Submitted by ALEXANDRE CONCEICAO (alemgc@gmail.com) on 2017-09-14T01:37:47Z No. of bitstreams: 1 dissertacao-alexandre-conceicao.pdf: 2127579 bytes, checksum: 809d5da780c4b1ae1c60b356b8ef1fbf (MD5) / Approved for entry into archive by Thais Oliveira (thais.oliveira@fgv.br) on 2017-09-14T17:48:55Z (GMT) No. of bitstreams: 1 dissertacao-alexandre-conceicao.pdf: 2127579 bytes, checksum: 809d5da780c4b1ae1c60b356b8ef1fbf (MD5) / Made available in DSpace on 2017-09-15T12:16:46Z (GMT). No. of bitstreams: 1 dissertacao-alexandre-conceicao.pdf: 2127579 bytes, checksum: 809d5da780c4b1ae1c60b356b8ef1fbf (MD5) Previous issue date: 2017-08-23 / Using dynamic factor models, one can analyse stocastic processes having multiple dimensions, that being the case on financial markets when considering the series formed by the stock prices. By using the model predictions it is possible to create trading strategies and measure their performance. With the closing prices of the stocks belonging to the IBX100 index from 2010 till 2016, the model parameters were estimated and trading strategies were proposed. The performance indicators of some strategies were superior than those of the IBOVESPA index on the given period of time, offering anualized returns of 27% and a 48% maximum drawdawn against 1.12% anualized return and 48% maximum drawdawn on the IBOVESPA. Therefore, those models are capable of capturing the price dynamic in such a way that their predictions can be used to create trading strategies having a performance higher than that of the IBOVESPA index. / O uso de modelos de fatores dinâmicos permite analisar processos estocásticos com grande número de dimensões, sendo exatamente esse o caso do mercado financeiro quando se consideram as séries formadas pelos preços de ações. Ao se utilizar as previsões feitas pelos modelos, é possível criar estratégias de trading cuja performance pode ser aferida. De posse dos preços de fechamento de ativos pertencentes ao IBX100 no período de 2010 até 2016, os parâmetros do modelo foram estimados e estratégias de trading foram propostas. Os indicadores de performance de algumas das estratégias superaram aqueles do índice IBOVESPA no período estudado, oferecendo retorno anualizado de até 27% e máximo drawdawn de 21%, contra um retorno de 1.12% e max drawdawn de 48% do IBOVESPA. Portanto, esses modelos são capazes de capturar a dinâmica dos preços de ações na medida em que suas previsões podem ser utilizadas para criar estratégias de trading com performance superior à do Índice IBOVESPA.