遺傳演算法在財務預測之應用

An Application of Genetic Algorithms on Financial Forecasting

每股盈餘是公司的重要財務資訊之一，它一方面可以提供給投資者作為投資決策之參考，另一方面可以提供給管理者作為管理評量的參考指標之一。過去在每股盈餘等財務預測往往以統計方法進行，因此在自變數選擇上常受到限制，同時有些預測模式其輸出結果往往只能以成長或衰退等二元式的結果表示。而另一方面，以類神經網路預測方式的預測模式可能因變數增加，使得網路變的較複雜。本研究嘗試以人工智慧中的遺傳演算法來作為預測的工具，發展財務預測模型，來預測每股盈餘，解決過去預測方式的限制或缺點。同時也將對過去的遺傳演算法稍做修正，並嘗試以實際值的編碼方式進行編碼，以符合需求。最後進一步比較遺傳演算法和其他預測方式，瞭解以遺傳演算法做於預測每股盈餘工具的特性及優缺點。

Earnings per share (EPS), reflecting the operating performance of a company, is one of the important financial indicators of a company’s financial health. On the one hand, EPS provides information to investors for making investment decisions；on the other hand, it is an index of measuring management performance. In the past, financial forecasting was often done by statistical models. However, these statistical models usually incorporate a limited number of input variables. Besides, some statistical models only provide dichotomous output, such as either “growth” or “decline”. This research attempts to develop a financial forecasting model to forecast a EPS via Genetic Algorithms, which constitute a new area in artificial intelligence. The model can avoid most of the limitations and disadvantages of the traditional models. Here, the genetic algorithms are modified and the real numbers are used to code as a gene of a chromosome to meet the requirements of financial models. Finally, we compare the genetic algorithms financial forecasting model with the other ones in order to understand the features, advantages and disadvantages of genetic algorithms as a financial forecasting tool.