In a way, the burden of proof's the other way round, and we should ask "What was right with the system", which highlights the answer, in this case, as it was "nothing".
Specifically -
It had no underlying basis in reality - nothing to do with market structure or movements or behavior.
It was using an indicator for trade entries (many have good uses, but trade entries isn't one of them).
It wasn't a good indicator anyway.
It wasn't using the indicator for the purpose for which it was designed.
But most important of all, it was based (as most forum thread "systems" using Bollinger bands are, on a fundamental misunderstanding of the concept of mean reversion. So it never had a chance.
1