running lions
In my eyes these "inexplicable reason" can be traced back to a general problem with simulation programs and that they are taken to holy. If they use for example random numbers, you must not forget that these numbers are not really random, they are only distributed as if they where random. Are you really sure, that this makes no difference?
It is easy to misinterpret the results of computer simulation software. I feel remind to the great "butterfly effect", that in realety does not exists: no butterfly has ever been responsable for a windstorm. It is only, that weather forecast simulation programs contains recursive structures, and if you would add such a little parameter as a butterfly, then the result of that program can be totally different. This does not mean that a butterfly makes a difference, this means only that the weather forcast simulation program is simply imperfect. These programs are still not able to handle such big parameters as e.g. the sunwind correct. And remember all the billions of parameters that are surely bigger than butterflys (for example running lions). And IMHO if you want to add a parameter for one butterfly, what is with all the billion others. And if it would be possible to add all these myriads of parameters (just imagine), then I am deeply convinced that butterflys would regain their innocence.
I do not want to compare random based Black Jack simulation programs with chaos theory based simulation programs. I do want to sensibilyze you not to suprainterpret results of simulation programs. E.g. if they assume perfect players, they make a big error. Perfect players are less probable than that the dealer would not shuffle.
It has much more influence on your result if you just believe that your Box is in favor to the bank, but that's another theme...