A neural fuzzy system for asset valuation

Asset valuation has long been a challenging and important task in finance. It is commonly performed prior to the sale of an asset and it may consist of both subjective and objective measurements. Suppose that the value of our asset depends on the currency fluctuations on the global financial markets. Suppose further that we are in the position to derive fuzzy rules for modeling the partially known causal links between the exchange rates and the asset values. In this work we suggest the use an ANFIS architecture based on Mamdani inference mechanism for asset valuation, since it can work with arbitrary membership functions. After learning the shape parameters of fuzzy numbers one can easily quantify the functional relationship between the exchange rates and the asset prices.