There is a method some call ‘Monte Carlo’ —
where you would research systems like this with a substantial amount of data left out at random.
If the question is, ‘how will this system stand against the tests of time?’, this is a fair way of gauging that. Not perfect, but there is no perfect in gambling…
As for bur char’s statement about it not standing up against the tests of time. I’m not sure about that; however, a healthy grasp of ‘regression to the mean’ is good to keep in your back pocket. Will 68% success rate hold up even with simple parameters? Usually not with the same degree of success. Especially unlikely if adding a parameter so obscure (no exactly 2 run wins after a Tuesday is like saying, ‘ The Patriots are 10-0 when my dog pooped 3 times on a wednesday after 5 pm in the front yard’