A roulette wheel is a device of physical construct and as such, despite all the precision engineering in the world, a bias (however small) will exist. The wheel doesn’t turn on a perfectly horizontal plane, the pockets or frets aren’t identically sized, nor are the diamonds for that matter.
It was only twenty odd years ago that Billy Walters famously took Atlantic City’s Lucky Nugget for $4 million after working out its old wheels had a significant bias. Of course the Nugget and casinos around the world were very quick to upgrade to new super-duper low profile wheels designed by George Melas to put an end to this type of advantage play.
So while no wheel is perfect, these days at any casino worth it’s salt, including live roulette operations, we’re talking about imperfections so infinitesimal that they hardly bear mention…or do they.
You see there is a school of thought amongst certain roulette advantage players that despite today’s precision wheels, there is always that exception to the rule, and if it can be found, there’s good money to be made provided the bias is significant enough to compensate for the house edge that should exist (were that wheel perfect). Detecting roulette wheel bias is something some guys spend a lot of time doing. And the ways they have devised to do it are either very interesting, or totally incomprehensible depending on your grasp of mathematics.
It’s not possible go in to the casino with high tech measuring tools and ask for a few days to test their table – for some reason casinos aren’t up for this, so advantage players are left to use their powers of observation and mathematics to determine whether they are onto a biased wheel. And we’re not talking a few spins here. Depending on who you talk to to an analysis based on anything less than 3,000 spins is meaningless. Upwards of 10,000 starts to become more robust.
The broad idea is that if you can identify significant deviation from the statistically expected distribution of outcomes, and if this deviation is skewed to a sector of the wheel, then bingo – there’s an opportunity.
Now how this deviation is quantified and determined to be worthy of advantage play is all a bit too complicated for me and involves statistical methods like chi-squared distribution and confidence interval analysis…none of which I feel remotely qualified to talk about. But the clear lesson to be taken here is that some very smart folks, believe not only that biased wheels exist, but importantly, that techniques exist to identify them.
I just wished I’d paid more attention in maths class!