The overall optimal portfolio is found using a hybrid(memetic) genetic algorithm where the fitness function of each chromosome (possible subset of stocks) equals the minimal tracking error achievable.
This minimal tracking error is determined by solving the corresponding quadratic
programming problem.
We have shown the experimental setup of the experiments as well as the simulation results.
We conclude that the performance of the tracking portfolio found is much better than that of randomly selected portfolios and that of low capitalized portfolios which can be derived from the AEX-index.
In addition, it is shown that the performance of high capitalized portfolios is similar although worse than the optimal GA tracking portfolio.
This research shows that even a small stock index like the AEX-index can be tracked quite well by a small subset of its composing stocks.
We also analyzed two types of crossover operators.
Generally spoken, the two-point order-based crossover performed much less than the two-point bit equalizer crossover operator.
We further conclude that for our problem, that of tracking the AEX-index, the two-point bit equalizer crossover operator with optimized parameter settings always yields the optimal solution using a relatively small number of generations namely, on average, less than 23 generations.
The overall optimal portfolio is found using a hybrid(memetic) genetic algorithm where the fitness function of each chromosome (possible subset of stocks) equals the minimal tracking error achievable.
This minimal tracking error is determined by solving the corresponding quadratic
programming problem.
We have shown the experimental setup of the experiments as well as the simulation results.
We conclude that the performance of the tracking portfolio found is much better than that of randomly selected portfolios and that of low capitalized portfolios which can be derived from the AEX-index.
In addition, it is shown that the performance of high capitalized portfolios is similar although worse than the optimal GA tracking portfolio.
This research shows that even a small stock index like the AEX-index can be tracked quite well by a small subset of its composing stocks.
We also analyzed two types of crossover operators.
Generally spoken, the two-point order-based crossover performed much less than the two-point bit equalizer crossover operator.
We further conclude that for our problem, that of tracking the AEX-index, the two-point bit equalizer crossover operator with optimized parameter settings always yields the optimal solution using a relatively small number of generations namely, on average, less than 23 generations.
การแปล กรุณารอสักครู่..