# Why BSTS model gives very different forecasts depending on length of testing set?

When I run a BSTS model in R, I get a finished model from a data set, and when I use that model to predict, for example, like this:

predict(model3,newdata=futuredata) -> pred_1

I get a set of predictions from pred_1\$mean

However, when I do this:

predict(model3,newdata=futuredata[1,]) -> pred_1_first

in order to get just the first period prediction for the future, the number I get is vastly different from the first number I get in pred_1

Since the newdata is the same data, just truncated, the prediction should be the same number, why different?????

hippie