Forecasting of Global Stock Market by Two Stage Optimization Model
Forecasting of Global Stock Market by Two Stage Optimization Model
复制标题
DOI：
--
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
J. Senoguchi
中科院分区：
文献类型：
--
作者：
J. Senoguchi
目前，美国股市约有一半的交易是基于高频算法交易，这使得养老基金等具有长期投资眼光的投资者难以获得稳定的回报。开发一种能够实现长期稳定回报的市场预测模型，不仅是支持养老金的一个重要问题，也是支持央行政策制定者或新私营企业的一个重要问题。为了通过预测模型获得长期稳定的投资业绩，需要提前去除样本数据中的噪声，提取出一种普遍的模式。但是，很难初步区分噪声和真模式，也很难提前去除噪声。本研究采用两阶段优化决策树将样本空间划分为8个子空间，并利用模式识别模型对每个子空间的通用性进行评价。然后，将通用性较低的子空间定义为噪声较大的子空间，剔除噪声较大的子空间，建立预测模型；结果表明，用这种方法构建的预测模型可以获得比常规方法更高的预测精度。同时，利用金融时间序列数据采用步进法对模型的预测精度进行评价时，获得了过去15年稳定超过基准资产收益率的投资业绩。
: Currently, about half of the transactions in the US stock market are based on high-frequency algorithmic trading, making it difficult for the investors with the long-term investment horizon, such as pension funds, to obtain stable returns. The development of a market forecast model that could achieve stable returns over the long term is an important issue in supporting not only pensions but also the central bank policy makers or new private businesses. To obtain stable investment performance by a forecast model over the long-term, it is necessary to remove noise from sample data in advance and extract a universal pattern. However, it is difficult to preliminarily distinguish between noise and true patterns and remove noise in advance. In this study, the sample space was divided into 8 sub-spaces using a Two Stage Optimization decision tree, and the versatility of each sub-space was evaluated by a pattern recognition model. Then, the sub-space with a low versatility was defined as the space with relatively large noise, and a forecast model was created by excluding the sub-spaces with large noise. It was found that the forecast model constructed in this way could obtain the prediction accuracy higher than that of the conventional method. Also, when the prediction accuracy of the model was evaluated by the walk-forward method using financial time-series data, investment performance that stably exceeded the return of benchmark assets was obtained over the past 15 years.