Your backtest looks good.
That is exactly what makes this the dangerous part.
The equity curve is clean. The drawdown looks survivable. The logic still makes sense when you read it back the next morning.
So you fund the account, you let it run, and you watch it bleed in a way the backtest never did.
Here is the part that surprises people: almost none of that is the strategy being wrong.
It is the gap between a strategy that works on historical data and a system that works when a computer runs it against a live venue with real money. That gap has a small number of specific, findable failure points. Every one of them can be checked before you fund anything.
That is all this is. Seven checks, in order, run before your bot touches real money.
What you'll be able to do after 10 minutes with it:
Tell in about two minutes whether your edge survives contact with the spread, because there is a specific signature in the live equity curve that separates an execution-cost problem from a broken signal.
Stop your account from being most fragile at the exact moment it has already lost money, by fixing the position sizing rule that quietly does the opposite of what most people think it does.
Know what your bot does when the connection drops mid-position, before the market is the thing that tells you.
Find out whether your out-of-sample test was ever actually out of sample, which is the failure that is invisible from the inside and survives every other check on this list.
Fund your first live run at a size you can watch go to zero without it changing anything about your life.