I did not pass the Aquafunded evaluation on skill. I passed it because a position fell through a gap in my own logic during a day-transition glitch, and I was too much of a perfectionist to notice for four months. Mark Twain once said he’d been through some terrible things in his life, some of which actually happened — and this story sits right on that line, because it happened, it went in my favor, and I am under no illusion that it was earned.
Every trader has a version of this story they don’t tell at meetups, because the version they tell is the one where the edge worked and the discipline held and the equity curve looks like something you’d frame. Mine has a bug in it. Here’s the honest version of what happened, warts, rollover gaps, and all.
The Setup
My strategy was swing trading — hourly timeframe, a script I’d built myself on TradingView, nothing exotic. I had a TradingView webhook to MatchTrader running through AlgoWay, closed the laptop, and mostly stopped watching it the way you’re supposed to with swing setups. It was a fairly ordinary TradingView webhook automation setup: the script made the decision, the alert carried it, and the execution platform was supposed to do the rest. That part was intentional. What happened next was not.
The Bug
At some point during a daily rollover, MatchTrader closed for the day-transition window. TradingView, of course, has no idea a broker platform just went dark for maintenance — it doesn’t pause, it doesn’t wait politely, it just keeps stamping candles and firing whatever alerts are due to fire. So a flat/close command went out into a platform that wasn’t there to receive it. The close never landed.
I run positions through AlgoWay in hedge mode, which checks position sizing before it opens or closes anything, matched against a tv_order_id parameter in the JSON payload so it knows which position a given signal is actually supposed to act on. Because the original close attempt fired into the rollover gap and never resolved, this particular position fell outside that matching logic entirely — it was technically still open, but no longer something my usual flow was tracking or trying to close automatically.
That is one of the less glamorous realities of webhook trading automation: the logic can be perfectly clear on the sending side while the receiving platform is simply unavailable. The command existed. The trade existed. The connection between the two disappeared for a few minutes, which was enough.
And here’s the part that’s genuinely on me: I don’t use closeall to clean up loose ends. I’m the kind of trader who wants to know exactly why a position is being closed, not sweep it shut with everything else, the way a man who irons his own shirts doesn’t take kindly to a dry cleaner steaming the wrinkles out with an industrial press. So instead of nuking the account back to a known state, I just… didn’t notice. For four months. There is a particular kind of confidence that comes from being meticulous about ninety-nine things and having the hundredth one quietly wander off while you weren’t looking, and I recommend it to no one, though I can’t say it didn’t work out.
What the Market Did While I Wasn’t Looking
This is the part that still doesn’t feel real when I type it out. During roughly the window this position went untracked, the market went through an actual phase shift — a real trend change on the 4-hour chart, not a fakeout. Somewhere out there, presumably, a very disciplined trader closed a similar position right on schedule, exactly as the textbook says, and missed the whole move because the textbook was written before this particular rollover gap existed. I did not close mine on schedule. I did not close mine at all, on account of not knowing it was open.
The position I’d forgotten about was sitting on the right side of that shift the entire time, patiently compounding without me, the way houseplants do when you go on vacation and come back surprised they didn’t die. I found it, still open, still running, and it had turned into just under $2,000 in unrealized profit.
I closed it manually, the same way I close everything — deliberately, one position at a time. That single trade, combined with the rest of my open positions, was enough to carry the account through the evaluation phase.
What I Actually Learned From This
I want to be careful here, because the honest lesson is not “leave your bugs alone and hope.” A silent failed close during a rollover gap is a real problem, and in a different market phase it would have just as easily been a $2,000 loss sitting there instead of a gain — the bug didn’t know which direction the market was going to move, it just happened to be on my side that time.
It also changed the way I thought about TradingView alert automation. Sending the alert is only half the job. If automated execution depends on a broker platform being reachable at a particular moment, then the trading window itself becomes part of the strategy whether you planned for it or not.
Since then I’ve gone looking at the platform-level controls for exactly this scenario: blocking trading during the day-transition window, padding a few minutes before and after daily rollover so nothing fires into a dead platform, and being deliberate about session filters instead of leaving everything on “all time” by default. Those settings exist precisely because rollover windows are where broker platforms and always-on charting tools like TradingView stop agreeing with each other.
The uncomfortable truth is that my own habit — refusing to closeall and always closing positions individually, on purpose — is what let this ride long enough to matter. That habit is usually a discipline. This one time, it was luck wearing discipline’s clothes.
What I actually did afterward was first fix the problem where it had happened. I added protection around the daily rollover window directly into my own Pine scripts, so they would stop firing execution commands into MatchTrader during the period when the platform could be unavailable.
That solved the problem for me, but after a while it seemed like the wrong place for the lesson to end. Why should only people using my own scripts get that protection? So when we continued developing the AlgoWayWS-MT5 EA, we brought the same idea down to the execution layer for everyone. The EA now includes a Daily Trade Window, Autoclose mode, and separate London, New York, and Tokyo session controls, giving AlgoWay users a way to define when trading should and should not be active regardless of how their original strategy was written.
What I’d Tell the Version of Me From Four Months Ago
Fix the rollover gap before it fixes itself in your favor once and teaches you the wrong lesson. Set the day-transition buffer. Know what your platform does with a close command that fires into a window when the broker is offline. I got away with it — genuinely, materially got away with it — and the only reason I’m writing this instead of writing a very different post is that the market happened to turn the way it did.
Twain also said a lie can travel halfway around the world while the truth is still putting its boots on, and I think about that a little differently in trading terms: a bug can travel a very long way while you’re still congratulating yourself on your discipline. Next time, the math doesn’t have to be that generous. Set the buffer. Don’t let the story you tell at the meetup be the version where you got lucky and called it a plan.
