Step 4 · Simulation

Step 4 computes on two different bases, and the view keeps them apart.

• Exactly one history. It carries the account path, the curve family, the binding rule, the time to target and the breakdown by strategy.
• Reassembled series. They carry only the block "how these numbers come about", the funnel and the robustness band.

The chosen entry day does not affect the second part. The series are reassembled, and in them there is no day anyone could have entered on.

The three outcomesCopy Anchor Link

Every reassembled series ends in exactly one of three states: it passes, it breaches, or it is still running at the horizon. The three numbers partition the same series completely.

The third state is not an edge case. How large it is depends on the horizon and not on the mechanism: a deadline means fewer days, and then more series stay undecided.

The Wilson intervalCopy Anchor Link

With 1,000 drawn series and a share around 0.44, the standard error is roughly 1.6 percentage points, so the 95 percent interval is about plus/minus three points. A number with one decimal claims thirty times that resolution, which is why the interval sits at the number.

Wilson is used, not the normal approximation. With zero events the normal approximation yields an interval of width zero and thereby claims perfect precision; Wilson yields an asymmetric interval with positive width.

The interval has the same denominator as the number it belongs to. An interval belongs to a number and must contain it.

It says how precisely that number is determined, and nothing about whether the model is right.

If no series breaches at allCopy Anchor Link

That is not a pass rate of a hundred percent but a calculation without an event. The statement is only that the threshold bites less often than this number of series can resolve. No standard error is shown, because it would be exactly zero and would claim perfect precision.

The funnel, from passed to paid outCopy Anchor Link

Five stages, each a subset of the previous one:

1. Target. The evaluation is passed.
2. Buffer. In addition, the funded account reaches the buffer.
3. Trading days. In addition, the qualifying trading days accumulate.
4. Consistency. In addition, the consistency threshold holds once.
5. Payout. All three are open at the same time.

Each stage carries two rates: the share of all series and the share of the previous stage. Showing only the first leaves open whether the chain breaks early or late; showing only the second turns a rare chain into a frequent one.

The last stage requires no new condition but simultaneity. What is lost there are accounts where each condition was met once, but never all three on the same day.

The order of the middle three stages is a choice, not a fact. An account can have its trading days together before it reaches the buffer.

Funded accounts only get whatever is left of the history after the target was passed. Passing late leaves little series, and the fact that nothing materialises then says something about the length of the history and nothing about the account.

The payout conditionsCopy Anchor Link

The calculation runs on the funded account. It starts on the day the evaluation is passed, at zero. Buffer, trading days and consistency count from there anew. The account can also breach, and that outcome is reported separately.

The three conditions must be met at the same time, not one after another: a losing day pushes the balance below the buffer and the chain starts over.

The day the conditions open is not the payout day. On the day the buffer gate opens, the balance sits exactly at the buffer, and the buffer stays in the account: there is nothing to withdraw.

The three values are documented, not assumed. They vary across firms, and anyone who can set them will set them until the result pleases. The view therefore names the documented range and computes with a clearly named profile.

The robustness bandCopy Anchor Link

The limit above the slider comes from one ordering of days. The band is its range across the reassembled series.

For planning, the left number counts, not the middle. It says what would still have held if the days had fallen unfavourably.

Beside it sits the hardest real stretch of the same length, and that is regularly harder than the drawn series at their lower edge. That is a property of the method: the draw reassembles blocks, and in doing so every cluster of bad days longer than one block falls apart. The band is therefore friendlier than reality rather than stricter.

The limit above the slider stands on the whole history and therefore on a different length than the band. Which of the two comes out stricter cannot be predicted: the floor depends on the start of the series, and whoever made a profit first falls from a greater height.

The model boundaryCopy Anchor Link

At some risk factors the analysis says it cannot compute anything above that. That does not mean things are bad up there. A model boundary is therefore labelled and not folded into the number.