Cost-fatal

That's the whole definition. The raw entries and exits had a genuine edge, the equity curve climbed before a cent of cost came off. Then we added real bid/ask spread, commission, and slippage, and the curve flattened or fell. Every trade the strategy called correctly still cost more to enter and exit than it earned. Not bad logic. A bad bill.

It's the failure everyone expects to be the big one. Trading costs are the standard warning attached to every backtest, and for good reason — they're invisible until you model them honestly. But cost-fatal is only our fourth most common reject reason. The louder story is real. It just isn't the majority story. Most rejects never had a real edge to lose in the first place, so there was nothing for costs to eat.

When costs are the actual killer, timeframe is usually why. A strategy holding a position for hours absorbs a few points of spread across dozens of trades and barely notices. A strategy re-entering every few minutes pays that spread on every single round trip, and the edge per trade shrinks toward the size of the cost per trade. Eventually it crosses zero.

Sub-15-minute FX is the cleanest case we've seen. Strategies built on that timeframe lose to cost so consistently that we stopped testing FX and CFD strategies below the 30-minute chart entirely. The entry logic wasn't wrong. The spread simply arrives before the edge can pay out, and no amount of tuning the signal changes that arithmetic.

The uncomfortable part: a system can be right about direction on most of its trades and still lose money every month, purely to its own transaction costs. Gross returns look like something worth trading. Net returns tell the truth. Cost-fatal is exactly the gap between those two numbers, and you only see it if you price real spread and real slippage into every fill instead of a frictionless market that was never there.