Yes its a bit insidious as you wouln't know as the backtest may look ok, that what is scary about it, I guess that's why you should always test on a demo first. I only became wary of this when I bought an
EA then tested and noticed:
1) The backtests could not match the data on the website
2) The stop &
profit targets were different on different sets of data at different time periods
3) The backtest settings could be changed but made no difference to the backtest results (ie the actual settings were hard coded internally)
4) The vendor promises regular 'updates' ie newly optimised data
Unfortunately only those with access to the underlying code will ever know.
Buyer beware!!!!!