Predicting spot exchange rates in a nonlinear estimation framework using futures prices

One-step Prediction of Financial Time Series

“…On the basis of the study, he concludes that the outperformance of the nearest neighbour method is not statistically significant compared with the forecast from a random walk model. A similar conclusion is reached by Parhizgari and De Boyrie (1997) when comparing the outof-sample performance of various forecasting models against a random walk model on exchange rate data. The main motivation for the present study is the observation that most studies reported in the literature use mean square forecast error as the criterion to compare the forecasting performance of various models.…”

Mean reversion in stock index futures markets: A nonlinear analysis

“…Section 5 reports the results of summary statistics and univariate unit root tests applied to basis data, cointegration tests applied to a regression involving the spot price and the futures price, linearity tests applied to the basis data, and the estimation results from employing nonlinear models to characterize the basis of the S&P 500 and the FTSE 100 indices. In Section 6 Monte Carlo integration methods are used to calculate the half-lives implied by estimated nonlinear models for the basis, further examining how the nonlinear estimation results can improve the profession's understanding of the dynamics 2 The literature related to the present study is very large and includes, among others, Blank (1991), Brennan and Schwartz (1990), Chan (1992), Dwyer et al (1996), Figlewski (1984), Mougoue (1997a, 1997b), Gao and Wang (1999), Kawaller (1991), Kawaller, Koch, and Koch (1987), Klemkosky and Lee (1991), Lekkos and Milas (2001), Ramaswamy (1988), Miller, Muthuswamy, andWhaley (1994), Modest and Sundaresan (1983), Parhizgari and de Boyrie (1997), Sarno and Valente (2000), Stoll and Whaley (1990), Brorsen (1993, 1994), Yadav et al (1994). 3 See, e.g., Dumas (1994) and Sofianos (1993) on this point.…”