There is no broker connection, and there is not going to be one that trades for you. What matters instead is understanding where your broker will differ from the test, because that difference is the whole gap between a backtest and a result.
Spread
We test on one side of the market — the bid — and add a fixed spread per instrument on top. So the cost is applied, but the number is ours, not the one your broker quoted at that second. A real spread widens around news and at the session roll; the modelled one does not, which means a strategy that trades through those moments is being tested kindly. This is the largest single gap between a test here and a live account, and it is why every result is also shown at 1.5 times those costs.
Slippage and requotes
We apply a realistic delay on every trade, but we do not model a requote or a partial fill on large size, because we have no order book depth to model them from. A strategy that trades small and infrequently is barely affected. One that trades often, or in size, is.
How to use the stressed number
It is there for exactly this. The 1.5× run is a deliberate answer to "what if my execution is worse than yours" — if the strategy still works at those costs, the gap between our feed and your broker is unlikely to be what breaks it.