Due to the continuous development and growth of domestic and international securities market, investors are more inclined to use professional trading tools to manage investments. So, for major institutions and investors, the transaction strategy of gradually formulating and managing liquidity assets will be meaningful. Therefore, we developed three models: the first model is price prediction model; the second model is investment return based model, and the third model is risk control model. firstly, the daily price data of gold and bitcoin given in the topic for the five-year trading period from November 9, 2016 to October 9, 2021 are preprocessed with missing values, etc., while first-order differences are performed, and the ARIMA model is used to verify the validity of the predicted prices by validating the original series of data with smooth and intrinsic trends. Next, the ARIMA parameters are fitted using historical data, and XGBoost machine learning training data are introduced to triple-fold cross-validate the results, combining ARIMA and XGBoost machine learning to derive predicted prices for daily gold and bitcoin transactions, laying a good foundation for the establishment of the return model and the risk model. We determine the rise and fall of gold and bitcoin in each trading day based on the predicted price data of gold and bitcoin in the next 5 days, and get the median M0.5 of the rise and fall, which reflects the expected return in a specific period by the increase or decrease of the rise. Then, according to the Apriori algorithm, the frequency item set is obtained. According to the plus positioning function, the investment income amount is obtained, and the investment revenue is combined based on this strategy. The result of the benefit is shown in Fig 9. The optimization of the model's accuracy and maximization of the model's accuracy and maximization of investment strategic revenue is proved by the comparative analysis of the prediction model and the revenue of the investment.

Yunyu Li. Research on Quantifier Trading System Based on Time Series. Academic Journal of Computing & Information Science (2023), Vol. 6, Issue 3: 102-110. https://doi.org/10.25236/AJCIS.2023.060313.

[1] Sang H K, Lee H S, Han J K, et al. Pattern Matching Trading System Based on the Dynamic Time Warping Algorithm [J]. MDPI AG, 2018(12).

[2] Young-Seok C. Information-Theoretical Quantifier of Brain Rhythm Based on Data-Driven Multiscale Representation [J]. Biomed Research International, 2015, 2015:1-8.

[3] Lei Y, Liu T, Zhu H. Research on the business model of distributed power trading based on blockchain technology[J]. Journal of Physics: Conference Series, 2021, 1800(1):012014 (8pp).

[4] Contreras, Ivan, Ignacio, et al. A hybrid automated trading system based on multi-objective grammatical evolution [J]. Journal of intelligent & fuzzy systems: Applications in Engineering and Technology, 2017, 32(3):2461-2475.

[5] Lam C, Yao Q, Bathia N. Estimation of latent factors for high-dimensional time series [J]. LSE Research Online Documents on Economics, 2011, 98(4):901-918(18).

[6] Yu W, Yuan W. Research on Network Trading System Using Blockchain Technology[C]// 2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS). IEEE Computer Society, 2018.

[7] Wang J, Yang Q, Jin Z, et al. Research on quantitative trading strategy based on LSTM [C]// 2020 Asia-Pacific Conference on Image Processing, Electronics and Computers (IPEC). 2020.

[8] Liao S. Comparison Research on Quantifier between Japanese and Chinese [J]. Modern Chinese, 2018.

[9] Fu Z, Jiang C. Research on Negative Construction of "One Quantifier N" in International Chinese Teaching [J]. Theory & Practice in Language Studies, 2015, 5(10):2064.