Resumen

MOTA-HERNANDEZ, Cinthya I.; ALVARADO-CORONA, Rafael  y  JIMENEZ GARCIA, Benita Martha. RNA's en el Análisis a Subyacentes de Divisas y Deuda Cotizadas en el MexDer. RevActaNova. [online]. 2019, vol.9, n.2, pp.190-203. ISSN 1683-0789.

This research paper reports a model for forecasting the trend in the currency assets (dollar and euro) and two debt (CETES and TIIE) used in futures quoted in the Mexican Derivatives Market (MexDer), applying the approach of Artificial Neural Networks (RNA's). One of the main contributions suggests that the proposed model based on Artificial Intelligence makes approximations with an error lower than 19%. The topologies used for the comparison of data are: Perceptron Multilayer and Recurrent networks with time series, which offer the possibility of obtaining approximations within an acceptable range in the forecast in the financial area, considering that the greater the risk the greater the gain. As previously mentioned, 7 variables that represent currencies are evaluated: Dollar and Euro; the debt indices: TIIE at 28 and 91 days, the CETES at 28 and 91 days and inflation is added because it is one of the important macroeconomic variables. Because it is about predicting short, medium and long-term variables over time, the input data for each network model takes into account: the current date, current inflation, the current dollar as a variable representing the exchange rate and the prediction date. The RNAs were trained with historical data in which the entries consider dates from 1969 and the outputs with historical data from 2003. The data of the variables obtained will be compared with real data for the years 2016, 2017 and 2018 and part of the 2019, to validate the RNA considered optimal.

Palabras clave : Analysis; ANN's; Foreign Exchange; MexDer.

        · resumen en Español     · texto en Español     · Español ( pdf )