Publicação

Markovian model for forecasting financial time series

Ver documento

Detalhes bibliográficos
Resumo:The study aims to create a Markovian model for forecasting financial time series and measure its effectiveness on stock prices. In the study, the new forecaster was inspired by several machine learning techniques and included statistical approaches and conditional probabilities. Namely, Markov Chains and Hidden Markov Chains are the main inspiration for machine learning techniques. To be able to process time series with Markov Chains like algorithm, new transformation developed with the usage of daily stock prices. Thirteen years of daily stock prices have been used for the data feed. For measuring the effectiveness of a new predictor, the obtaıned results are compared with conventional methods such as ARIMA, linear regression, decision tree regression and support vector regression predictions. The comparisons presented are based on Mean Absolute Percentage Error (MAPE) and Root Mean Square Error ( RMSE). According to the achieved results, the new predictor performs better than decision tree regression, and ARIMA performs best among them.
Autores principais:Hasanbas, Ersin
Assunto:Time series Machine learning Cadeia de Markov -- Markov Chain Forecasting Séries temporais Aprendizagem das máquinas Previsão
Ano:2020
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso aberto
Instituição associada:ISCTE
Idioma:inglês
Origem:Repositório ISCTE
Descrição
Resumo:The study aims to create a Markovian model for forecasting financial time series and measure its effectiveness on stock prices. In the study, the new forecaster was inspired by several machine learning techniques and included statistical approaches and conditional probabilities. Namely, Markov Chains and Hidden Markov Chains are the main inspiration for machine learning techniques. To be able to process time series with Markov Chains like algorithm, new transformation developed with the usage of daily stock prices. Thirteen years of daily stock prices have been used for the data feed. For measuring the effectiveness of a new predictor, the obtaıned results are compared with conventional methods such as ARIMA, linear regression, decision tree regression and support vector regression predictions. The comparisons presented are based on Mean Absolute Percentage Error (MAPE) and Root Mean Square Error ( RMSE). According to the achieved results, the new predictor performs better than decision tree regression, and ARIMA performs best among them.