大數據在金融業中的應用：
以TB銀行為案例分析

大數據（Big Data）在金融方面一般分成二大應用：（1）金融業本身的數據管理；（2）程式交易或自動化交易系統。金融業本身的數據管理：屬於內部數據的分類與分析管理，這方面最常見的應用是客戶管理。在對單一客戶提出理財建議時，可以先透過集團內部所有客戶資料先分析，例如銀行交易、保險、證券交易、其他金融資產交易，以及個人資料分析，如年齡、家庭等，來對掌權客戶，提供理財服務或風險管理服務。程式交易或自動化交易系統：透過歷史資料、實時市場監控等，可以對投資組合自動化管理，以達到交易目的。同樣是以報酬率最大化為目標，但投資組合可以是高風險組合，也可以是低風險組合，程式或自動化交易的重點，在於訓練軟件自動建立投資部位、調整、紀錄，擬定最大風險、停損、分散交易等策略，再透過不斷自我回測，以隨時應對市場變化，最終達到報酬率最大化並打敗市場平均的成果。本研究從以上兩個方面分別做文獻回顧，再從實際事務數據來分析，在建立自動交易系統時的策略與限制，來分析個案。

關鍵字

大數據；程式交易；理財機器人；馬可夫鏈；金融業

並列摘要

Big Data is generally divided into two major applications in finance: (1) data management in the financial industry itself; (2) program trading or automated trading systems. Data-management in the financial industry itself: Classification and analysis management of internal data. The most common application in this area is customer management. When making financial advice to a single customer, you can first analyze all customer data in the group, such as banking transactions, insurance, securities trading, other financial asset transactions, and personal data analysis, such as age, family, etc. Provide financial services or risk management services. Program trading or automated trading system: Through historical data, real-time market monitoring, etc., the portfolio can be automatically managed to achieve trading purposes. The same is to maximize the rate of return, but the portfolio can be a high-risk combination, or a low-risk combination, the focus of the program or automated trading, is that the training software automatically establish investment positions, adjustments, records, formulate the maximum risk, stop Loss, decentralized trading and other strategies, through continuous self-testing, to respond to market changes at any time, and ultimately achieve maximum return and defeat the average market results. This study reviews the literature from the above two aspects, and then analyzes the actual transaction data, and analyzes the cases when establishing the automatic trading system.