Research on Quantifier Trading System Based on Time Series

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