GAT Performance with increasing RF

[kweg7]

Hi Anastasios. Suppose I run the GAT engine in relation to the Australian Saturday Lotto game and select a good GAT which shows almost 60% for 3 plus numbers. I take this to show that this GAT will, on average, give me 3 plus numbers each second or third DRAW. (Australian lotto pays around $10 for 3 drawn numbers). Will the performance of this particular GAT be maintained if I use it for say the following 8 or 9 draws increasing the RF each time. In other words, is there any limit to how high the RF number can go before the 60% result ( in my example above) no longer holds. Can you use such a GAT over say 6 months despite the RF becoming higher and higher. Thank you.

[lottoarchitect]

Think of this like that, if this 60% has been obtained by the use of a lengthy test range and the hits are spread out throughout the test range then by all means keep using RF for as long as it keeps delivering. If however this 60% was generated using a small test range like 5-10 draws, then there isn't much history used to evaluate performance. Generally the advice however is to keep using a particular GAT for the next 100/X draws (the 100/X rule mentioned in these forums). This means, if your X=60%, you should keep using that GAT for at least 100/60 = about 2-3 draws or a bit more. If it doesn't deliver during that time, it may have gone cold. The reason is primarily GAT attempts to find "local" behavior in the draws which means a set of a few consecutive draws can share something in common which GAT tries to find and use, and the predictions offered using RF rely on whatever has been detected. Under that premise, after a while these "dynamics" fade away, thus insisting on a GAT that didn't delvier over the 100/X plus a bit more might not be the best course of action. Still, you can use RF for as long as you want. If you plan to go that route, try to use a smaller stat. data value to minimize averaging effects thus every proposed draw is mostly affected by that "local" behavior. I can't tell if you will get something better doing so however.

[kweg7]

Thanks Anastasios.