Backtesting does not predict the future, nothing does, however...
The market stays consistant enough in its patterns and actions to allow us to test a theory on backdata, and have a reasonable idea that it will work in the future. Unless the market makes a drastic change overnight. Most changes are slow enough.
Lets face it pivots still work, Fibbonacci still works, historic Highs and Lows still work, S & R still works. EMA crossovers still work and this seperation trade still works. So If I backtested variations of any of those it should be reliable enough. Unless, S & R and trends stop happening tomorrow and all we ever see is inside candles forever more.
Maybe the reason why the "backtesting does not work" mantra is so strong, is most don't really understand the point of backtesting, it's limitations and the advantages to your trading. The reason for this is most people have the wrong focus and are obsessed with pips rather than profitability.