Let me give you a simple example of why the assumptions matter in a model though almost the only thing people pay attention to is the headline finding. Let's assume I run the following model: I + S = P where P = -10. I then tell you -10 is the number of pounds I forecast 1/n

Losing during corona quarantine. You report this around the world as Balding forecast to lose 10 pounds during corona quarantine. That's the only thing people pay attention to and the only questions that get asked. Is this a good forecast? Based upon the information provided 2/n

You can't answer the question if it is a good and reasonable forecast. What if I tell you now that I=ice cream and S=salad. Now we are getting some place. We are actually looking at what I might eat. Now what if I told you I set the ice cream parameter to 1 because I promise 3/n

To lose weight and salad to m-1 where m is all meals in quarantine except the one time I have ice cream. All of a sudden it makes sense why my forecast produces a 10 pound weight loss but it raises questions about three models assumptions. I mean ice cream only once 4/n

During the entire corona quarantine? That's just not realistic. The forecast is only as good as the assumptions. Anyone can make a forecast. It is vital trust we look at the assumptions trust make up there model to see how accurate the assumptions are that make up the model.