The benefits most operators stay for are the ones that cannot be shown.

The first

Less improvisation under stress, visible only as a decision not taken.

Under pressure the expensive move is inventing a response — a new sizing rule after a loss, an exception for a setup that looks unmissable, a temporary suspension of a limit that suddenly seems arbitrary. When the response already exists in the gate architecture, the invention does not occur. That absence is the benefit, and it is definitionally unobservable: there is no record of the rule that was not written, and the operator's own recollection is the worst possible witness because the pressure that would have produced the improvisation is precisely what distorts recall.

The second

Confidence attached to process rather than to recent results.

Confidence built on a winning run is fragile in a specific way: it is highest exactly when skill is most overestimated, and it collapses on the first ordinary losing sequence. Confidence attached to positive expectancy, controlled drawdown, stable execution and consistent adherence behaves differently — it survives a bad month because the bad month does not contradict it. The difficulty is that the two are indistinguishable from outside and from inside while both are holding. Only a serious drawdown separates them, at which point the distinction is no longer an abstraction, and the demonstration cost more than anybody wanted to pay for it. This is the awkward shape of the second benefit: it can only be confirmed by the event it exists to survive, so it is either untested or expensively proven, and never comfortably established in advance.

The third

The question changes shape, and the change is permanent.

Before, the operative question is whether this setup feels strong enough to size up. After, it is which gate applies, what tier is authorised, how much active exposure remains, what the branch expectancy currently is, and what the throttle permits. That is not a rephrasing of the same question — it has a different answer procedure, a different failure mode, and a different relationship to the operator's mood. It also does not revert. Operators who leave the system report keeping the framing, which suggests something real happened and provides no way at all to measure it.

The fourth

The recurring argument about whether the method is broken simply stops.

Most discretionary traders run a continuous background dispute with themselves about whether the current stretch means anything. It consumes attention, it resolves nothing, and it terminates in whichever direction the mood was already leaning. What replaces it is not certainty — the benchmark and the diagnostics frequently return that the stretch is uninformative — but a verdict from something that is not the operator, produced by a standard set before the stretch began. The argument ends because it now has an answer procedure, and the relief is substantial and is not a data field anywhere in the system.

• Prevented improvisations leave no record; recall is the worst available witness.
• Process-attached and result-attached confidence look identical until a drawdown.
• The reframed question does not revert, even for operators who stop using the system.

Why these are published anyway

Omitting them would misrepresent the product in the other direction.

It would be tidier to list only the six benefits that survive an audit and let the rest go unmentioned. That would understate the product materially, because these four are what retention actually runs on — nobody maintains a demanding weekly workbook for three years because the fee-drag figure reconciles. The honest presentation is to publish them, mark them clearly as unverifiable, and decline to argue for them with evidence that does not exist. A reader can then weigh them as testimony, which is what they are, rather than mistaking them for measurements. Testimony from an interested party is weak evidence and is not zero evidence, and labelling it correctly is what allows a reader to discount it appropriately instead of discounting everything.

The key idea

Unverifiable is not the same as unreal, and the distinction is the buyer's to make.

The reason to separate these four rather than blend them into the list is that they carry a different kind of risk. A verifiable benefit that fails to arrive is a defect and is actionable. An unverifiable benefit that fails to arrive is a disappointment with nothing to point at, and no remedy — which is exactly why it should be identified as such before purchase rather than discovered afterwards. Read that way, the four are an honest description of what is being hoped for, sitting next to six descriptions of what is being bought.