What it proves
That the code runs unattended, that signals arrive when the backtest says they should, that fills at live prices resemble the simulated ones, and that the drawdown is one you can sit through. It does not prove profitability; it adds weeks of unseen data.