摘要 :
The essence of the present study is to investigate the applications of multi-source data fusion methodology in economic systems.To introduce the ideas of data fusion, an element ary discussion of terminologies and structures is pr...
展开
The essence of the present study is to investigate the applications of multi-source data fusion methodology in economic systems.To introduce the ideas of data fusion, an element ary discussion of terminologies and structures is provided from engineering perspective. These developments in turn motivate the first applicat ion, on obtaining optimal volatility based on decisions t aken by stock exchange agents monitoring the market for volatility. This is made possible by putting the decision problem into Bayesian framework and having it analyzed.Single market model is a linear equation with unknown parameters. To estimate the parameters and their variances, the model is put into the Linear Minimum variance framework and parameters estimated. Extension to multi-factor market model is also discussed. Investigation when the matrix of systematic errors is non-singular as a way of generalizing the model is considered. Also modeled is the wish of any investor preferring higher returns to lower returns; this is achieved by constraining the parameters in the model.Finally, a portfolio management model is analyzed. The focus is on the case where the multiple returns are correlated. This generalizes the model, whereby a model with uncorrelated multiple returns is merely a special case. Pseudo-inverse technique and quadratic programming theory are used in obtaining the general solutions. The notion is further extended to constrained parameter model. Overall, generalized portfolio weights are obtained. Multiple expected returns with linear equality and linear inequality constraints are also studied.
收起