查詢結果分析
來源資料
頁籤選單縮合
| 題 名 | 基因表達規劃法之動態交易策略設計研究--以臺灣股市為例=A Study of Dynamic Trading Strategy Design Based on Gene Expression Programming--Evidence from the Taiwan Stock Market |
|---|---|
| 作 者 | 林青峰; 林湘霖; 李明祐; | 書刊名 | 輔仁管理評論 |
| 卷 期 | 32:2 2025.05[民114.05] |
| 頁 次 | 頁65-107 |
| 分類號 | 312.83 |
| 關鍵詞 | 隨機森林; 基因表達規劃法; 交易策略; Random forest; LGBM; Gene expression programming; Trade strategy; |
| 語 文 | 中文(Chinese) |
| 中文摘要 | 隨著金融科技發展,人工智慧被應用於各類金融商品上,如何從複雜多變的金融市場找出一套能夠獲利的交易策略是一項挑戰。本研究採用基因表達規劃法(GEP)從過去歷史資料中搜尋有效的交易策略,以交易訊號、資金管理與風險控管三種基因構建一套交易模型。此外,隨機森林(RF)與LGBM兩種決策樹方法,在每一期交易標的中動態更新投資組合,期望透過此種方法提高交易模型的獲利與降低交易時的風險。實驗結果發現,藉由動態選股結合GEP交易模型(基本面與技術面)的方法,相較於純GEP模型(技術面)與選股模型(基本面),確實能有效的提升GEP交易模型各期的交易績效,如獲利、勝率與風險等。 |
| 英文摘要 | With the development of financial technology and artificial intelligence being applied to various financial products, it is a challenge to find a set of profitable trading strategies from the complex and ever-changing financial market. This study uses Gene Expression Programming (GEP) to search for effective trading strategies from past historical data, and builds a set of trading models based on three genes, namely, trading signals, capital management and risk control. In addition, two decision tree methods are used to dynamically update the investment portfolio in each period, namely, Random Forest and LGBM. Hoping through these two methods, the profit of the GEP trading model could be improved, and the risk of trading could be reduced. The results of this research found that combining dynamic stock selection with the GEP trading model, it can indeed effectively improve the trading performance of each period of the GEP trading model. Compared with the pure GEP model (technical aspect) and stock selection model (fundamental aspect), it can indeed effectively improve the trading performance of the GEP trading model in each period, such as profit, winning rate and risk. |
本系統中英文摘要資訊取自各篇刊載內容。