Standard financial techniques neglect extreme situations and regards large market shifts as too unlikely to matter. Such approach accounts for what occurs most of the time in the market, but does not reflect the reality, as major events happen in the rest of the time and investors are ‘surprised’ by ‘unexpected’ market movements. An
alternative fuzzy approach permits fluctuations well beyond the probability type of uncertainty and allows one to make fewer assumptions about the data distribution and market behaviour.
Fuzzifying the present value criteria, we suggest a measure of the risk associated with each investment opportunity and estimate the project’s robustness towards market uncertainty. The procedure is applied to thirty-five UK companies traded on the London Stock Exchange and a neural
network solution to the fuzzy criterion is provided to facilitate the decision-making process. Finally, we suggest a specific evolutionary algorithm to train a fuzzy neural net - the bidirectional incremental evolution will automatically identify the complexity of the problem and correspondingly adapt the parameters of the fuzzy network.