大额资产清仓的自动化交易策略

市场的机构投资者经常需要清仓手中持有的大额资产, 因此清仓的交易策略成为了关心的问题. 以工商银行的股票为例，给出适用于计算机执行的自动化清仓策略. 首先将高频的工商银行股票历史数据在每个交易日分别划分出48个交易期, 将问题简化为处理每个交易日交易期的数据. 在此基础上, 综合考虑用神经网络模拟预测清仓时股票价格随时间下降的风险和用信息流理论模型衡量的价格冲击和交易时刻, 并通过优化模型得到清仓持续的交易日天数. 此后, 再制定出每个交易日的具体自动化交易策略.在制定日内交易策略 时, 首先用神经网络对交易时刻做出预测, 然后综合考虑使用 VWAP 预测出的交易量和通过 Kalman 滤波方法修正过的期权定价公式预测出的各时刻股票的初始价格, 最终给出详细的交易策略及交易的成本.

Automated Trading Strategies for Large-Value Asset Clearing

Institutional investors in the market often need to liquidate large assets held in their hands, so the trading strategy has become a concern. This paper takes the stock of Industrial and Commercial Bank of China as an example to give an automated clearing strategy suitable for computer execution. First, the high-frequency Industrial and Commercial Bank of China stock historical data is divided into 48 trading periods on each trading day, and the problem is simplified to processing data for each trading day. On this basis, comprehensive consideration is given to the use of neural network simulation to predict the risk of stock price decline over time during liquidation and the price shock and transaction time measured by the information flow theoretical model, then the number of trading days for liquidation is obtained through the optimization model. After that, a specific automated trading strategy is worked out for each trading day. The neural network is used to predict the trading time, and then the trading volume predicted by VWAP and the initial price of the stock at each time predicted by the option pricing formula modified by the Kalman filtering method are comprehensively considered, and finally a detailed trading strategy and the cost of the transaction are given.