Testing is necessary over an extended period and with data from a wide variety of sources to avoid the dangers of reverse engineering and self delusion. CEH was right in suggesting at least 30 samples of 100 spin tests with a positive outcome in at least 28 of those samples as a first priority. Of course 3000 spins is still a small sample, and in my view there should be 10 sets of 30 samples as an ideal benchmark with sourcing from ten different live casinos. I do not consider randomly generated data suitable and would rather I personally sourced the live spin data.
In my own library I have data from Ritz London, Sky Casino Auckland, Christchurch Casino, Hamburg, Wiesbaden, Baden Baden, Sydney Star City, Jupiters Casino Queensland, Dublin Fitzwilliam, Queenstown Sky City NZ.
In recent observation of the remarkable work in VR8 and Hexagon I note the wisdom of being able to read/ respond to outcomes sufficiently quickly to shut down live play when a trend pattern is broken and while a new trend is forming live play should be halted. This can be done on internet live casinos where very smart programs can assist the player, but in live play I am interested only in play without any technical assistance. Therefore it is essential to add to play efficiency that corrective phases are mitigated. Such player discretion is improved with experience and it may be that earlier tests of a potential CWB must be sufficiently powerful to handle the correction waves and the 100 spin samples will encounter such from time to time, and therefore must be able to deal with such. It is anticipated therefore that in real time play results can be further improved.
I hope to be able to publish some results soon of my own work in one such field.