I think I've got my head around the reasoning on this one, but it's not what I inferred from what was originally written around "soaking up spins".

If we take an example of a single number straight up, then there's roughly a 95% chance of it showing within 145 spins - this is two standard deviations south from the expectation of a little less than four shows, based on 1/37. So if my "luck" was running at -2SD, I'd be betting (and losing) up to 145 units before I got paid out 35+1? Covering additional numbers, giving a greater probability of hitting, will (hopefully) produce returns that will offset this anticipated loss? Am I on the right track?

Problem, as I see it, is that the additional numbers wagered may also be subject to negative variance and lose successively too, in which case adding more numbers to offset potential losses doesn't achieve it's aim and is just exposing more money to the house edge?

Also, on the subject of "edge" I think there may be some misunderstanding on my part with the terminology used here. Covering more numbers certainly provides the benefit of an increased probability of hitting and reducing the risk of losing all on any one spin, but doing so won't affect the house edge - the number of slots on the wheel and the pay table doesn't alter. As dobblesteen has pointed out - applying any "system" will never provide an edge over the house.

Playing around with methods that hit a balance between risk and potential return is a conundrum that has been around for as long as the wheel of doom itself. I spend more time than is decent thinking about it - as a means of paying less for my entertainment.