基于投资者视角的大宗商品市场预测研究

In recent years, investing in commodity markets has been an important issues interested by both academics and participants. With the development of commodity index investment, commodity prices display an unprecedentedly unstable state driven by macro fundamentals and derivative market speculations, causing a huge investing risk. There have been many studies on commodity market forecasting but very few studies detect the predictability from the perspective of investors. The innovations of this proposal are as follows: On the topic issues, we perform the portfolio allocation and risk analysis based on the forecasts of commodity returns, volatilities and correlations from the investor perspective, providing important suggestions and guidance for commodity investing. On the methodological issues, we introduce time-varying sparsity and mixed data sampling techniques to existing predictive models and combine a series of volatility models and high-dimensional correlation models to propose a set of new forecasting methodology. This method can describe the dynamic effect of different determinants on futures markets such as macro fundamentals, technical factors and investor activities and endogenously select relevant predictors at each point of time. These advantages help us to improve forecasting accuracy and the performance of portfolio allocation and risk management.

近年来，商品市场投资成为学术界和实务界共同关心的重要议题之一。随着商品指数化交易的兴起，在宏观经济和衍生品市场投机行为的共同作用下，商品价格也呈现出前所未有的不稳定状态，造成巨大的投资风险。国内外学者虽然对商品市场预测展开大量卓有成效的研究，但是几乎没有从投资者视角出发，系统性地探究商品市场可预测性。此次项目申请的创新在于：在研究问题层面，本课题拟从投资者视角出发，通过商品收益率、波动率和相关性的预测进行资产最优配置和风险测算，为大宗商品市场投资提供重要的借鉴和指导作用。在研究方法层面，通过在现有预测模型基础上引入时变稀疏技术和混频方法，结合多种低频和高频波动率模型和高维相关模型，提出一套新的预测方法，能够刻画宏观因素、技术因素和期货市场投资行为等混频变量对商品期货市场的动态影响，内生地在不同时点选择合适的预测变量，提高商品市场预测精度，改进资产配置和风险管理效率。

近年来，商品金融化的发展使得商品市场投资成为学术界和实务界共同关心的重要议题之一。随着商品指数化交易的兴起，在宏观经济和衍生品市场投机行为的共同作用下，商品价格也呈现出前所未有的不稳定状态，造成巨大的投资风险。基于这一考虑，课题组从投资者视角研究商品市场可预测性。课题组充分利用宏观基本面、技术指标和期货交易等混合频率的高维信息，通过信息筛选和预测组合技术，在同一个模型框架下根据投资者的实际需求预测商品期货超额收益率、波动率和相关系数。将商品收益率和波动率预测结果应用于资产组合管理、风险管理和期权定价等投资者关注的实践，提高商品市场预测的实用价值。课题组还将商品市场预测方法模型进一步应用于金融市场预测，证明方法创新的普适性。