Quote:
Originally Posted by SiberianPIMP
I mean that we are maybe calculating probabilities wrong. The numbers that we get with no. of combinations/all combinations are maybe not exact number.
As a physical matter, it's possible that coins are not perfectly balanced for flipping. Depending on the methodology used, it's possible to bias results slightly depending on the way that the coin is manufactured. For example:
* The edges of coins are very slightly beveled so that they release from the mold properly, which slightly biases them if they spin on a table before settling.
* The weight of each face of the coin may vary slightly, causing bias as the coin bounces/spins before settling.
* There may be physical defects for any specific coin that is being flipped that may cause bias.
But none of these physical features will be captured in a simulation. There have been various studies of this bias that you can look up for yourself. These are all dependent upon the specific coin itself.
However, the larger point is that the math is still precise in the sense that once the bias is known (say, 50.5% heads/49.5% tails), the math can be worked out. And the bottom line is that if you have a coin that's biased towards heads, then if the payout is even money you gain an EV advantage by just betting heads ALL THE TIME. All the fancy waiting to see X tosses before making a wager has ZERO impact on the outcome as long as you believe that the coin flips are truly independent of each other (by definition of what independent events are).
All of this is understood under the existing framework of probability. As much as you may want to believe you have some super-clever insight that completely destroys everything we know about probability, the reality is that it's highly unlikely (especially given the depth of knowledge we have in this area) and that it's far more likely that whatever "system" you think you come up with is either wrong or easily explained in the existing knowledge framework.
Edit:
Quote:
If you run a simulation like descbribed in this post(for a billion trials)? You allways get a number that is lower than 37,5%, right? Or no?
No. The fact that you think two consecutive results that are just below 37.5% (on the order of 0.001%) is a pattern that can be extrapolated is a strong sign that your basic understanding of probability is flawed. There was about a 25% chance of that happening.