How do we develop a way of assessing forecasting performance that is responsive to results, but not completely driven by accurately forecasting a particular event?
As forecasters how do we avoid being judged based only on our last forecast's prediction?