Making Profit in Stock Investment Before XD Dates by Using Genetic Algorithm - Innovative Management in Information and Production - page 336

well prepared for the investment or speculation. With more bidding volumes from
more confident investors, then the price can rise sharper in the this period than in
the second half.

5.3 Comparison Between 30, 40, and 50 days Before

the Selling Point

It is hypothesized that if the genetic algorithm is robust, then the results from the 30,
40, and 50 days before XD dates should not be different. For example, if the regime
switching is at 25 days before XD dates, then all the three models should point out
similarly that regime switching takes place at day 25. The results do not confirm
this hypothesis. Many solutions from data of 30, 40, and 50 days differ. However,
it can be understood that the longer data allows an investor to find a better buying
position that lies further from the XD dates, for example, when the model using data
of 30 days found that day 25 is the regime switching but the model with data of 50
days may find another lower regime at day 46. This lower regime at day 46 can be
lower than the lower regime found at day 25. Technically, it is the multiple regimes,
containing more than one lower and one higher regime.

5.4 Best Buying Strategy of Each Stock

All the eight stocks are profitable for the investment before XD dates in the first
half of the year. Good stocks for the short-term investment which an investor will
buy and hold the stocks around 15 days until the selling day before XD dates, are
ADVANC and PTT. For medium-term investment, 16-30 days are CPALL and SCC.
The stocks for the long-run investment that an investor must buy more than 30 days
prior to the XD dates are CPF, IVL, KBANK, and TCAP.
The rate of return is attractive with more than 100 % per year in two stocks,
IVL and TCAP. CPF yields the return almost up to 100 %. SCC is also good for
the investment with its 76 % of the return. Other stocks yield around 50 %. For all
stocks, the annual rate of return is around 76 %. The average buying days are around
31 days before the selling day (Table 3 ).

5.5 Performance of Genetic Algorithm for Detection

of Regime Switching

Sudtasan and Suriya ( 2012 ) found in the detection of regime switching before the
year end that the performance of genetic algorithm is low, around 37 % of the highest
potential profit. This study breaks this record when finding the performance of 62 %
when applying the algorithm with zero mutation in the first half of the year.