3 forecast experts, need to fire 1.
Look at the predicted forecast for the last 9 years, compute the square error for each expert.
Compute the pair-wise correlation matrix between experts. Found out the most uncorrelated pair is brad and Shahrukh, thus giving clue about reducing risk through diversification. Looked at the chart to find out that maybe the two combined forecast may lead to a more accurate forecast through regression.
Compute the regression between forecast and true values for each expert and compute the sum of square error.
Compute the regression between paired forecasts and true values for pair of two experts and compute the sum of square error.
The more accurate (by far) forecast is obtained with the pair of forecast of Brad and Shahrukh, thus we should fire Jackie Chan, as suggested by the correlaion matrix.