Since I take a statistical approach to sports betting it's only natural that I should do the same for roulette. In my limited experience of playing roulette and reading roulette forums I've formed the opinion that there is no simple selection process which will consistently win. Not surprising really because individually none of them has an edge unless you're using some kind system based on the physical conditions, but for a purely statistical approach I've found that the following works pretty well. It uses the idea of 'hot' numbers and applies to any bet on the table but the higher odds bets will require a lot more work, so my example uses the even chances red & black. The step by step procedure is

1) Create a number of simple selections based on patterns or other criteria. In this example I've used 6 selections.

2) Generate a list of wins and losses for each selection from the red & black outcomes. This list should be 'rolling' so after a maximum number of outcomes (I used 27 in this example) the first element should be removed.

3) For each selection process calculate the percentage of wins and update this on every spin.

4) For each selecton process give the next bet for that process, so for red & black there will be only 2.

5) To make a bet, look for the majority of predictions being on one outcome and calculate the average percentage win. If it is higher than the average percentage win for the minority predictions, make the bet.

So on this spin 4 of the 6 selections are predicting black, and the average win percentage is (0.51 + 0.66 + 0.66 + 0.77)/4 = 0.65. This is obviously higher than the average win percentage of red, so the bet is on black. If there are an equal number of predictions on red & black - this would be 3 of each here -then no bet is made.

My findings have been that this 'system' results in fewer losses and less extreme deviations than anything else I've tried. Much of the time it wins using level stakes. It's really more of a framework than a system because it can be applied to any bet.