No sweat. use some common sense and get each gang of bots some separated data, a gang for NGDP, one for unemployment, some for the markets. Moedl the yield curve as a bunch of bonds sold at dollar price, leave the bots to compute the yield curve. But get all your separate data botified then gang these together with buffer bots, leading to the mother bot sitting near the abyss. Try to get a good series of city data. But put it all together as sequential bets in MacroBits money. Your model becomes a series of finite polynomials, describing all the probability functions with MacroBits as the finite line.
Pretty sure Atlanta GDP Now does something like this, Goldman Sachs should have one. All we are doing is Ito's calculus, by adapting in real time. The better version of a Huffman encoder/decoder. When they say 'activities indexing' model, they generally will be doing maximum likelihood, finite probability distributions.
No comments:
Post a Comment