APPLYING THE PRINCIPLES OF FORECASTING TO DANIEL MURPHY. I arrived at my forecast about Daniel Murphy by giving weight to his OPS for the last 28 days of the regular season. It was 0.854. I then reduced that figure by .05 to give more weight to his career average OPS 0f 0.755 and his OPS for 2015 of 0.770. I deliberately (and whimsically) chose to have a prediction that is not divisible by 5 because Tetlock thinks that forecasters who have a more “granular” approach are more accurate.
Tetlock stresses the importance of having a suitable question. This question has a reasonable time frame. Tetlock has come to believe that forecasting for more than one year is not useful. Angela Chen cites Tetlock’s conclusion that the most important thing in improving forecasting is to reward yourself only for the accuracy of the prediction and not the reasoning process. The answer here will be precise. The question involves assessing data from different time periods just as many of Tetlock’s questions do. The Murphy question differs from Tetlock’s questions in that it does not permit updating the forecast as more information comes in.
It occurs to me that the rotisserie baseball I have been playing for 35 years is quite similar to answering the questions in Tetlock’s forecasting competitions.