Thanks Baco65,

I'm still wondering how one might apply this philosophy or execute it when dealing with statistics.

Isn't it simply another way of working out the expected price given the strike rate and the RF variation is the profit?

Example:

Favourites
Expected frequency 30% or 33.33
Needed average dividend to break even $3.33

But if the favourites you pick have a lower or higher strike rate, you recalculate the needed dividend.

If you can get 50% win strike rate, you need $2.00 average dividend.

If the actual strike rate goes up or down this is the relative frequency variation...or am I mistaken?

Perhaps the best thing I found in my search was this statement...

"As the number of trials of an experiment is increased,then the long-term relative frequency becomes a better estimate of the probability."

Simply put more data, more confidence and greater likelihood of future success. Something I've lived by for the last couple of years.