Just tested a new system, 59 spins each, not a huge amount, my goal was to win or lose $1000. This is the exact system tested on both RX , RS and with Real money. Three very different results.
With real money -$567.
Rx +$1007
and RS -$969
Drastic difference. I would say a bit of variance since only 59 spins. But results are too drastic in opposite directions for that.
Here are the charts for the RX and RS. The real money one I didn't chart it, Once it started going south more than 20% I don't bother.
(link:://:.pichost.org/images/2018/09/21/sourceb8f49.png) (link:://:.pichost.org/image/th33K)
(link:://:.pichost.org/images/2018/09/21/source68821.png) (link:://:.pichost.org/image/thA7B)
Remember if testing , test as many different methods as you can. If they work on all they will work with Real money too.
This is down to the quality of the data
You will find this same issue in tests with financial markets
You have to pay for clean data
For gambling games, it's important to do the tests based on a wide array of sources
I recommend to do the test with data from the wheels you play, this is the final phase of the test exercise after the initial broader test has already confirmed the result you want
Such tests are required for statistics based methods only