I read the broker emulator fill choice as the deciding point before a ported result is compared with TradingView. Your 9 and 21 EMA cross shows how identical entry lines can still fill differently. How do you handle entries when the isconfirmed filter removed the live signal but the historical run has no realtime bar?
Good question — it's the same gap the post flags with the isconfirmed variant: the name converts cleanly, but in a historical Python run every bar is already closed, so the filter passes on bars where the live signal may have vanished before the close. I don't handle it automatically — no translator can, because there's no realtime bar offline. What I do is treat isconfirmed-gated entries as close-only and check them against TradingView's historical run; if a strategy genuinely behaves differently live, that difference has to be re-engineered in the feed, not in the translation.
Arnold Holm
I read the broker emulator fill choice as the deciding point before a ported result is compared with TradingView. Your 9 and 21 EMA cross shows how identical entry lines can still fill differently. How do you handle entries when the isconfirmed filter removed the live signal but the historical run has no realtime bar?