The original rules came out of a journal, not off a whiteboard — and then they were written down and stopped moving.

The extraction

Trades first, grouped by volatility state, graded on management outcome — and the rule fell out.

The sequence is the opposite of how trading rules are usually formed. Rather than deciding that trend conditions deserve more room and then looking for confirmation, the journal was grouped by the ATR state each trade occurred in and the management outcomes were compared within each group. What emerged was that the two states genuinely wanted different treatment: controlled static structure in one, continuation logic and more breathing room in the other. The rule is a description of what the evidence showed, which is why it survived the subsequent decade of tooling built around it.

Why freezing is the second half

A rule that keeps adjusting cannot be tested, because there is never a stable thing to test.

Extraction alone is not enough. If the rule continued to be tuned as new trades arrived, then at any moment the current version would have almost no evidence behind it and the accumulated history would describe versions no longer in force. Freezing it — writing down the levels, assigning a version, and leaving it alone — is what allows a year of trades to be evidence about one specific rule. The discipline feels conservative and is the opposite: it is the only arrangement under which a rule can ever be shown to be wrong.

What re-audit actually looks like

Group new trades by state, compare management outcomes, and check the original finding still holds.

Because the doctrine was formed by a repeatable procedure, the procedure can simply be run again on later evidence. Group the recent journal by ATR state, compare management outcomes within each group, and ask whether the two states still want different treatment and whether the levels still sit in the right place. That is a bounded, concrete review with a defined output. It is available because the original process was recorded, and it is not available for the large majority of trading rules, which exist as convictions whose origins nobody wrote down.

• The re-audit is the original extraction, run again on newer trades.
• Its output is either confirmation, or a specific level that has moved.
• A rule with no recorded derivation has no re-audit procedure at all.

The state must be recorded at entry

The whole procedure depends on one field captured before the outcome was known.

The load-bearing input is the ATR state written into the journal at entry. If the state were inferred later from the chart, the grouping would be contaminated by hindsight in the usual direction — trades that ran would be remembered as having occurred in trend conditions, which would guarantee the finding regardless of whether it were true. Capturing the state before the trade resolves is what makes the two groups independent of their outcomes. It is one field, entered in a moment when nobody has any incentive to enter it dishonestly, and the entire doctrine's auditability rests on it.

What this does not license

Provenance makes a rule testable. It does not make it correct, and it does not make it permanent.

It is worth resisting the stronger claim. That the doctrine came from evidence does not mean it is right now, or right for every instrument, or right at the specific levels chosen — the original sample was one operator's book over a bounded period, and the levels are round-ish numbers that fell in the right region rather than optima computed to a decimal. What provenance buys is narrower and more valuable: the rule states where it came from, so a future disagreement with it can be settled by running the same procedure rather than by two people asserting things at each other.

The key idea

Write down where a rule came from, or its only future defence will be that it has always been there.

Most trading plans are a collection of provisions whose origins have been forgotten, which means none of them can be removed safely — nobody knows which are load-bearing and which are scar tissue from a bad month in a market that no longer exists. Recording the derivation costs a paragraph at the moment the rule is written and is the difference between a plan that can be simplified with evidence and one that can only accumulate. The ATR doctrine's real inheritance is not the two break-even levels. It is that somebody wrote down how they got them.