Aplicação de redes neurais para previsão de contrato de dólar futuro no mercado brasileiro

Resumo

The objective of this work is to develop a neural network to predict the movement of the 1-month BRL/USD future contract and to develop trading strategies for futures and options, thus providing a tool for estimating the movement of exchange rates and trading strategies based on these assets. This tool can help companies that need to hedge assets and liabilities as well as market players who need to increase performance of their portfolios. In this work, we use as input market data available on Bloomberg from January 2001 to September 2013 and for the calculation of the financial results of the strategies we use reference prices provided by BM&F. The best neural network obtained has an accuracy rate slightly above 70%, the financial result of strategies outperforms the usual market benchmark, the CDI. The options strategy presented a better result than the future strategy, both for the period of 2013 and for the whole period considered, generating a return of 884% versus 573% in the period from September 2005 to September 2013, with the CDI in this period yielding 132%. From January to September 2013, the options strategy also outperformed the future strategy, yielding 31% versus 24% whereas the CDI yielded 6% in the same period.

Ficha do documento

Tipo

Dissertação

Ano

2014

Instituição

Fundação Getulio Vargas

Fonte

Repositório da FGV

Idioma

Português

Acesso

Acesso aberto

Identificador

oai:repositorio.fgv.br:10438/11502

Temas

Economia

Palavras-chave

Redes neuraisDólar futuroOpçãoEconomiaRedes neurais (Computação)Mercado futuroCâmbioMercado financeiro - Brasil