台股投資更聰穎：從機器學習到區塊鏈結合遊戲化應用

Abstract

本研究旨於實現新型區塊鏈投資平台的概念性驗證 (Proof-of-Concept)，並結合智能投資及遊戲化概念。雖然目前證券或軟體應用市場上存在著相關的產品和研究，但很少有研究致力於這三個概念的結合與應用。本研究由兩部分實驗組成：(1) 透過機器學習方法實作一套量化交易策略，找出一籃未來最具漲勢的股票並做多進行買入，持有N天後賣出，此階段的實驗結果會作為下一部分區塊鏈應用中的一項重要元素；(2) 建構基於區塊鏈的遊戲平台原型，並結合上一部分的量化投資策略，實現新穎的智能投資服務。在第一部分中，預測模型的表現穩健，相較於大盤指數皆能在不同的市場狀況之下獲得超額報酬；在第二部分的實驗中提出並實現了區塊鏈平台原型，透過遊戲化的概念包裝能為參與者提供更多的誘因參與此區塊鏈生態系統，而建構基於區塊鏈與智能合約的去中心化基礎之上，安全性、公平性與可靠性也能被充分實現。

This study is a proof of concept (PoC) of constructing an investment gaming platform based on Blockchain, quantitative trading and gamification. Although there exist some related products and researches in the current marketplace. But few studies work on the combination of these three concepts. This study consists of two parts: (1) developing a quantitative trading strategy through machine learning approach to find out the most profitable stocks in the future and the results will be used as a feature of the proposed gaming platform; (2) building a blockchain-based gaming application incorporating with the concept of smart trading. In the first part, the performance of the predictive model is quite well and the model can beat the market under different market situations. In the second part, a prototype is proposed and implemented. Through gamification, it is capable of providing more incentives for people to participate. Safety and reliability can also be ensured by the nature of Blockchain and smart contracts.