Abstrakt
We create a new way of comparing models for forecasting stock prices. Our idea was to create a simple game in which the individual models would compete against each other.
We were inspired by the heterogeneous agent models and we created an artificial market. Models act in our artificial market as a forecasting strategies of each agent who trades on the market.
Each agent uses his own model for predicting future prices of risky asset and its dividends. Delayed prices of risky asset and dividends provided the basis for predictions.
The way how agents trade affects the price of risky asset, which in turn influences their expectations and therefore their decisions whether to buy or sell. Moreover, each agent can recalculate his strategy, if he is not satisfied with its performance.
So the forecasting strategies and the artificial market evolve side by side. The models we confront are neural networks VARMA models.
The winning model is the one which earns the most money.