ABOUT ME

我是陳俊瑋，臺北科技大學碩士班學生，目前在我愛交易實驗室從事量化交易研究，聚焦於機器學習、策略優化與資金管理。

我始終相信一件事：真正穩健的策略，往往是最簡潔的那一個。這正是奧卡姆剃刀定律的精神。

我的策略大多帶有「集成式學習」的風格，簡單來說就是「弱加弱加弱，等於強」。 例如，基本面選股強但反應遲鈍，技術指標時效好但雜訊多；經典規則策略勝率穩定但獲利有限，純機器學習潛在報酬高但容易過擬合。這些方法單獨來看，表現都十分有限。但正因為它們犯錯的時機不重疊，組合在一起後，個別的失誤會相互抵銷，真實的市場訊號則會從多個角度同時被捕捉，逐步浮現。

在研究之外，我持續撰寫論文導讀、記錄策略實驗的心得與失敗教訓，致力於將學術洞察轉化為具實務價值的應用。我相信，分享失敗和分享成功一樣重要。

如果你對量化策略、資金管理或學術研究實務化有興趣，也歡迎與我交流。

My name is Jun-Wei Chen, a master's student at National Taipei University of Technology, currently conducting quantitative trading research at the I Love Trading Laboratory, with a focus on machine learning, strategy optimization and capital management.

I have always believed in one thing: the most robust strategies are often the simplest ones — and that is precisely the spirit of Occam's Razor.

Most of my strategies carry the flavor of ensemble learning, which can be summarized as: weak + weak + weak = strong. For example, fundamental analysis excels at stock selection but reacts slowly; technical indicators are timely but noisy. Classic rule-based strategies offer stable win rates but limited profitability, while pure machine learning holds high return potential but is prone to overfitting. Each of these methods, taken individually, delivers rather limited performance. But because their failure points do not overlap, combining them allows individual errors to cancel each other out, while genuine market signals — captured simultaneously from multiple angles — gradually emerge.

Beyond my research, I continuously write paper reviews and document the insights and lessons learned — including failures — from my strategy experiments. I am committed to translating academic findings into practically valuable applications. I believe that sharing failures is just as important as sharing successes.

If you are interested in quantitative strategies, capital management, or bridging academic research with real-world practice, feel free to reach out and connect with me.