Empirical Study of Stock Selection Model in Taiwan Stock Market

• Main Authors: Liu,Pei-Ling, 劉佩玲
• Other Authors: I-Cheng Yeh
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/03666120813867721856

碩士 === 中華大學 === 資訊管理學系碩士班 === 99 === This paper aims to study the performance of stock in Taiwan market of single factor stock selection model and multi-factor stock selection model, and these models under various simulation parameters, samples based on size and industry, as well as during different period, months, and bull/bear market. The test samples consist of all the listed stocks in Taiwan stock market. The backtest period is from January 1997 to September 2009, a total of 12.75 years. The results showed that (1) the relations between most factors and return are consistent with previous literature, such as Return on equity (ROE), sales growth rate (SGR), market value (MV), trading volume (TV), price-to-book value ratio (PBR), price-to-earning ratio based on last four-season earning(PE4), and price-to-earning ratio based on forecasted earning (PE). Only the market risk factor (β) is inversely proportional to return, which is inconsistent with CAPM theory. In addition, there is no clear relation between prior period returns and current returns, which showed there is neither momentum effect nor reversal effect. There still are some factors with the stock selection ability for large cap stocks, including β, PE4, PE, ROE, and SGR. However, the PBR lose its stock selection ability for large cap stocks, (2) whether for all-cap stocks or large-cap stocks, all the four multi-factor stock selection models have similar risk and higher return than all the single-factor stock selection models except to PE. Considering the rate of return, Growth Value Index based on forecasted earning (GVI) model was the best; considering the systematic risk, the probability that the monthly return is higher than 0, and the probability that the monthly return is higher than the market, then the intersection model was the best, (3) The ratio of stock selection and trading cycle are the two most important simulation parameters of stock selection model. Under monthly trading period, the lower the ratio of stock selection is, the higher the return rate is. Under the ratio of stock selection is only half of the normal ratio, the monthly trading period is still the best, followed by weekly trading period, and quarterly trading period is much lower, (4) GVI (0.25) have the highest annual rate of return in all the five samples based on different size of stocks, and the stock selection method based on value characteristics is more suitable for stocks of electronic industry, (5) Among all the six models under different periods, GVI (0.125), which combines value and growth viewpoint, is significantly effective and with good stability during the total period, whose significance is between 0.001 and 0.1, and (6) value stocks (low PB) are much better than growth stocks (high ROE) in bull market; on the other hand, growth stocks (high ROE) are much better than value stocks (low PB) in bear market.