| Autor: |
Beckmann, M., Künzi, H. P., Fandel, G., Trockel, W., Basile, A., Drexl, A., Dawid, H., Inderfurth, K., Kürsten, W., Ehrentreich, Norman |
| Zdroj: |
Agent-Based Modeling; 2008, p187-193, 7p |
| Abstrakt: |
The general sequence of activities in the SFI-ASM is summarized and represented in figure 11.1. At the beginning of period t, a new dividend dt is announced. For each agent i = 1..., N, the GA is invoked with probability 1/K to change an agent's rule set Ti. Afterwards, agents determine their active rule set ωiTi by comparing the condition parts of each trading rule j = 1..., 100 with the binary market descriptor. From that active rule set, agents select the rule with the highest forecast accuracy for their forecast production. The price formation process is then initiated by the specialist who announces a trial price Ptrial equal to last period's price. Based on this trial price, agents form their expectation about next period's price and dividend, determine their optimal demand for the risky stock, and then submit their offers and bids to the specialist. If the bids and offers cannot be matched, the specialist determines a trial price and the whole process starts all over. This iterative process ends when the offers are balanced by the submitted bids or after 10 unsuccessful trial rounds. In the latter case, one side of the market will be proportionally rationed. The last trial price is announced to be the stock price for period t and all trades between agents are executed at that price. Since at this time, the dividend and the stock price in period t are known to the agents, they are now able to update the forecast performance of all rules which were activated in the last period and actually produced a forecast about this period's price and dividend. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Supplemental Index |
| Externí odkaz: |
|
