Given the shape of the curve modeling the DJIA over the last several decades, it makes sense to use an exponential growth model for predictive purposes. Plug the year in for x to get the DJIA value.
y = (2 Ã— 10^(-101))e^(0.1206x)
Read more: https://www.letsrun.com/forum/flat_read.php?thread=5369837&page=737#ixzz4N50pv9qh
No this was the original question, and as I pointed out a weak model. Hussman's models uses more data, larger time frame, and has 93% predictability. I wouldn't say your's is totally worthless, but it can at least serve as a bad example. Last several decades, oh, is that two, perhaps three. OK, use 1930-1960?