隨著金融自由化,越來越多的金融性商品可供投資人選擇,這些金融性商品,各有不同的風險和其報酬,但個別投資人的專業知識及時間資金有限,相對可選擇的投資項目較少,風險不易分散;而「共同基金」其特就就是集合投資者的資金,由專業經理人代為管理,依循既定的投資理念與策略。 馬可夫決策過程(Markov Decision Process)是由馬可夫鏈(Markov Chain)所衍生的模型之一,此模型已廣泛地被使用在各種領域,因此本文希望利用馬可夫決策過程,再藉由基金多角化投資的概念,降低風險,並建構一決策模型,提供投資人在買賣基金時所能參考的客觀參考,結果發現在為期3年的測試期中,其績效仍不及買入持有策略。
We developed a model by Markov Decision Process which is developed from Markov chain, and used the model to make the decision in fund market. The model is composed of 14 different state indexes. At first we defined NAV in 14 different state index, and calculated the switch times between each state. We can calculated the transition probability matrix by the switch times between each state, and use the matrix to calculate the expected return at different state index. The model will make decision by the expected return. The model will provide an objective suggestion for the investors. Lastly, by the evidences I conducted under my methodology, people can find the MDP model shows no significant advantage against the normal strategy ‘Buy and Hold’ in a continued growth market.