Using Support Vector Machine for Option Trading on Settlement Day

Money management is one of the research topics to experts and scholars (Ralph in The mathematics of money management: risk analysis techniques for traders. Wiley, Hoboken, NJ, USA [1]; Gottlieb in Journal of Applied Probability: 787–795 [1]). The content includes the control of the portfolio and position sizing, which is more important than the study of trading strategies. Money management originated from Kelly criterion, but the theoretical Kelly criterion still has a considerable gap with the real trading. In this paper, we use Ralph Vince’s Optimal fraction (Opt. f ), which is closer to the real trading, and the options trading strategy for money management. Due to the characteristic of options, the prices of out-of-the-money will be zero on settlement date. Therefore, the small fluctuations (−20 points to +20 points) of the Taiwan Capitalization Weighted Stock Index (TAIEX) may cause the prices of out-of-the-money to largely fluctuate. However, we find through historical data that if the options close to the settlement date, the price fluctuations in and out-of-the-money are likely to multiply. In view of this, this paper uses the support vector machine (SVM) algorithm to construct the prediction model and estimate the TAIEX fluctuations probability distribution in the last hour of the settlement date. This distribution is then converted into in- and out-of-the-money options fluctuation distribution. If SVM model can conduct the classifications accurately, we can apply Ralph Vince’s Opt. f management position sizing, so that the option trading strategy on the settlement date is profitable.