This is how I would verify your results in figure 1: For the DJF season 2004- 2005, the forecast approximately states that the Western third of the continental United States will = +1, the Central third will = 0, and the Eastern third will = -1. Unfortunately the observed results indicate the Western third of the continental United States = +1, the Central third = +1, and the Eastern third = +1. That’s a Win, a Loss, followed by a Double Loss, leaving a net residual error equivalent to two thirds of the continental United States. Now the next year 2005 -2006 you do a little better, getting around two thirds of the continental United States correct. While this might seem like an achievement, after two seasons, the net skill of your model in forecasting DJF temps is close to ZERO.