國立高雄大學圖資館 |

Language: English

Back

一個使用機器學習模型的資金管理系統的比較性研究 = A Comparat...

國立高雄大學資訊工程學系碩士班

一個使用機器學習模型的資金管理系統的比較性研究 = A Comparative Study of Money Management Systems Using Machine Learning Models

Record Type:	Language materials, printed : monographic
Paralel Title:	A Comparative Study of Money Management Systems Using Machine Learning Models
Author:	洪嘉澤,
Secondary Intellectual Responsibility:	國立高雄大學
Place of Publication:	[高雄市]
Published:	撰者;
Year of Publication:	2013[民102]
Description:	77面圖，表格 : 30公分;
Subject:	遺傳演算法
Subject:	money management
Online resource:	http://handle.ncl.edu.tw/11296/ndltd/03453786328125688851
Notes:	參考書目：面63-66
Notes:	102年10月31日公開
Summary:	在這篇論文中，我們提出一個應用遺傳演算法與機率整合體來建立一個兼具降低風險與提升績效的交易組合模型的比較性研究。為了驗證資金管理模型的效用，我們使用技術指標來決定交易系統中的進出市場時機。我們同時採用多種不同的目標函數以及績效指標來驗證模型的優劣，包含複利報酬率、Sharpe ratio、Calmar ratio以及Sortino ratio。首先實驗結果顯示，以複利報酬率比較，遺傳演算法與機率整合體所演化出的模型明顯優於買進持有策略。另外我們實驗結果也表明，使用機率整合體所演化出的模型會比遺傳演算法更為強健。最後實驗結果顯示，百分率模型是本文中最有效的資金管理系統。我們期望本文所提出的方法可以推進機器學習方法於計算財務學的應用研究。 In this thesis we present a comparative study of money management systems using two machine-learning approaches—Genetic Algorithms (GA) and Probability Collectives (PC). To investigate the effect of money management models, we purpose using technical indicators to determine the market timing for our trading systems. We also use a combination of various objective functions and performance indicators to evaluate our models, including annualized returns, Sharpe ratio, Calmar ratio as well as Sortino ratio. Our results showed that (1) both the GA and the PC outperformed the benchmark in terms of return and risk; (2) the models optimized by the PC are typically more robust than those by the GA; (3) the percent risk model tends to be the most effective in all the money management models we tested. We thus expect this proposed methodology to advance the current state of machine learning for computational finance.

一個使用機器學習模型的資金管理系統的比較性研究 = A Comparative Study of Money Management Systems Using Machine Learning Models
洪, 嘉澤

一個使用機器學習模型的資金管理系統的比較性研究 = A Comparative Study of Money Management Systems Using Machine Learning Models / 洪嘉澤撰 - [高雄市] : 撰者, 2013[民102]. - 77面 ; 圖，表格 ; 30公分.
參考書目：面63-66102年10月31日公開.
遺傳演算法money management

一個使用機器學習模型的資金管理系統的比較性研究 = A Comparative Study of Money Management Systems Using Machine Learning Models
LDR:03090nam0a2200289 450 001 389685
005 20170214090331.0
009 389685
010 0 $b 精裝
010 0 $b 平裝
100 $a 20170214d2013 k y0chiy05 e
101 1 $a chi $d chi $d eng
102 $a tw
105 $a ak am 000yy
200 1 $a 一個使用機器學習模型的資金管理系統的比較性研究 $d A Comparative Study of Money Management Systems Using Machine Learning Models $z eng $f 洪嘉澤撰
210 $a [高雄市] $c 撰者 $d 2013[民102]
215 0 $a 77面 $c 圖，表格 $d 30公分
300 $a 參考書目：面63-66
300 $a 102年10月31日公開
314 $a 指導教授：黃健峯博士，張志向博士
328 $a 碩士論文--國立高雄大學資訊工程學系碩士班
330 $a 在這篇論文中，我們提出一個應用遺傳演算法與機率整合體來建立一個兼具降低風險與提升績效的交易組合模型的比較性研究。為了驗證資金管理模型的效用，我們使用技術指標來決定交易系統中的進出市場時機。我們同時採用多種不同的目標函數以及績效指標來驗證模型的優劣，包含複利報酬率、Sharpe ratio、Calmar ratio以及Sortino ratio。首先實驗結果顯示，以複利報酬率比較，遺傳演算法與機率整合體所演化出的模型明顯優於買進持有策略。另外我們實驗結果也表明，使用機率整合體所演化出的模型會比遺傳演算法更為強健。最後實驗結果顯示，百分率模型是本文中最有效的資金管理系統。我們期望本文所提出的方法可以推進機器學習方法於計算財務學的應用研究。 In this thesis we present a comparative study of money management systems using two machine-learning approaches—Genetic Algorithms (GA) and Probability Collectives (PC). To investigate the effect of money management models, we purpose using technical indicators to determine the market timing for our trading systems. We also use a combination of various objective functions and performance indicators to evaluate our models, including annualized returns, Sharpe ratio, Calmar ratio as well as Sortino ratio. Our results showed that (1) both the GA and the PC outperformed the benchmark in terms of return and risk; (2) the models optimized by the PC are typically more robust than those by the GA; (3) the percent risk model tends to be the most effective in all the money management models we tested. We thus expect this proposed methodology to advance the current state of machine learning for computational finance.
510 1 $a A Comparative Study of Money Management Systems Using Machine Learning Models $z eng
610 0 $a 遺傳演算法 $a 機率整合體 $a 資金管理 $a 技術指標 $a 績效指標
610 1 $a money management $a technical Indicators $a Probability Collectives multi-agent systems $a Genetic Algorithms $a performance indicators
681 $a 008M/0019 $b 464103 3443 $v 2007年版
700 1 $a 洪 $b 嘉澤 $4 撰 $3 614545
712 0 2 $a 國立高雄大學 $b 資訊工程學系碩士班 $3 353878
801 0 $a tw $b NUK $c 20131011 $g CCR
856 7 $z 電子資源 $2 http $u http://handle.ncl.edu.tw/11296/ndltd/03453786328125688851