Can they not just back test it using the old and constantly updated data; almost everything has a probability but probability is useless. Statistics are function of probability but not probability but rather an Expectation function. Statistics are good for generalization but not necessarily fantastic for a single sample.
Nevertheless statistic are typically the only option that there is to predict like polls so such predictions are statistical. The best that can be done is to back test the old data constantly and I guess they have to decide what the boundaries will be, lax or stiff.. The algo has to be constantly back tested to see if it still relevant and has acceptable risk or be changed otherwise.
1 comment
[ 0.21 ms ] story [ 17.2 ms ] threadNevertheless statistic are typically the only option that there is to predict like polls so such predictions are statistical. The best that can be done is to back test the old data constantly and I guess they have to decide what the boundaries will be, lax or stiff.. The algo has to be constantly back tested to see if it still relevant and has acceptable risk or be changed otherwise.