資訊管理學報
陳明琪;林逾先;郭人介;
頁: 153-178
日期: 2008/01
摘要: 本研究的主要目的是運用蟻群最佳化演算法，針對台灣股票市場建構一個理性的投資決策系統，探討ACO系統最適化參數及其在股市投資決策上之投資績效，並研究系統參數變動對投資績效之影響。模型中以股價、20日移動平均線、KD線與成交量等技術分析指標為判斷因子，根據蟻群最佳化之路徑選擇機率進行投資。設定成交量為費洛蒙濃度訊息、KD值為能見度資訊。研究結果發現ACO系統最適化參數為股票個數n=10、費洛蒙殘留係數ρ=0.9、費洛蒙刺激係數α=1、能見度刺激係數β=2。另外，ACO系統之投資績效優於標準系統及大盤系統的投資績效，且ACO系統在研究期間之投資報酬率高達108.64%，遠超過銀行定存之獲利。 ACO系統的參數測定顯示費洛蒙殘留係數ρ值的改變，會影響ACO系統之投資績效，ρ值越高，ACO系統搜尋最佳投資組合之能力越好。且費洛蒙刺激係數α值與能見度刺激係數β值的變動，會影響ACO系統之投資績效，當α=1、β=2時，ACO系統之投資績效最高；當α=1、β=1時，ACO系統之投資績效最差。
關鍵字: 蟻群最佳化;KD指標;成交量;20日移動平均線;
Abstract: This paper studies the optimization of ACO (Ant Colony Optimization) system parameters aiming at constructing a rational investment decision-making system for Taiwan stock market. It studies the effect of changing system parameters on the performance of investment return. Our model takes into account the technical indicators that include stock price, 20-day moving average, KD line and trading volume as the determining factors. Specifically, the trading volume is the phenomenon, the data of stochastic line KD is the visibility. The results of our research reveals that optimal parameters of ACO system are 10 for stock kinds, 0.9 for pheromone trail rate ρ, 1 for pheromone trail weighting parameter α , and 2 for visibility weighting parameter β. Furthermore, the ACO system adopted in this research has better investment return performance than that from the standard system and stock market. The 108.64% investment return rate generated from this research is far better than the profit obtained from fixed term deposit. Our research results indicate that the change of pheromone trail rate ρ will affect the optimal return of the portfolio under ACO system. Specifically higher ρ value leads to stronger capability of searching the optimal investment portfolio through ACO system. The variation of pheromone trail weighting parameter α and visibility weighting parameter β will affect the portfolio performance through ACO. It is found that ACO yields the best investment performance when α=1 and β=2. On the other hand, it generates the least performance when α=1 and β=1.
Keywords: Ant colony optimization;KD technical indicators;stocks trading volume;20-day moving average;
瀏覽次數: 12823 下載次數: 226
引用 導入Endnote
頁: 153-178
日期: 2008/01
摘要: 本研究的主要目的是運用蟻群最佳化演算法，針對台灣股票市場建構一個理性的投資決策系統，探討ACO系統最適化參數及其在股市投資決策上之投資績效，並研究系統參數變動對投資績效之影響。模型中以股價、20日移動平均線、KD線與成交量等技術分析指標為判斷因子，根據蟻群最佳化之路徑選擇機率進行投資。設定成交量為費洛蒙濃度訊息、KD值為能見度資訊。研究結果發現ACO系統最適化參數為股票個數n=10、費洛蒙殘留係數ρ=0.9、費洛蒙刺激係數α=1、能見度刺激係數β=2。另外，ACO系統之投資績效優於標準系統及大盤系統的投資績效，且ACO系統在研究期間之投資報酬率高達108.64%，遠超過銀行定存之獲利。 ACO系統的參數測定顯示費洛蒙殘留係數ρ值的改變，會影響ACO系統之投資績效，ρ值越高，ACO系統搜尋最佳投資組合之能力越好。且費洛蒙刺激係數α值與能見度刺激係數β值的變動，會影響ACO系統之投資績效，當α=1、β=2時，ACO系統之投資績效最高；當α=1、β=1時，ACO系統之投資績效最差。
關鍵字: 蟻群最佳化;KD指標;成交量;20日移動平均線;
The Application of Ant Colony Optimization System on the Investment Strategies at Taiwan Stock Market
Abstract: This paper studies the optimization of ACO (Ant Colony Optimization) system parameters aiming at constructing a rational investment decision-making system for Taiwan stock market. It studies the effect of changing system parameters on the performance of investment return. Our model takes into account the technical indicators that include stock price, 20-day moving average, KD line and trading volume as the determining factors. Specifically, the trading volume is the phenomenon, the data of stochastic line KD is the visibility. The results of our research reveals that optimal parameters of ACO system are 10 for stock kinds, 0.9 for pheromone trail rate ρ, 1 for pheromone trail weighting parameter α , and 2 for visibility weighting parameter β. Furthermore, the ACO system adopted in this research has better investment return performance than that from the standard system and stock market. The 108.64% investment return rate generated from this research is far better than the profit obtained from fixed term deposit. Our research results indicate that the change of pheromone trail rate ρ will affect the optimal return of the portfolio under ACO system. Specifically higher ρ value leads to stronger capability of searching the optimal investment portfolio through ACO system. The variation of pheromone trail weighting parameter α and visibility weighting parameter β will affect the portfolio performance through ACO. It is found that ACO yields the best investment performance when α=1 and β=2. On the other hand, it generates the least performance when α=1 and β=1.
Keywords: Ant colony optimization;KD technical indicators;stocks trading volume;20-day moving average;
瀏覽次數: 12823 下載次數: 226
引用 導入Endnote