Why past results genuinely do not predict future ones

The phrase is a legal requirement and it is also true. Outcomes carry enormous variance around any real level of skill, so a past figure is a weak estimate of a future one.

A 60 per cent month at a true 54 per cent rate is entirely ordinary and will not repeat on demand. The published figure captures both the skill and the noise, and the noise is usually the larger term.

It does not mean past results are worthless. Over sufficient volume they are the only evidence there is, which is why the count matters more than the percentage.

And that is the trap. The warning appears so universally that it has stopped being read, which is the standard failure mode of every mandated disclosure.

What to do with it. Read past results as evidence about process, weighted by sample size, and never as a forecast of the next period.