Thanks RLogic for your excellent and helpful summary.
My own work 15-20 years ago was built on two professional mentors MS and BC who dramatically shifted my views and also the influence of the professional player Dr Martin Blakey with his money management, enthusiasm and lifestyle. After that through empirical testing I devised many bets and some I will share with you here. Some such as WF have been outlined here before, and others such as the D+C and Loss Bet do relate to the context above in this thread.
My own private bet exploded into reality when I realised it worked flat staking, and that there was no need to progress. In practical play however I tend to step bet values in the context of favourable/ unfavourable trending.
In reference to RLogic's most helpful post, and thanks to CS1968 ( since re-named) and his extraordinary progress, I do observe from experience through the long practice of the 'thought pattern' (ie way of thinking - tilt ) that clusters in varying formats have a bell curve of distribution, and to observe the centreline of the that bell curve can tell you when same can switch to change (ie run to switch).
Note in the model matrix proposed by CS1968 a theory of common duration extent may be 4 spin outcomes, or 5 outcomes as trigger to switch to change.
That matrix works with D+C , dozens/ colours and may overlap with EC bets. There are 9 movements or variable forms with all containing 4 inside numbers. Zero is set aside. My view on this is to play Zero as an independent single number target that transcends shorter games and which game may have a life of its own within but most probably without other sessions in this matrix form. Because it is a game targeting only one number, and given a suitable bank, it is quite a good bet. The goal would simply be to break even. Like wise with 00 as well as 0- the two could be played in a sort of parallel universe zero game.
My observations regarding the switch bet ( that gives great returns on only 4 target numbers with this CS1968 matrix ) are based on initial views only but a sort of 'idea level' operation. In real life most such ideas come to nought, but in this case the research into this might be time very well spent. It is sad that in some forums I note some world weariness where some writers who have been around too long seem to lament that the enthusiasm and naivety of research and writing 5-10 years ago is no longer with us, as if we are now all wise to the impossibility to beat the game, or that anyone who dares to entertain such thoughts must be a scammer. Amusingly some have accused me of such despite having given away so much 'free' information - must be a long scam then. This is nonsense of course because now more than ever roulette opens many parallel paths into winning bets. Far more than the 4 or 5 Simon and CEH mentioned. Always it is important to note the fraudsters may never have found a true bet, but just raked in all the offerings from those on the trail. They may have had the signs and some principles but it is pleasing to note that there are some of us who have truly developed worthy bets despite being de-frauded or let down many times by time wasters or lazy individuals who want it all spoon fed. Yes I could name about a dozen and some are still readers and regular visitors to this forum.
In contrast with the CS1968 switch bet noted above ( my terminology) which targets 4 numbers, my own bet targets 9 numbers, but because it overlays several sets or matrices, it can produce a target every spin if sufficient sets are monitored. If playing one set, then a target is set every 7 spins or so.
In my own work which has 4 variables abcd as opposed to the CS1968 bet which has 3 variables abc, the duration of a cluster runs from 1 spin** to a maximum once observed of 20 spins with a median at about 4-5 spins. Thus we know change will follow soon after. On that basis a technique can be built.
**formation of a cluster by definition may take 2-3 spins and the term 1 spin is applied after that formation phase
The technique, then the strategy is formed around the bet characteristic performance.
It is not hard once you get the 'thought pattern'.
By the way, the bet efficiency of a 4 number target is superior to that of a 9 number target, but another variable is the consistency of the bet which needs to be measured over a very large sample size, or better, multiple samples, at least 10,000 spins per sample.