By the time drawdown becomes a deployment figure, the drawdown is gone.

The chain

Four conversions, each one narrowing what the next can know.

The first stage produces a continuous number — the percentage below the anchor — which carries full resolution and could distinguish any two account states. The second locates that number in one of seven bands, and at that instant every distinction inside the band is destroyed. The third converts the band into a brake tone and a maximum tier, which are categorical rather than numeric. The fourth converts the tier and gate row into an authorised pool. Information is discarded at every step and never regenerated, so a question asked at stage four cannot be answered with anything the earlier stages knew. This is normal for a routing pipeline and it has consequences worth being explicit about.

The largest discard

Two accounts six percentage points apart receive the same deployment.

The recovery band spans from just past minus seven to minus thirteen, which means an account that has barely crossed the boundary and one sitting a fraction above the next one are treated identically by everything downstream. Their tier ceilings are the same, their pools are the same, their per-trade sizes are the same. That is a substantial difference in actual capital position producing no difference at all in authorised deployment. It is the direct cost of banding, and the compensating benefit is what banding always buys: an unambiguous state with a name, no borderline cases requiring interpretation, and a deployment figure that does not change every time equity moves a fraction.

Position within band is invisible downstream

The gate cannot tell whether the account is arriving or leaving.

A subtler loss is directional. An account at minus twelve that has been recovering from minus eighteen and an account at minus twelve that has been sliding from minus eight occupy the same band, and the pipeline treats them the same way — same brake tone, same ceiling, same pool. Their situations are not the same. One is a system healing and the other is a system deteriorating, and the difference is exactly the sort of thing that should inform posture. The pipeline cannot express it because the gate is computed from a level rather than from a trajectory, deliberately: a state that depended on direction would need a lookback window, and a window is a parameter, and a parameter at the gate is the thing the ratchet was built to avoid.

Where the lost information lives

Other instruments carry what the pipeline discards, and they are not the gate.

None of this means the discarded information is unavailable — it means it is not available to the deployment decision, which is a different claim. Position within the band and direction of travel both survive elsewhere. The gate dwell reading distinguishes a state touched briefly from one lived in for weeks. The drawdown percentile comparison places the current figure against modelled expectations rather than against a boundary. The structural diagnostics read trajectory across rolling windows. Each of those informs review posture and none of them touches the tier ceiling, which is the separation working exactly as designed: the pipeline is lossy because deployment authority is meant to be coarse and unarguable, and the fine reading belongs to the layers whose job is diagnosis.

What follows for the operator

Do not read the pool as a fine-grained statement about the account.

The practical translation is a caution against over-interpreting deployment figures. An unchanged pool across two weeks does not mean the account's condition was unchanged; it means both readings landed in the same band, which is compatible with meaningful movement in either direction. A pool that drops does not mean a sudden deterioration; it may mean a boundary was crossed by a fraction after a long slow slide. The deployment figure is a routing outcome, not a health metric, and questions about how the account is actually doing belong to instruments that kept the resolution. Reading the pool as a condition report is the most common way the pipeline's coarseness gets mistaken for the account's stability.

• Banding destroys position within band, and nothing downstream can recover it.
• The gate reads a level, not a trajectory — healing and sliding look identical.
• Dwell, percentile bands and the structural pipelines carry what the gate discards.

The key idea

Coarseness at the authority layer is a feature paid for in resolution.

It would be straightforward to compute a continuous deployment figure directly from the drawdown percentage and skip the banding entirely, and the result would be more precise and considerably worse. It would move constantly, it would produce no nameable states for a doctrine to attach to, and it would put every borderline case back into the operator's hands. Discretising into seven bands buys an unambiguous answer to what state is my capital in, and the price is that everything downstream knows less than the first stage did. Knowing which questions the pipeline can no longer answer is the difference between using the output correctly and reading meaning into a number that has none left.