The trade was up $458. It closed at $46.

Ask a trader how their system is doing and they will tell you the net. Ask what the trades were worth at their best moment and most cannot tell you, because nobody logs it.

That second number is where the money is.

Give-back, and why net profit hides it

Every trade has a peak — the best unrealized profit it ever showed. The gap between that peak and what you actually banked is the give-back. Sum it across every trade and you get a number that is usually far larger than the net, and it is invisible on a standard performance report.

Do that across fifty trades and you have a system that looks profitable, feels frustrating, and is leaving most of its edge on the table. The entry was right every time. That was never the problem.

Why this matters more on a prop account than anywhere else

On an account whose threshold ratchets on unrealized profit, give-back is not a missed opportunity. It is a direct cost. The peak raised your floor permanently; the give-back removed the profit that justified it. You are strictly worse off than if the trade had never gone in your favour at all.

That is the argument for treating give-back as a first-class metric rather than a curiosity. On a cash account it costs you upside. On a trailing account it costs you upside and lifespan.

The trap in fixing it

The obvious fix is to bank profit sooner. It is also how most people destroy a working system, because a trailing exit has an internal coherence that is easy to break:

The arming threshold must be smaller than the risk. If a trail only engages after the trade has earned more than it was risking, it will rarely engage at all — and it will engage least on exactly the volatile days that produce the biggest peaks. A threshold that scales with volatility while the stop stays fixed is the specific version of this mistake, and it is silent: the system looks fine and the trail simply never fires.

The arming threshold must also exceed the give-back distance. Otherwise the trail locks in a loss at the moment it engages, which is worse than having no trail.

A percentage give-back is self-coherent. A fixed-tick one is not. Giving back a share of the peak always locks something positive. Giving back a fixed number of ticks can lock a negative if the peak was small.

We have broken every one of these and measured the damage. It is why we changed what we judge on.

What we actually optimize for

We rank changes to a strategy's trade management on how much of the available peak it captures and on intraday drawdown — not on net profit. Net over a few hundred trades is noisy enough that a handful of trades can flip its sign, and we have watched exactly that happen: the same change looked like a loss in one year and a gain in the next, while the capture and drawdown improvements held steady in both.

So when we say a build is better, we mean it kept more of what it earned and spent less of the account getting there. That claim survives resampling. “It made more money” frequently does not.

The dollar figures in this article are an illustration of the mechanic, not a report of any specific engine's results. Nothing here is a performance claim.