Merton) model to provide a dynamic adjustment strategy and final return. The delta operation in the BSM model includes the following: according to the volatility prediction results of the GARCH model, the delta value of the futures is calculated; according to the calculated delta value, the position of the futures contract is adjusted to maximize the return.

[0093] The specific steps of S5 are:

[0094] According to the volatility prediction results of the GARCH model, the delta value of the futures is calculated, and the strike price is set and the position of the commodity futures contract is adjusted according to the calculated delta value. The BSM model is used to formulate a detailed operation strategy to maximize profits. The Delta calculation formula in the BSM model is:

[0095] Δ=N(d1) (12)

[0096]

[0097] Where N(·) is the cumulative distribution function of the standard normal distribution, S is the futures price, K is the strike price, r is the risk-free interest rate, σ is the volatility, and T is the expiration time. The bullish futures price formula is:

[0098] C=SN(d1)-Ke -rT N(d2) (14)

[0099]

[0100] Futures trading strategies also include regular calibration and updating of VaR models, GARCH models and BSM models to ensure the accuracy and effectiveness of the models, and adjust model parameters according to market changes.

[0101] In view of the increasingly complex and volatile electricity market, and in view of the deficiencies of traditional fixed-price contracts and long-term and short-term electricity purchase and sales contracts in risk management, the present invention provides a futures portfolio weight optimization and trading strategy method based on electricity price forecasting by combining VaR model, LSTM model, GARCH model and BSM model. This method can effectively utilize the characteristics of market fluctuations, accurately hedge and manage electricity price risks, significantly improve the stability and security of electricity sales business, provide more flexible and efficient risk management tools, and help improve the market competitiveness and economic benefits of enterprises.

[0102] The following is an example with the help of a flowchart:

[0103] Data collection and processing: First, collect monthly transaction electricity price market data and futures commodity data. In this example, taking the electricity market data of a certain province as an example, the monthly centralized competitive transaction price in the long-term transaction is used as the monthly transaction price in this example; and obtain other relevant data to prepare for electricity price forecasting, including bilateral negotiated transaction prices, transaction volume, and market competition indexes on the power generation side and the power consumption side; after screening and processing, these data will be used to build a forecasting model to ensure the accuracy and relevance of the data, thereby improving the accuracy and reliability of electricity price forecasting.

[0104] By collecting the correlation between futures prices and monthly trading electricity prices, coking coal futures and liquefied petroleum gas futures were selected as the commodity combination after calculation, and their correlation with monthly trading electricity prices reached 0.79 and 0.64 respectively.

[0105] Risk interval calculation: After determining the types of commodities, its ratio needs to be calculated. Here, the Monte Carlo method is used to calculate the VaR value of the commodity portfolio. First, the ratio of coking coal and liquefied petroleum gas futures needs to be determined. Here, the number of trading days of the two in the next month is simulated, and 1,000 simulations are performed to calculate the daily average return and standard deviation of the commodity portfolio. The normal distribution is used to generate random numbers that conform to the distribution of the commodity portfolio return. Finally, the cumulative return of each simulation sequence is calculated. The simulation results are as follows: Figure 2 As shown. Through simulation and calculation, it is found that when the weight of coking coal is 0.58, the variance of the comprehensive yield of the two is the smallest, and when the weight of coking coal is 0.60, the VaR value calculated based on the Monte Carlo method is optimal. The following are the results under different combinations, including the hedging effect when only coking coal futures are used and the coking coal weight is 0.58 and 0.60 respectively.

[0106] Table 1 VaR values ​​under different weight combinations

[0107]

[0108] From the analysis results in Table 1, it can be seen that different weight combinations have a significant impact on the VaR value. The VaR value of using coking coal futures alone is the highest, indicating that the risk exposure is large; by optimizing the weight combination of the yield variance, the VaR value is significantly reduced, indicating that the risk management effect is significantly improved; the weight combination based on the Monte Carlo simulation method (0.6:0.4) further optimizes the risk management effect, and the VaR value under different confidence intervals is the lowest, showing the effectiveness of risk assessment and optimization strategies through a large number of random sample path simulations. Therefore, by optimizing the weight combination, especially based on the Monte Carlo simulation method, the VaR value can be significantly reduced, the risk management effect of purchasing futures products can be improved, and the stability and security of enterprises in market price fluctuations can be enhanced.

[0109] Monthly transaction electricity price forecast: This paper uses a multi-input single-output LSTM model to forecast monthly transaction electricity prices. Since the selected futures commodity has a high correlation with the monthly transaction electricity price, it is used as part of the multi-dimensional input. In addition, the bilateral negotiated electricity price, monthly transaction electricity volume, and market competition coefficients on the power generation side and the demand side are also selected as inputs.

[0110] The model training uses the data set of Guangdong power market from February 2017 to May 2023, and predicts the monthly transaction electricity prices from June 2023 to December 2023. The prediction results are shown in Table 2. By comprehensively considering these key factors, the model can more accurately reflect the changing trend of future electricity prices and provide reliable guidance for further operation strategies.

[0111] Table 2 Electricity price forecast results

[0112]

[0113] The specific trend chart is as follows Figure 3 As shown. The analysis results show that the electricity price forecasting model has a small prediction error, showing high prediction accuracy and model stability, and can effectively capture the overall trend of market electricity prices. Even in months with large fluctuations in electricity prices, the model can provide reasonable prediction values. This provides valuable decision support for the next step of specific purchasing strategies. Based on the predicted electricity price and contracted electricity purchase volume, the expected expenditure amount can be further calculated, thereby formulating a more accurate purchasing strategy and improving the profitability of power sales companies.

[0114] Determine the opening strategy: According to the correlation calculation, there is a high correlation between futures commodities and monthly trading electricity prices. Combined with market data, this correlation is specifically manifested as follows: when the predicted electricity price is lower than the electricity price of the previous month, the price of coking coal futures will usually rise, while the price of liquefied petroleum gas futures will fall, and the specific fluctuation range is related to the correlation. Based on this, formulate the opening strategy as shown in Table 3 and give the futures price.

[0115] Table 3 Opening strategy operation and futures price table

[0116]

[0117]

[0118] It can be seen that the operations made in the futures market based on the predicted electricity prices are basically in line with the market trend. Specifically, when the electricity price is predicted to fall, coking coal futures usually go long to take advantage of the rising price trend to profit, while liquefied petroleum gas futures usually go short to take advantage of the falling price trend to profit.

[0119] In addition, the operations performed in the futures market based on the predicted electricity prices are highly consistent with the actual market trends, which further verifies the effectiveness of the forecasting model and trading strategy.

[0120] Determine the number of open contracts required: By calculating the correlation coefficient between the commodity portfolio and the monthly trading electricity price, it can be obtained that it is 80%. This means that when the power purchase contract loses 1%, the corresponding commodity portfolio will lose 0.8%. Assuming the contract value is 1 million yuan, you can buy a commodity portfolio of 1.25 million to hedge. According to the data of the futures exchange, only 15% of the margin is required for trading, which actually only requires 187,500.

[0121] It can be found that this trading strategy can make efficient use of funds and only require a small investment to hedge a large amount of risk; it can flexibly adjust the number of positions and hedging ratios, respond quickly to market changes, and has little impact on the company's cash flow.

[0122] On this basis, the expected contract cost, the required futures amount and the actual cost after margin calculation are given according to the electricity price forecast results. The results are shown in Table 4.

[0123] Table 4 Amount used for risk hedging

[0124]

[0125]

[0126] Determine the strike price: Here, the GARCH model is used to model the volatility of futures prices, and the relevant data of coking coal futures and petroleum gas futures are calculated based on the residual probability distribution diagram and the probability-theoretical diagram, including Alpha yield, Beta yield and Omega yield. The Alpha yield represents the excess return of the investment portfolio relative to the market benchmark. If it is positive, it means that the performance of the asset is better than the expected return, that is, the investor can obtain excess returns; if it is negative, it means that the performance of the asset is not as expected. The specific calculation formula is:

[0127] α=r i -(β×r m +∈) (16)

[0128] where r i is the actual rate of return of the portfolio, r m is the market benchmark rate of return, and β is the market risk coefficient.

[0129] Beta returns represent the systematic risk of a portfolio relative to the market benchmark. The specific calculation formula is:

[0130]

[0131] Omega rate of return: represents the risk-adjusted rate of return of an investment portfolio at different return levels. The specific calculation formula is as follows:

[0132]

[0133] Where F(r) is the cumulative distribution function of the rate of return, and threshold is the expected rate of return.

[0134] The calculation results are given here.

[0135] Table 5 GARCH model calculation results

[0136]

[0137] By analyzing the results in Table 5, it can be found that both coking coal futures and liquefied petroleum gas futures show significant advantages in risk management and hedging. The Alpha yield of coking coal futures is 0.0501, providing a higher excess return, and the Beta yield is 0.9468, showing a high market correlation, which is helpful for market risk prediction. The Omega yield of liquefied petroleum gas futures is 0.0517, which is significantly higher than that of coking coal futures, providing a higher risk-adjusted return, and the Beta yield is 0.9485, which also shows a high market correlation. The combined use of these two futures can optimize the purchase strategy, obtain excess returns through coking coal futures, and enhance risk management through liquefied petroleum gas futures, thereby improving the stability and security of power sales companies in market fluctuations.

[0138] By combining these three yields, we can get the calculation formula for the exercise price K:

[0139]

[0140] Where S(1+α) reflects the potential excess return of the futures contract, P m is the market benchmark value, e rT Represents the impact of the risk-free interest rate during the maturity period.

[0141] Through this detailed calculation method, combined with Alpha, Beta, Omega yields and other market factors, the exercise price can be set more accurately, thereby optimizing the purchase strategy and improving the effectiveness of risk management.

[0142] Dynamically adjust daily specific operations: After determining the number of open contracts and the strike price, the number of positions and the proportion of commodity combinations need to be dynamically adjusted every day according to price trends. This process uses the Black-Scholes-Merton (BSM) model to adjust the strategy by calculating the daily Delta value. First, initialize the Delta value and cash account according to the option parameters, time series, and account settings. If the Delta value is less than 0.5, increase the number of positions by buying options; if the Delta value is greater than 0.5, keep the position unchanged and record the daily cash account balance and number of positions.

[0143] During the daily adjustment process, the system recalculates the Delta value based on the new price and expiration date, and determines the change in the Delta value. When the Delta value decreases, the number of positions is increased by buying futures, consuming cash; when the Delta value increases, the number of positions is reduced by selling futures, and cash is obtained. In this way, the number of positions is ensured to adapt to market conditions and dynamic trading strategy formulation is achieved.

[0144] At maturity, settlement is made based on the final futures price and strike price, the final cash account balance is calculated, and the total income and total rate of return are determined. This method has the characteristics of real-time adjustment, risk management, capital utilization and flexibility. It optimizes the use of cash accounts through daily adjustment strategies, maximizes capital utilization efficiency, adapts to risk management needs under different market conditions, and ensures the stability and security of the investment portfolio.

[0145] According to the method proposed in the present invention, the benefits of different purchase strategies are calculated as shown in Table 6.

[0146] Table 6 Comparison of benefits of different purchasing strategies

[0147] 2023.9 ​​179571.88 354380.69 262989.77 317824.32

[0149] 2023.10 -23099.04 38293.45 731.99 23268.87 2023.11 16773.16 136421.84 37674.91 96923.07 2023.12 41744.40 10286.78 97087.88 45007.22

[0152] Its specific yield fluctuations are as follows Figure 4As shown, it can be found that the use of a combination investment strategy disperses investment risks and improves the stability of returns. By allocating funds among different assets, the volatility and extreme risks of a single asset can be reduced, thereby maintaining a more stable return in market uncertainty. The data shows that combined products can provide relatively stable returns in most months, significantly reducing the risk of large losses that may be caused by a single asset.

[0153] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A futures portfolio weight optimization and trading strategy method based on electricity price forecast, characterized in that: The method comprises the following steps: S1. Obtain commodity futures data related to monthly transaction electricity prices, determine the correlation between the commodity futures data and the monthly transaction electricity prices, determine two futures commodities based on the correlation, construct a VaR model, and use the Monte Carlo simulation method to calculate the optimal VaR value of the commodity, obtain the weights corresponding to the two futures in the optimal case, and obtain the futures combination under the optimal weight; S2, using the futures data of the futures combination and other relevant data as inputs to the electricity price prediction model, and the electricity price prediction model outputs the predicted electricity price; S3. Provide the required number of opening contracts and opening strategy based on the predicted electricity price and correlation; S4. Build a generalized autoregressive conditional heteroskedasticity model to model volatility, give the Alpha yield, Beta yield and Omega yield of the futures portfolio, and get the exercise price based on the three yields; S5. Based on the BSM model, calculate the daily Delta value and adjust the strategy and final profit.

2. A futures portfolio weight optimization and trading strategy method based on electricity price forecast according to claim 1, characterized in that: The specific steps of S1 are: Calculate the correlation between commodity futures data and monthly trading electricity prices, determine two futures as a futures combination based on the correlation, use the Monte Carlo simulation method to generate random sample paths for the futures combination, simulate future price trends, calculate the corresponding loss value for each path, and calculate the VaR at a specific confidence level based on the loss value distribution of all paths to determine the optimal weight of the futures combination.

3. The method for optimizing futures portfolio weights and trading strategies based on electricity price forecasting according to claim 2, characterized in that: The electricity price prediction model is an LSTM model, in which the state update process of the memory unit is: f t =σ(W f ·[h t-1 ,x t ]+b f ) i t =σ(W i ·[h t-1 ,x t ]+b i ) the t =σ(W o ·[h t-1 ,x t ]+b o ) h t =o t *tanh(C t ) where x t is the current input data, h t-1 is the hidden state of the previous moment, C t is the current memory cell state, f t 、i t , o t are the activation values ​​of the forget gate, input gate, and output gate, respectively, W f , W i , W o , W C and b f , b i , b o , b C are weight and bias parameters.

4. The method for optimizing futures portfolio weights and trading strategies based on electricity price forecasting according to claim 1, characterized in that: The required number of open contracts and opening strategies are as follows: Determine the relationship between the predicted electricity price and the electricity price of the previous month, set the opening strategy of the futures portfolio based on the correlation, and obtain the required number of opening contracts based on the correlation.

5. The method for futures portfolio weight optimization and trading strategy based on electricity price forecast according to claim 1, characterized in that: The specific steps of S4 are: Build a generalized autoregressive conditional heteroskedasticity model GARCH to obtain the residual value and volatility of the futures portfolio, calculate the Alpha yield, Beta yield and Omega yield of the hedged commodity portfolio based on the residual value and volatility, and combine the three yields to obtain the exercise price K; The generalized autoregressive conditional heteroskedasticity model GARCH is: Among them, σ t represents the volatility at time t, ∈ t-1 is the residual of the previous moment, α0, α1 and β1 are model parameters; The exercise price K is specifically: Among them, α represents Alpha rate of return, β represents Beta rate of return, ω represents Omega rate of return, S represents price, P represents m is the market benchmark value, e rT represents the impact of the risk-free interest rate during the maturity period, and σ is the volatility.

6. A futures portfolio weight optimization and trading strategy method based on electricity price forecasting according to claim 5, characterized in that: The Alpha rate of return is: α＝r i -(β×r m +∈) Among them, r i is the actual rate of return of the portfolio, r m is the market benchmark rate of return, β is the market risk coefficient, and ∈ represents the residual.

7. A futures portfolio weight optimization and trading strategy method based on electricity price forecasting according to claim 5, characterized in that: The Beta rate of return is: Among them, Cov represents covariance, VaR represents VaR model, r i is the actual rate of return of the portfolio, r m The market benchmark rate of return.

8. The method for futures portfolio weight optimization and trading strategy based on electricity price forecast according to claim 5, characterized in that: The Omega rate of return is: Among them, F(r) is the cumulative distribution function of the rate of return, and threshold is the predetermined rate of return.

9. The method for futures portfolio weight optimization and trading strategy based on electricity price forecast according to claim 1, characterized in that: The specific steps of S5 are: The Delta value is calculated based on the strike price, and the Delta calculation is performed in the BSM model. The number of positions and the weight of the futures portfolio are adjusted according to the change in the Delta value.

10. A futures portfolio weight optimization and trading strategy method based on electricity price forecasting according to claim 9, characterized in that: The specific method of adjusting the number of positions and the weight of the futures portfolio according to the change in Delta value is: When the Delta value decreases, the number of positions is increased by buying futures, which consumes cash; when the Delta value increases, the number of positions is reduced by selling futures, which earns cash; The Delta value is: Δ=N(d1) Where Δ represents the Delta value, N(·) is the cumulative distribution function of the standard normal distribution, S is the futures price, K is the strike price, r is the risk-free interest rate, σ is the volatility, and T is the expiration time.

Images