Abstract:
Professionals extensively analyze financial markets in order to make profitable trading strategies, resulting from the price prediction models. The field of computational sciences has been developing prediction models for long but still, chaotic nature of financial market poses a challenge. During the last decade, extensive research has been carried out on forecasting markets using artificial techniques like neural nets, neuro fuzzy systems, statistical techniques and support vector machines. These techniques attempt to model the apparently chaotic behavior of stock markets in order to get any possible predictive value. A realistic hypothesis on nature of markets states that markets swings between two modes efficient and inefficient, intermittently. Among the various approaches being used currently, suffer from market efficiency. As implied by Efficient Market Hypothesis (EMH), no technique can continually beat the market, due to information seepage it becomes public, thus negates its potential. It requires an adaptive methodology that is able to sense the market modes and capture the inefficiencies in order to prolong the prediction capability. This research provides a system sensitive towards market efficiency and adaptive towards changing trends. To achieve this, it constantly monitors the affectivity of Prediction Model and dynamically switches to appropriate Prediction Model. The success of this research lies on the claim of dynamic switching from one prediction model to other depending upon different price formations.
