Confronto tra l’approccio tradizionale e le tecniche di Machine Learning per la modellizzazione della stagionalità nella valorizzazione degli swap indicizzati all’inflazione

An Inflation-Indexed Swap (IIS) is a derivative in which, at every payment date, a counterparty swaps an inflation rate with a fixed rate. For the calculation of the Inflation Leg cash flows it is necessary to build a mathematical model suitable for the CPI projection. For this purpose, analysts usually start to use market quotes for the Zero-Coupon swaps in order to derive the future trend of the inflation index, together with a seasonality model for capturing the typical periodical effects. Authors propose a forecasting model for inflation seasonality based on feed-forward artificial neural networks with circular neurons: a specific Machine Learning technique suitable for identifying the cyclic nature of time series. The paper is structured in four sections: section 1 illustrates the pricing methodologies for the two most popular IIS: the zero Coupon Inflation-Indexed Swap (ZCIIS) and the Year-on-Year Inflation-Indexed Swap (YYIIS); section 2 deals with the traditional base method for the forecast of CPI values (trend + seasonality); section 3 describes the architecture, the working principle and the validation of the neural network; section 4 concludes comparing the impact of the two seasonality models on the fair-value of an inflation- indexed Year-on-Year swap.