Title: Evidence from a horse-race on the top of intra-daily forecasting models for algorithmic trading
Authors: Beatrice Sagna - Universite Paris Dauphine PSL (France) [presenting]
Abstract: For every trader operating within stocks markets, a good prediction of intra-daily volume is a big concern. The better this prediction is, the better his (her) financial performances will be. In the literature, not much attention has been dedicated on predicting intra-daily volumes. The core is to propose a cartography of four main intra-daily forecasting models for volume existing in the literature based on four models. The contribution is to challenge these models in a horse race by replicating them considering two main criteria: accuracy and speed execution. The estimations provide evidence that models that take advantage of cross-sectional variations give faster estimations and more accurate predictions of intra-daily volumes than models that only consider time series variations. We discuss econometric insights and microstructure phenomenons to support such results.