The system rarely fails first. The operator does.

The chart is quiet for two weeks, so you take a “small” trade the system didn't signal. Flat is the signal — it is how the system skips the chop that bleeds accounts. If you trade beside it, you are running two systems, and the discretionary one has no backtest.

More signals feels like more opportunity; it is mostly more churn, more cost, and more whipsaw, on a chart the build was never tuned for. The state table even warns you. See timeframes.

Doubling risk after wins, halving it after losses — or worst, doubling it to “win back” a drawdown. The backtested numbers assume a constant budget; a moving one makes your results untestable and usually worse. Set risk per trade once, from the drawdown you can survive, and leave it.

A system that wins well under half its trades will regularly open with three or four losers in a row — that is arithmetic, not malfunction. Its edge lives across dozens of trades and the occasional big trend. Judge it on a year, or check the Strategy Tester's full history whenever you're tempted to judge it on a week.

Mid-bar, price can be through the line at lunch and back above it by the close — Vektor decides on closed bars only. Acting on an unconfirmed bar means trading signals that may never exist. Let the alert or the bar close tell you; that's the no-repaint guarantee working for you.

Each tweak looked reasonable; together they are an untested strategy with Vektor's name on it. Change one input at a time and re-verify, or reset to defaults — the settings page has the rules.

The most expensive pattern in trend following: abandon the system mid-drawdown, right before the trend that would have paid for it. If the drawdown is intolerable, the honest fix is a lower risk setting — decided calmly, in advance — not an exit from the system at its low.