Skip to main content

Table 9 Out-of-Sample Forecasting Analysis

From: Timing the market: the economic value of price extremes

  Panel A. Monthly Index Data Panel B. Quarterly Index Data
Out-of-Sample 1971.01-2015.12 1989.01-2015.12 1996.01-2015.12 1971Q1-2015Q4 1989Q1-2015Q4 1996Q1-2015Q4
R2oos (%) 1.07* 0.098 1.46 3.87** 0.97* 2.14*
CER (%) 0.378 1.03 2.41 2.95 4.20 4.85
SR p 0.059 0.076 0.099 0.201 0.227 0.226
SR bh 0.037 0.085 0.068 0.060 0.138 0.109
  1. Note. ***, **, * means respectively statistical significance at the level of 1%, 5% and 10%. The statistical significance of Roos2 is evaluated with the MSPE-adjusted statistic (Clark and West, 2007). The MSPE-adjusted statistic is conveniently calculated by first defining ft+1 = (rt+1-rmt+1)2-[(rt+1-rtp+1)2-(rmt+1-rtp+1)2].
  2. By regressing ft+1 on a constant and calculating the t-statistic corresponding to the constant, a p-value for a one-sided (upper-tail) test is obtained with the standard normal distribution