Wrong.
You are mushing together words "equal chance" with other words like "equal likelihood" in convoluting the conversation.
Statistics, as a process, is rigid. It will produce an outcome directly relates to your input information.
You are evaluating a system for bias, but instead of opening the analysis to multiple variables in order to determine their interaction within the model, you say "hey, 50/50 (or some other proportional metric) results happen X number of times after Y number of trials" then compare what we have currently against trials conducted assuming equal inputs.
That is fundamentally flawed as a method to evaluate causality. In fact, it may be step 1 or 2 in 20 steps, many with several iterations, in discerning causality with any reliability.