WIT Press

Application Of Fuzzy Models And Neural Models In Financial Time Series


Free (open access)








559 kb

Paper DOI



WIT Press


M. A. F. Allemão, A. G. Evsukoff & N. F. F. Ebecken


Brazilian bank institutions daily perform thousands of trips for money collection and delivery at their branches. The amount spent in these operations is very high and has grown in recent years (U$ 35 million in 1999 and U$ 75 million in 2002). The reduction of money supplying and collecting trips at agencies is a predominant factor in reducing the operational costs of the system. There is a need to make reliable forecasts based on the behavior observed in the agency’s past. According to the Bank’s analysts, the incoming resources are very strong on Mondays and decrease on Tuesdays. Wednesdays are considered balanced days. Thursdays have paying characteristics and Fridays too; the last days of the week have more resources coming out than any other day. Regarding the monthly cycle, the behavior of an agency has strong payments in the beginning of each month, growing slightly to balance the remaining days and money coming out at the end of the month. Concerning the months of the year, agencies have larger withdrawals in the summer months, especially from December to January. The balance is observed from the third or fourth month until approximately the ninth and tenth month of the year. Considering the use of the Vague Logic concepts to translate into mathematic terms the inaccurate information contained in sentences expressed in natural language and in creating models based on previous premises to map the agency’s behaviour, the aim is to help analysts in the task of defining the optimal amount of money to be remitted. Neural models have been developed to compare, they have reached lower performance rates than those verified with the use of vague algorithms. Keywords: fuzzy logic, neural networks, time series, forecast, money prediction, agency’s behavior.


fuzzy logic, neural networks, time series, forecast, money prediction, agency’s behavior.