Dynamic Hedge Fund Style Analysis with Errors-in-Variables

This paper revisits the traditional return-based style analysis (RBSA) in presence of time-varying exposures and errors-in-variables (EIV). We first apply a selection algorithm using the Kalman filter to identify the more appropriate benchmarks for the analysed fund return. Then, we compute their corresponding higher moment estimated errors-in-variables, i.e. the measurement error series introducing the (cross) moments of order three and four. We adjust the selected benchmarks by subtracting their higher moments estimated EIV from the initial return series, to obtain an estimate of the true uncontaminated benchmarks. We finally run the Kalman filter on these adjusted regressors. Analysing EDHEC alternative indexes styles, we show that this technique improves the factor loadings and permits to identify more precisely the return sources of the considered hedge fund strategy