A symbolic representation for any data can be used as a tool for reducing the irrelevant noise. Any methods of reducing noise are extremely useful in the field of financial data, where a good trading signal is crucial to achieving the profits for the long term trading approach. In this article, we use the concept of symbolic representation to transform the market situation described as a time series of successive price changes into the simplified representation of this situation. Every element of such symbolic representation is further treated as an attribute in the decision table. On the basis of historical data transformed in the same manner, we try to identify the market situations leading to the increase in the instrument value on the forex market. We use a set of well-known classifiers built and trained with the use of historical data. Finally, we use these classifiers to estimate the possible efficiency of the present market situation. There is no need to exactly identify the quality of the signal. We are interested in price direction rather than the exact price of the instrument, thus we use the concept of fuzzy accuracy. Fuzzy accuracy allows us to properly classify objects belonging not only for the actual decision class but also for the neighboring decision classes. The presented approach is verified with the use of the large set of data collected from the forex market.
展开▼