El enfoque de espacio de estado en el análisis de las series de tiempo financieras

Otros responsables
Fecha
09/2017
Editorial
Universidad Nacional de La Matanza. Departamento de Ciencias Económicas
Revista
Revista RINCE
Título del libro
Nombre del evento
Fecha del evento
Institución organizadora
ISSN
1851-3239
eISSN
ISBN
Idioma
Español
Tipo del evento
Clasificación temática
Palabras clave
Resumen
El objetivo de este trabajo es examinar los métodos para tratar una gran variedad de datos con irregularidades que suceden en series de tiempo. Los modelos autorregresivos integrados de promedios móviles (ARIMA) son frecuentemente considerados como los que proveen la base principal para el modelado de series de tiempo. Ahora bien, dada la tecnología actual, puede haber alternativas más atractivas. Una nueva y poderosa solución fue ideada por Kalman (1960) y Kalman y Bucy (1961), usando la llamada representación de espacio de estado de una serie de tiempo. Esto provee una descripción muy compacta del modelo y está basado en el resultado conocido que dice que cualquier ecuación en diferencias (o diferencial) lineal de orden finito puede ser escrita como una ecuación vectorial en diferencias (o diferencial) lineal de primer orden. La ventaja de esta última representación es que involucra solamente dependencia de un paso, lo cual conduce a un algoritmo simple para calcular las predicciones de valores futuros de la serie conocido como el algoritmo del filtro y suavizador de Kalman. Presentaremos las ideas básicas de modelado estructural de series de tiempo y haremos notar la relación con los modelos autorregresivos integrados de promedios móviles.
The aim of this paper is to examine the methods for treating a variety of data with irregularities occurring in time series. The autoregressive integrated moving average models (or ARIMA models) are often considered to be the ones that provide the main basis for modeling any time series. However, given current technology, there may be more attractive alternatives. A powerful new solution was devised by Kalman (1960) and Kalman and Bucy (1961), using the so-called state space representation of a time series. This provides a very compact description of the model and is based on the known result that says any difference equation (or differential) that is linear and has a finite order can be written as a linear vector (or differential) equation in differences of first order. The advantage of the latter representation is that it involves only one step dependence, which leads to a simple algorithm to calculate predictions of future values of the series, known as the algorithm of the Kalman filter and smoother. We will introduce the basic ideas of structural modeling of time series and we will present the relationship with autoregressive integrated moving averages or ARIMA models.
The aim of this paper is to examine the methods for treating a variety of data with irregularities occurring in time series. The autoregressive integrated moving average models (or ARIMA models) are often considered to be the ones that provide the main basis for modeling any time series. However, given current technology, there may be more attractive alternatives. A powerful new solution was devised by Kalman (1960) and Kalman and Bucy (1961), using the so-called state space representation of a time series. This provides a very compact description of the model and is based on the known result that says any difference equation (or differential) that is linear and has a finite order can be written as a linear vector (or differential) equation in differences of first order. The advantage of the latter representation is that it involves only one step dependence, which leads to a simple algorithm to calculate predictions of future values of the series, known as the algorithm of the Kalman filter and smoother. We will introduce the basic ideas of structural modeling of time series and we will present the relationship with autoregressive integrated moving averages or ARIMA models.
Colecciones
Aprobación
Revisión
Complementado por
Referenciado por
Licencia Creative Commons
Excepto donde se indique lo contrario, la licencia de este ítem se describe como by-nc-sa
