so yeah, looks like this idea of a roll of a dice is potentially a no dice. it does make you wonder tho if there are calculations that would allow you to review random distribution in any way. my original thinking tbh was how this would stack up if used against a BIG dataset and fed into a machine learning model. you'd use the trap wins (as the features) in concatenated sequences (123,236,366,666,662) etc where each group was composed of the last 3 results. using this model, you'd input the last TWO resuts (concateneated - i.e. 12, 23, 36 etc) and it would use the model to attempt to predict the 3rd digit. (in my view, this would predict a potential trap to avoid in terms of a lay bet, rather than trying to pop the winner).
as i said, i set this thread up for a bit of fun and am getting exactly the feedback i'd hoped for in terms of proof that randomness can come in many forms, therefore some sort of pattern recognition might be a fun approach.
anywayz ....
) model against apparently unconnected data events. in my other life, we use ML in an application that reviews all the tv advertising across numerous channels in an attempt to identify advertisers misrepresenting the demograph that they have pitched their campaign against (in tv world, advertising is sold at varying rates, depending on the demograph that you are targetting, therefore ABC1 adults is theoretically the most expensive target, so some advertisers position their campaign to capture that audience without actually paying the full cost).
