High-frequency transaction strategy optimization method, system and device, medium and product

Abstract

The invention discloses a high-frequency transaction strategy optimization method, system and device, a medium and a product, and relates to the technical field of high-frequency transaction management. The method comprises the steps of obtaining market microstructure data and market transaction data, inputting the market microstructure data and the market transaction data into a market change trend prediction model to obtain a market change trend prediction result, and calculating the market change trend according to the market change trend prediction result. High-frequency transaction strategy parameters are adjusted through a strategy parameter adjustment algorithm, and the adjusted high-frequency transaction strategy parameters are sent to a user transaction platform for a user to refer to and execute transaction operation; and an execution result of user transaction operation is monitored in real time, and the market change trend prediction model and the strategy parameter adjustment algorithm are updated through an online learning mechanism, so that the key technical problems of slow response, inaccurate optimization parameters and the like of the current traditional high-frequency transaction strategy optimization method when facing market change are solved, and the user experience is improved. Therefore, the high-frequency transaction effect of the user is obviously improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of high-frequency transaction management, and particularly relates to a high-frequency transaction strategy optimization method, system, device, medium and product. BACKGROUND

[0002] High-frequency trading is a trading strategy that uses advanced technology and complex algorithms to perform a large number of transactions in a very short time. Its main feature is through high-speed data processing and ultra-low latency transaction execution, which can complete a large number of transactions in milliseconds or even microseconds. This strategy relies on the automatic execution of algorithms and programs to generate trading signals by analyzing market data in real time, quickly capturing small price differences in the market, and achieving profits.

[0003] In recent years, high-frequency trading has become increasingly prominent in the financial market. With its high-speed computing power and complex algorithmic models, it has become an important part of the modern financial system. According to relevant research, high-frequency trading not only significantly improves market liquidity, but also improves market efficiency by optimizing the price discovery mechanism. Common high-frequency trading strategy types include arbitrage strategies, trend-following strategies, and market-making strategies. Arbitrage strategies aim to profit from price differences between different markets or assets without risk, such as foreign exchange trading strategies based on statistical arbitrage models that have been widely used in global financial markets. Trend-following strategies analyze technical indicators (such as moving averages, MACD, etc.) to predict price changes and develop buy / sell rules accordingly.

[0004] However, the optimization process of the above two high-frequency trading strategies has the following defects: Existing trading strategy optimization mostly focuses on static parameter adjustment, lacking dynamic adaptation to market real-time changes. For example, although the traditional genetic algorithm can optimize the trading strategy to a certain extent, its convergence speed is slow and it is easy to fall into local optimal solution, which is difficult to meet the real-time and accuracy requirements of high-frequency trading. Secondly, the shortcomings of algorithm design itself are also important reasons for the problem. The optimization algorithm design of many current high-frequency trading strategies lacks global optimization ability, especially in multi-objective optimization. SUMMARY

[0005] The purpose of the present application is to provide a high-frequency trading strategy optimization method, system, computer device, computer readable storage medium and computer program product, to solve the problem of low dynamic adaptation to market real-time changes and lack of global optimization ability in the existing optimization method of high-frequency trading strategy.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: In a first aspect, a method for optimizing a high-frequency trading strategy is provided, comprising: obtaining market microstructure data and market transaction data in real time, the market microstructure data including order book data, trade-by-trade data, and cancellation data, and the market transaction data including transaction price, transaction time, and transaction frequency; inputting the market microstructure data and the market transaction data into a market change trend prediction model to obtain market change trend prediction results, the market change trend prediction results including market volatility, liquidity indicators, and trend direction, and the market change trend prediction model being a recurrent neural network model trained based on historical transaction data; adjusting high-frequency trading strategy parameters, including transaction frequency, transaction volume, transaction price, and transaction timing, according to the market change trend prediction results through a strategy parameter adjustment algorithm; sending the adjusted high-frequency trading strategy parameters to an electronic trading platform through a low-latency communication protocol to perform corresponding trading operations; monitoring the execution results of the trading operations in real time and updating the market change trend prediction model and the strategy parameter adjustment algorithm through an online learning mechanism.

[0007] Based on the above invention, a new solution for optimizing a high-frequency trading strategy is provided, that is, by integrating market microstructure data (order book, trade-by-trade, and cancellation data) and market transaction data (price, time, and frequency), combined with a recurrent neural network model trained based on historical transaction data, market volatility, liquidity indicators, and trend direction can be comprehensively predicted, overcoming the prediction lag and deviation problems caused by traditional methods relying on a single data source or simple models. The fusion of multi-dimensional data and the time series processing capability of deep learning models significantly improves the accuracy and foresight of trend prediction, providing a reliable basis for strategy optimization. At the same time, based on the market change trend prediction results, the transaction frequency, transaction volume, transaction price, and transaction timing are adjusted in real time through a strategy parameter adjustment algorithm. This mechanism can dynamically adjust the trading strategy according to market volatility, liquidity changes, and trend direction, compared to static strategies or manual intervention adjustments, this method has millisecond-level response capability, significantly improving the real-time adaptability and trading efficiency of the strategy.

[0008] In one possible design, the market change trend prediction model is a neural network model based on deep learning, specifically selected from one or a combination of the following models: recurrent neural network, long short-term memory network, gated recurrent unit, and Transformer model.

[0009] In one possible design, the training method of the market change trend prediction model includes the following steps: The historical market microstructure data and historical market transaction data are preprocessed, including data cleaning, missing value filling, normalization processing and feature extraction; The preprocessed data is divided into a training set and a validation set in a ratio of 7:3; An optimization algorithm based on adaptive matrix estimation is used to train the initialized market trend prediction model, wherein the mean squared error between the predicted value and the actual value is selected as the optimization objective. The trained market trend prediction model is prevented from overfitting by the validation set.

[0010] In one possible design, the adjustment of high-frequency trading strategy parameters according to the market trend through a strategy parameter adjustment algorithm includes: When the market volatility prediction value output by the market trend prediction model exceeds the dynamic threshold, the trading frequency or trading volume is reduced; When liquidity anomalies are detected and the market trend prediction model predicts that the price will fall, the trading price is adjusted to maintain liquidity; When the model predicts a market trend reversal, the trading direction is adjusted in advance or the current transaction is terminated.

[0011] In one possible design, the dynamic threshold σ(t) is determined by the following formula: σ(t)=α·σ_{historical}+(1-α)·σ_{recent}; Wherein α is a decay factor, and α∈[0.6,0.8], σ_{historical} is the median of the past 30-day volatility, and σ_{recent} is the exponential average number index of the recent 5-minute volatility.

[0012] In one possible design, the historical transaction data further includes external influencing factor data, which includes at least one of macroeconomic indicators, policy change information, and market sentiment index.

[0013] The second aspect provides an optimization system for high-frequency trading strategy, comprising: A data collection module for real-time acquisition of market microstructure data and market transaction data; A market trend prediction module for inputting the market microstructure data and the market transaction data into a market trend prediction model to obtain a market trend prediction result, A strategy adjustment module for adjusting high-frequency trading strategy parameters through a strategy parameter adjustment algorithm according to the market trend prediction result; An automation execution module is configured to send the adjusted high-frequency trading strategy parameters to an electronic trading platform to perform corresponding trading operations through a low-latency communication protocol. A monitoring and feedback module is configured to monitor the execution results of the trading operations in real time and update the market trend prediction model and the strategy parameter adjustment algorithm through an online learning mechanism.

[0014] In a third aspect, the present application provides a computer device comprising a storage module, a processing module and a transceiver module connected in sequence, wherein the storage module is configured to store a computer program, the transceiver module is configured to transceive messages, and the processing module is configured to read the computer program and execute the optimization method of the high-frequency trading strategy as described in the first aspect or any possible design of the first aspect.

[0015] In a fourth aspect, the present application provides a computer readable storage medium having instructions stored thereon, wherein the instructions, when executed on a computer, perform the optimization method of the high-frequency trading strategy as described in the first aspect or any possible design of the first aspect.

[0016] In a fifth aspect, the present application provides a computer program product comprising a computer program or instructions, wherein the computer program or the instructions, when executed on a computer, implement the optimization method of the high-frequency trading strategy as described in the first aspect or any possible design of the first aspect.

[0017] The above-mentioned scheme has the following beneficial effects: The present application creatively proposes an optimization method of a high-frequency trading strategy, which has the following beneficial effects: 1. By collecting multi-dimensional market microstructure data (order book, transaction by transaction, and order cancellation data) and market transaction data (price, time, and frequency) in real time, and combining a market trend prediction model trained based on historical data, the trend prediction is performed, which overcomes the limitations of traditional methods relying on a single data source or static rules. The market trend prediction model can comprehensively output key prediction results such as market volatility, liquidity indicators, and trend direction, significantly improving the ability to capture complex market changes and providing high-credibility decision-making basis for strategy adjustment.

[0018] 2. Based on the predicted market trend, the trading frequency, trading volume, trading price, and trading timing are optimized in real time. When the volatility is above a threshold, the risk exposure is reduced, when the trend reverses, the trading direction is quickly adjusted, or when the liquidity is abnormal, the liquidity is actively maintained. This breaks the passivity of traditional strategies, enabling the system to respond to market mutations within milliseconds, maximizing trading opportunities and effectively controlling risks, and significantly improving the flexibility and profitability of the strategy.

[0019] 3. By studying users’ operations and results on the trading platform, we ensure the effectiveness and stability of the high-frequency trading strategies subsequently provided to users before they are actually deployed.

[0020] 4. By monitoring trade execution results (such as profit and loss, latency, and risk indicators) in real time and incorporating online learning mechanisms to provide feedback and optimize trend prediction models and strategy adjustment algorithms, the system continuously learns from users' actual trading data, dynamically corrects model deviations, and adapts to long-term market evolution or unexpected events. It also provides personalized optimization based on users' trading habits, ensuring the long-term effectiveness and robustness of strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 A flowchart of a method for optimizing a high-frequency trading strategy according to an embodiment of the present application.

[0023] Figure 2 A schematic diagram of the structure of the optimization system for high-frequency trading strategies provided in an embodiment of the present application.

[0024] Figure 3 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention will be briefly introduced below in conjunction with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structures of the drawings is only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these embodiments without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0026] It should be understood that although the terms first, second, etc. may be used herein to describe various objects, these objects should not be limited by these terms. These terms are merely used to distinguish one object from another. For example, a first object can be referred to as a second object, and similarly, a second object can be referred to as a first object without departing from the scope of the exemplary embodiments of the present invention.

[0027] It should be understood that, for the term "and / or" which can appear in the present text, it is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent three cases of A alone, B alone or A and B existing at the same time; for example, A, B and / or C can represent any one of A, B and C or any combination thereof; for the term " / and" which can appear in the present text, it is another description of the relationship of another associated object, which means that there can be two relationships, for example, A / and B can represent two cases of A alone or A and B existing at the same time; in addition, for the character " / " which can appear in the present text, it generally represents an "or" relationship between the associated objects before and after it.

[0028] As Figure 1 shown, the high-frequency trading strategy optimization method provided by the first aspect of the present embodiment can be executed by a computer device with certain computing resources, such as a server, a personal computer (PC, which refers to a multi-purpose computer with size, price and performance suitable for personal use; desktop, notebook computer to small notebook computer and tablet computer and ultrabook, etc.), a smart phone, a personal digital assistant (PDA) or a wearable device, etc. Figure 1 As shown in the figure, the high-frequency trading strategy optimization method includes but is not limited to the following steps S1-S6.

[0029] S1. Real-time acquisition of market microstructure data and market transaction data, the market microstructure data including order book data, transaction-by-transaction data and cancellation data, the market transaction data including transaction price, transaction time and transaction frequency.

[0030] In the step S1, the sources of market microstructure data and market transaction data mainly include real-time transaction data provided by the exchange, historical data from financial data service providers, and unstructured information supplemented by third-party data platforms. Specifically, the exchange data covers core indicators such as price fluctuations, trading volume changes, and order book depth, which can reflect the instantaneous state and trend of the market. For example, order book data includes bid and ask prices, order quantities, and order times, and each transaction data includes transaction price, transaction volume, and transaction time; cancellation data includes the time, price, and quantity of canceled orders. Financial data service providers provide more comprehensive historical data, such as transaction prices, specific transaction times, and transaction frequencies, including price sequences at different time scales, technical indicator calculation results, and macroeconomic variables, providing sufficient sample support for strategy model training. In addition, as a preferred solution, unstructured data such as news data and social media sentiment can also be included in the collection range to enhance the sensitivity of the strategy to external environmental changes. The integration of the above multi-source data not only improves the comprehensiveness of the data, but also lays a solid foundation for the subsequent construction of the strategy model.

[0031] S2. Input the market microstructure data and the market transaction data into a market change trend prediction model to obtain a market change trend prediction result; In the step S2, the market change trend prediction result includes market volatility, liquidity indicators, and trend direction, and the market change trend prediction model is a recurrent neural network model trained based on historical transaction data.

[0032] In order to further optimize the data analysis results, the market trend prediction model of the present application is a neural network model based on deep learning, specifically selected from one or a combination of the following models: recurrent neural network, long short-term memory network, gated recurrent unit and Transformer model, for processing complex patterns in time series data. Given the characteristics of high-frequency trading, such as large data volume, high time sensitivity and complex and variable market environment, the present application selects a neural network model based on deep learning as the core algorithm model. The deep learning model has a strong non-linear fitting ability and processing ability for high-dimensional data, and has shown significant advantages in prediction and decision-making problems in the financial field. Preferably, the long short-term memory network is widely used in trend prediction and transaction signal generation in the financial market because it can effectively capture long-term dependencies in time series data. In addition, the long short-term memory network solves the gradient vanishing and gradient explosion problems in traditional recurrent neural networks through a gating mechanism, making it more suitable for processing time series data in high-frequency trading. Combined with the high requirements of high-frequency trading for real-time and accuracy, the long short-term memory network model can complete complex calculations in a short time and provide reliable decision support, providing a solid technical foundation for the optimization of subsequent high-frequency trading strategies.

[0033] The market trend prediction model used in the present application takes the long short-term memory network as the core architecture, and its structure mainly includes the input layer, the hidden layer and the output layer. The input layer is responsible for receiving multi-dimensional data related to high-frequency trading, such as price fluctuations, trading volume, technical indicators, etc. These data are normalized to improve the convergence speed and learning efficiency of the model. The hidden layer is composed of multiple long short-term memory network units, each containing an input gate, a forget gate and an output gate, which are used to control the flow and update of information. Among them, the input gate decides whether the information at the current time enters the cell state, the forget gate controls the retention degree of historical information, and the output gate decides how the cell state affects the output result. The parameter settings of the hidden layer include the number of units, the type of activation function and the weight initialization method, which are all optimized through multiple experiments to determine the best configuration. The output layer generates trading signals such as buy, sell or hold based on the calculation results of the hidden layer, and uses the Sigmoid activation function to map the output value to a probability distribution between 0 and 1, indicating the possibility of different trading operations. The information is transmitted between the layers through full connection to ensure that the model can fully utilize the potential patterns in the input data.

[0034] Preferably, the training steps of the market trend prediction model of the present application include the following steps S201-S204, but are not limited to the following steps.

[0035] S201, pre-process the historical market microstructure data and historical market transaction data, including data cleaning, missing value filling, normalization and feature extraction.

[0036] In step S201, there are often noise data, missing values and outliers in market microstructure data and market transaction data, which will seriously affect the accuracy and effectiveness of subsequent data analysis. First, obvious noise data is removed by a filter based on statistical rules, such as price fluctuations beyond a reasonable range or order records that do not conform to trading logic. Second, for missing value problems, the system uses a combination of linear interpolation and K-nearest neighbor algorithm to fill in. For missing values in time series data, linear interpolation can estimate the data trend at the previous and next time points; while for missing values in multi-dimensional feature data, K-nearest neighbor algorithm fills in by calculating the average value of similar samples, thereby maximizing the preservation of the original features of the data.

[0037] S202, the preprocessed data is divided into training set and validation set according to the proportion of 7:3.

[0038] S203, the adaptive moment estimation-based optimization algorithm is used to train the initialized market trend prediction model, wherein the mean square error of the predicted value and the actual value is selected as the optimization target.

[0039] In step S203, adaptive moment estimation refers to a class of algorithms that can dynamically adjust the optimization parameters, and the typical representative is Adam (Adaptive Moment Estimation) optimization algorithm. This kind of algorithm calculates the first moment (mean) and second moment (variance) of the gradient to adaptively adjust the learning rate of each parameter. In the model training process, the model parameters (such as the weights and biases of neural networks) are iteratively updated to minimize the value of the loss function. Adaptive moment estimation algorithm can more efficiently complete this optimization process, which usually converges faster and more stably than traditional algorithms.

[0040] Specifically, first, the parameters (such as weights and biases) of the market trend prediction model are randomly initialized. The training process starts from this initial state, and the parameters are gradually adjusted by the optimization algorithm to make the model accurately fit the data. The loss function (Loss Function) is the core objective of model training, which is used to quantify the difference (error) between the model prediction results and the actual observed values. The goal of adaptive moment estimation is to minimize the value of this loss function, thereby improving the prediction accuracy of the model. The mean square error is calculated by calculating the average of the squared differences between the predicted value and the actual value, which measures the overall prediction error of the model. The square operation amplifies the influence of large errors, prompting the model to focus on and correct the prediction results with large deviations. During the training process, the optimization algorithm continuously adjusts the model parameters, so that the value of the mean square error gradually decreases, that is, the prediction results of the model become more and more close to the true market trend.

[0041] S204, prevent overfitting of the trained market trend prediction model by the validation set.

[0042] In step S204, overfitting refers to the phenomenon where a model learns too much about the details and noise of the training data during the training process, resulting in a model that is overly complex and performs well on the training set but significantly worse on new data (validation / test set). An overfitted model has poor generalization ability and cannot effectively predict unseen data. When training a market trend prediction model, a validation set is used to evaluate the model in real-time to detect and prevent overfitting, ensuring that the trained model maintains good prediction performance on actual market data (i.e. unseen data).

[0043] S3. Adjust high-frequency trading strategy parameters, including trading frequency, trading volume, trading price, and trading timing, through a strategy parameter adjustment algorithm based on the market trend prediction results.

[0044] In step S3, market trend prediction results refer to the prediction results about the future market trend output by the market trend prediction model after analyzing real-time market microstructure data and transaction data. These results may include market volatility prediction, price direction, liquidity trend, trend reversal signals, and other key information. The strategy parameter adjustment algorithm is a specially designed algorithm module that automatically and dynamically adjusts key parameters in high-frequency trading strategies based on market trend prediction results. This algorithm may be based on a rule engine, optimization algorithm (such as genetic algorithm, reinforcement learning), machine learning model or expert system, with the goal of achieving real-time optimization of strategy parameters to match the current market state. High-frequency trading strategy parameters mainly include but are not limited to the following parameters: Trading frequency: the number of transactions initiated per unit of time, one of the core features of high-frequency trading. Adjusting the frequency can control the trading activity, such as increasing the frequency to capture more opportunities when market volatility is high, or reducing the frequency to reduce costs when volatility is low.

[0045] Trading volume: the number of assets (such as stocks, contracts) traded per transaction or per period. Adjusting the trading volume can balance risk and return, such as increasing the volume when liquidity is sufficient and the trend is clear, or reducing the volume to avoid slippage when liquidity is tight.

[0046] Trading price: the bid or ask price. Adjusting the price based on trends (such as raising the sell price when predicting an uptrend, or lowering the buy price when predicting a downtrend) can optimize profit margins.

[0047] Trading timing: the specific time point or condition trigger for placing orders (such as price breaking through key levels, volatility exceeding thresholds). Precise timing can improve the success rate of transactions.

[0048] Specifically, real-time data is analyzed by market trend prediction models (such as the machine learning-based method described above in the embodiments) to generate trend prediction results (such as "the price will rise by 5% in the next 5 minutes" "liquidity will soon decrease" etc.). The strategy parameter adjustment algorithm monitors the trend prediction results and triggers the parameter adjustment mechanism when certain trend signals are detected (such as volatility breaking thresholds, trend reversal predictions). The algorithm calculates new parameter values based on pre-set rules, optimization objectives (such as maximizing returns, minimizing risk) or learned strategies. For example: If the price is predicted to rise rapidly, the algorithm may increase the trading frequency and volume, and adjust the trading price (such as setting a higher sell price).

[0049] If the liquidity is predicted to decrease, the algorithm may reduce the trading frequency and volume to avoid high costs due to insufficient liquidity.

[0050] If a trend reversal signal is detected, the algorithm may immediately adjust the trading direction (from long to short) and optimize the trading timing (wait for the price to pull back to a key support level before entering).

[0051] The adjusted parameters are passed to the high-frequency trading strategy execution module, which updates the strategy configuration in real time and executes trading operations in virtual markets or real markets through an automated system.

[0052] Further, during the process of adjusting high-frequency trading strategy parameters, the strategy parameter adjustment algorithm can set a dynamic threshold. When the market volatility prediction value output by the market trend prediction model exceeds the dynamic threshold, the trading frequency or volume is reduced.

[0053] The calculation formula of the dynamic threshold is as follows: σ(t) = α · σ_{historical} + (1 - α) · σ_{recent}; Where α is the decay factor, and α ∈ [0.6, 0.8], σ_{historical} is the median of the past 30-day volatility, and σ_{recent} is the exponential average number indicator of the recent 5-minute volatility.

[0054] S4. The adjusted high-frequency trading strategy parameters are sent to the user trading platform through a low-latency communication protocol for user reference before executing trading operations.

[0055] In the step S4, the policy optimization module generates adjusted parameters (e.g., transaction frequency is increased to 100 times per second, transaction price is adjusted to the current market price + 0.5%) based on market data. Through the optimized communication protocol (e.g., UDP), the parameters are sent to the user trading platform with extremely low latency (e.g., 1-2 ms). After receiving the parameters, the trading platform displays them to the user in the form of a visual interface (e.g., dashboard, chart) or alert. The user quickly evaluates the parameter suggestions based on the reference information. The user may choose to execute the transaction, adjust the parameters and then execute, or refuse to execute (e.g., if the user finds that the market has abnormal fluctuations, the user may suspend the strategy). If the user approves, the platform submits orders to the exchange through the trading interface to complete the transaction operation.

[0056] S5. Real-time monitoring of the execution results of user trading operations, and updating the market trend prediction model and the strategy parameter adjustment algorithm through online learning mechanism.

[0057] In the step S6, the transaction execution results are used as new "real feedback data" to optimize the parameters of the prediction model (e.g., the weights of the neural network) through online learning mechanism combined with real-time market data, to improve the prediction accuracy of the model on market trends. For example, if the transaction results show that the model has a large prediction bias for a certain type of market fluctuations, the online learning will adjust the model parameters to reduce the error in similar future situations. By analyzing the actual effect of the strategy parameters (e.g., transaction frequency, price) in the transaction results, the online learning mechanism can optimize the logic or parameters of the strategy parameter adjustment algorithm. For example, if increasing the transaction frequency leads to an increase in cost but no increase in revenue, the algorithm will learn this association and be more cautious in future frequency adjustments; or if a certain trend prediction parameter adjustment rule leads to a loss, the algorithm will correct the rule.

[0058] Specifically, the user executes transactions according to the strategy parameters, and the system records the execution results of each transaction (e.g., transaction details, profit and loss, etc.) in real time. The transaction results are combined with real-time market data (e.g., order book, price fluctuations) to form complete feedback data containing "prediction - execution - result". The feedback data is input into the online learning module, and the model parameters are updated using incremental learning algorithms (e.g., online gradient descent, recursive least squares) to correct prediction bias. Based on the feedback results, the parameter adjustment algorithm is adjusted through reinforcement learning (e.g., Q-learning), rule optimization algorithm, or meta-learning (Meta-Learning) (e.g., adjusting the threshold, weight, or logical rules of parameter adjustment). The updated model and algorithm are used again for subsequent strategy parameter generation and transaction execution, forming a closed-loop cycle of "prediction → execution → feedback → learning → optimization" to continuously improve system performance.

[0059] The embodiment also compares the optimization method of the high-frequency trading strategy with the optimization method of the traditional high-frequency trading strategy. The dynamic optimization method of the high-frequency trading strategy has the following beneficial effects compared with the prior art: first, by monitoring market data in real time and dynamically adjusting the trading strategy, the trading efficiency is significantly improved, so that the strategy can respond to market changes more quickly; second, the adaptive algorithm model based on machine learning can accurately adjust the trading parameters, thereby improving the profitability; finally, the method continuously adapts to changes in the market environment through a closed-loop optimization process, enhancing the adaptability and robustness of the strategy. In summary, the method provides an efficient and accurate solution for the dynamic optimization of high-frequency trading strategies, and has important practical value and application prospects.

[0060] As shown in Figure 2 The second aspect of the embodiment provides a virtual system for implementing the optimization method of the high-frequency trading strategy of the first aspect, comprising: a data collection module for acquiring market microstructure data and market transaction data in real time; a market change trend prediction module for inputting the market microstructure data and the market transaction data into a market change trend prediction model to obtain a market change trend prediction result, a strategy adjustment module for adjusting high-frequency trading strategy parameters through a strategy parameter adjustment algorithm according to the market change trend prediction result; a strategy reference module for sending the adjusted high-frequency trading strategy parameters to a user trading platform through a low-latency communication protocol for user reference before performing a trading operation; a monitoring and feedback module for monitoring the execution result of the user trading operation in real time and updating the market change trend prediction model and the strategy parameter adjustment algorithm through an online learning mechanism.

[0061] The working process, working details and technical effects of the aforementioned system provided by the second aspect of the embodiment can be referred to the optimization method of the high-frequency trading strategy of the first aspect, which will not be repeated here.

[0062] As shown in Figure 3As shown, the third aspect of the present embodiment provides a computer device for executing the optimization method of the high-frequency trading strategy according to the first aspect, comprising a storage module, a processing module and a transceiver module connected in sequence, wherein the storage module is configured to store a computer program, the transceiver module is configured to transceive messages, and the processing module is configured to read the computer program and execute the optimization method of the high-frequency trading strategy according to the first aspect. For example, the storage module can include, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a flash memory, a first-in first-out memory (FIFO) and / or a first-in last-out memory (FILO), etc.; and the processing module can be, but is not limited to, a microprocessor of the STM32F105 series. In addition, the computer device can further include, but is not limited to, a power module, a display screen and other necessary components.

[0063] The working process, working details and technical effects of the aforementioned computer device provided by the third aspect of the present embodiment can be referred to the optimization method of the high-frequency trading strategy according to the first aspect, which will not be described here again.

[0064] The fourth aspect of the present embodiment provides a computer-readable storage medium storing instructions of the optimization method of the high-frequency trading strategy according to the first aspect, i.e., the computer-readable storage medium stores instructions, and when the instructions are run on a computer, the optimization method of the high-frequency trading strategy according to the first aspect is executed. The computer-readable storage medium refers to a carrier storing data, which can include, but is not limited to, floppy disks, optical disks, hard disks, flash memories, USB flash drives and / or memory sticks, etc. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable systems.

[0065] The working process, working details and technical effects of the aforementioned computer-readable storage medium provided by the fourth aspect of the present embodiment can be referred to the optimization method of the high-frequency trading strategy according to the first aspect, which will not be described here again.

[0066] The fifth aspect of the present embodiment provides a computer program product comprising a computer program or instructions, which, when executed by a computer, implement the optimization method of the high-frequency trading strategy according to the first aspect. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable systems.

[0067] Finally, it should be noted that the above description is only the preferred embodiment of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for optimizing a high-frequency trading strategy, characterized in that: The following steps are involved: Real-time acquisition of market microstructure data and market transaction data, including order book data, transaction data, and order cancellation data, and market transaction data including transaction price, transaction time, and transaction frequency; Inputting the market microstructure data and the market transaction data into a market trend prediction model to obtain a market trend prediction result, wherein the market trend prediction result includes market volatility, liquidity indicators, and trend direction, and the market trend prediction model is a recurrent neural network model trained based on historical transaction data; Adjusting high-frequency trading strategy parameters using a strategy parameter adjustment algorithm based on the market trend forecast results, wherein the high-frequency trading strategy parameters include trading frequency, trading volume, trading price, and trading timing; The adjusted high-frequency trading strategy parameters are sent to the user's trading platform via a low-latency communication protocol for reference before executing trading operations; The execution results of user trading operations are monitored in real time, and the market change trend prediction model and the strategy parameter adjustment algorithm are updated through an online learning mechanism.

2. The method for optimizing a high-frequency trading strategy according to claim 1, wherein: The market change trend prediction model is a neural network model based on deep learning, specifically selected from one or a combination of the following models: recurrent neural network, long short-term memory network, gated recurrent unit and Transformer model.

3. The method for optimizing a high-frequency trading strategy according to claim 2, wherein: The training method of the market change trend prediction model includes the following steps: Preprocess historical market microstructure data and historical market transaction data, including data cleaning, missing value filling, normalization and feature extraction; The preprocessed data is divided into training set and validation set in a ratio of 7:3; The market trend prediction model initialized by training is trained using an optimization algorithm based on adaptive moment estimation, wherein the loss function uses the mean square error between the predicted value and the actual value as the optimization target; The validation set is used to prevent the market change trend prediction model from overfitting after training.

4. The method for optimizing a high-frequency trading strategy according to claim 1, wherein: The adjusting of high-frequency trading strategy parameters by a strategy parameter adjustment algorithm according to the market change trend includes: When the market volatility forecast value output by the market change trend forecast model exceeds a dynamic threshold, reducing the trading frequency or trading volume; When liquidity anomalies are detected and the market trend prediction model predicts that prices will fall, adjust the transaction price to maintain liquidity; When the model predicts a market trend reversal, adjust the trading direction or terminate the current transaction in advance.

5. The method for optimizing a high-frequency trading strategy according to claim 4, wherein: The dynamic threshold σ(t) is determined by the following formula: σ(t)=α·σ_{historical}+(1-α)·σ_{recent}; Where α is the attenuation factor, and α∈[0.6,0.8], σ_{historical} is the median volatility of the past 30 days, and σ_{recent} is the exponential average volatility index of the last 5 minutes.

6. The method for optimizing a high-frequency trading strategy according to claim 1, wherein: The historical transaction data also includes external influencing factor data, and the external influencing factor data includes at least one of macroeconomic indicators, policy change information, and market sentiment index.

7. A high-frequency trading strategy optimization system, characterized in that: include: Data collection module, used to obtain market microstructure data and market transaction data in real time; A market change trend prediction module is used to input the market microstructure data and the market transaction data into a market change trend prediction model to obtain a market change trend prediction result. A strategy adjustment module, configured to adjust high-frequency trading strategy parameters through a strategy parameter adjustment algorithm based on the market trend prediction results; A strategy reference module is used to send the adjusted high-frequency trading strategy parameters to the user trading platform through a low-latency communication protocol for the user to refer to and then perform trading operations; The monitoring and feedback module monitors the execution results of user trading operations in real time, and updates the market trend prediction model and the strategy parameter adjustment algorithm through an online learning mechanism.

8. A computer device, characterized in that: The method comprises a storage module, a processing module and a transceiver module which are communicatively connected in sequence, wherein the storage module is used to store computer programs, the transceiver module is used to send and receive messages, and the processing module is used to read the computer programs and execute the method for optimizing high-frequency trading strategies as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the method for optimizing a high-frequency trading strategy according to any one of claims 1 to 6 is executed.

10. A computer program product comprising a computer program or instructions, characterized in that When the computer program or the instruction is executed by a computer, the method for optimizing a high-frequency trading strategy according to any one of claims 1 to 6 is implemented.

Images