The conditional
A result is not interpretable until the state it occurred in is known.
Nine R gained under Growth was produced at full deployment authority against an account at its peak: it is expansion evidence and can reasonably inform confidence in the method. The identical nine R clawed back under a compressed gate was produced at a reduced tier against a damaged account: it is recovery, which is encouraging and is not licence. Neither reading is available from the number. Both require the state, and the state is not something the operator can supply from memory reliably because the gate may have moved during the week being reviewed. Recollection of one's own capital state is unreliable in a specific direction: operators remember the state they felt they were in, which is usually the more favourable one.
State first, then results, then interpretation. Each step is conditional on the one before it, which is why the sequence is fixed rather than a suggestion about where to start.
