| 摘 要: 人工智能技术和量化投资领域的结合，诞生了各类基于机器学习算法的价格预测模型。为研究不同机器学习算法在股指期货价格预测中的应用效果，采用支持向量回归、长短期记忆网络、随机森林及极端梯度提升树四种常用的机器学习算法构建价格预测模型，对沪深300股指期货价格进行预测研究，并利用贝叶斯算法对模型进行超参数优化，对比贝叶斯优化对于以上四种机器学习算法预测精度的提升效果。研究结果表明，随机森林和极端梯度提升树因其模型自身的优点，可以实现对金融时序数据的准确预测，而贝叶斯优化利用高斯过程，不断更新先验，可以显著提高支持向量回归预测效果，均方误差(MSE)、平均绝对误差(MAE)、对称平均绝对百分比误差(SMAPE)和损失适应度(LOSS)分别降低了78.6%、94.7%、95.1%和97.0%。 |
| 关键词: 机器学习；支持向量机；长短期记忆网络；随机森林；极端梯度提升树 |
|
中图分类号: TP312
文献标识码: A
|
| 基金项目: 青海民族大学研究生创新项目基金(65M2022186). |
|
| Research on Stock Index Futures Price Prediction Model based on Machine Learning Algorithms |
|
YANG Xuewei
|
(School of Economics and Management, Qinghai University for Nationalities, Xining 810007, China)
1661315325@qq.com
|
| Abstract: With the combination of artificial intelligence technology and quantitative investment, various price prediction models based on machine learning algorithms have emerged. In order to study the effect of different machine learning algorithms on stock index futures price prediction, this paper proposes to use four commonly used machine learning algorithms, namely SVR (Support Vector Regression), LSTM (Long Short-Term Memory), RF (Random Forest) and XGBoost (Extreme Gradient Boosting), to construct a price prediction model, so as to predict the stock index futures price of Shanghai and Shenzhen 300. Bayesian algorithm is used to optimize the hyperparameters of the model, and the improvement effect of Bayesian optimization on the prediction accuracy of the four machine learning algorithms is compared. The research results show that RF and XGBoost can achieve accurate prediction of financial time series data due to their own advantages, while Bayesian optimization can significantly improve the prediction effect of support vector machines by using Gaussian process and constantly updating the prior. MSE, MAE, SMAPE and LOSS are reduced by 78.6%, 94.7%, 95.1% and 97.0% respectively. |
| Keywords: machine learning; SVR; LSTM; RF; XGBoost |
