| dc.description.abstract | Abstract
In recent years, with the rise of Financial Technology (FinTech) and Artificial Intelligence (AI), the smart money set off a battle for position in Taiwan′s financial markets. Investors confused on financial operators has launched a variety of Robo-advisor, advertised big data algorithm and custom-made portfolio statement, today. The reason is that the investment strategy which implementing is a black box, how do you make decisions about investment tools that cannot be clearly to state to the public? The proposes of this article try to use a simple program backtesting method to explain the investment strategy and steps of the financial robot, making its implicit investment strategy more clear and clear.
In this study, the best empirical scientific evidence to ＂Robo-Advisor＂. According to freely available market indicators signal on the fundhot web, FBI (Fundhot Bias Index) deviation from the rate corresponding to the closing price of the ETF historical, suitable for buy point in time. During the process, we designed the investment cases based on the fundhot′s investment attitude and general basic financial management knowledge, and used R language to make the FBI signal plus the mutual fund and ETF′s net value as the back test model to obtain the total profit of the simulation investment. The investment strategy performance is compared with the level of risk exposure, and the accuracy of the FBI index signal is estimated through the Keltner Channel(KC) trading simulation, and a practical SOP is created for its investment steps.
The FBI investment method, KC optimization, and MA60 strategy are used in the cases empirical method to verify the results of ETF case backtesting. FBI short-term backtesting returns perform well, and the KC’s strategy profitability is generally close to or even possible. Better than FBI and quarterly strategy. | en_US |