Separate a series frequency from model probability
A short series need not reproduce its stated probability exactly.
Observed relative frequency is the target outcome's share in a recorded series. The next outcome's probability depends on the assumed model. Separate assumptions from observations: a few tosses neither establish a real coin's fairness nor require later compensation.
What to do
- Record every outcome and specify what you count; do not change the denominator after counting.
- Write the model's assumptions separately, including coin fairness and independent tosses.
- Label the observed share and model probability, without promising a required number of heads next time.
Example
Fictional series H,H,T,H, with H=heads and T=tails: heads occur 3/4=0.75. Under the stipulated independent fair-coin model, next-head probability is 1/2=0.5. The model does not require exactly two heads in every four tosses.
My Day · Windows
My Day can store the series, assumptions and conclusion. It does not perform statistical verification for you.