Same tier, different gate: pool-budget row mechanics.

The two-axis grid

Tiers are columns. Gates are rows. Deployment lives at the intersection.

The Pool Budget Panel is a grid, not a list. One axis carries the seven tiers — the tactical answer to 'how strong is the evidence right now?' The other axis carries the gate states — the structural answer to 'how wounded is the account?' Every cell in that grid holds a cycle pool percent and a per-trade average. When the 7-Tier Decision Engine resolves a tier, it doesn't look up 'T4' in the abstract; it looks up T4 in the row belonging to the current gate. This is the mechanism that lets the same evidence quality produce different capital commitments depending on what the account can currently absorb.

The worked numbers

One tier, three gates, three different pools.

The clearest illustration from the panel's own budget rows: T4 Mid in the Growth row authorizes a 29% cycle pool at roughly 7.25% average risk per trade across the 4-trade cycle. The identical tier in the Buffer row authorizes 22% and 5.50%. In the Floor row, 18% and 4.50%. Nothing about the evidence changed between those three cells — the week could look identical on the scorecard. What changed is the account's structural capacity to absorb a losing cycle, and the rows re-price deployment accordingly.

Why not just lower the tier?

Because that would burn the evidence signal to pay for the capital signal.

A tempting simplification: skip the rows, and when drawdown deepens, just force a lower tier. But that collapses two independent pieces of information into one number. If Buffer-state pressure were expressed by demoting T4 evidence to a T2 label, the decision log would lie — it would record weak evidence when the actual evidence was mid-grade and the actual constraint was capital. Later analysis of tier performance would be poisoned: T2 results would mix genuinely weak-evidence weeks with strong-evidence-under-pressure weeks. Keeping the tier honest and letting the row carry the compression preserves both signals for every downstream module — the scorecard, the attribution analysis, the Monte Carlo comparison — that reads the log.

Row discipline

The row is selected by arithmetic, never by preference.

The active row follows mechanically from drawdown measured against the equity peak: the gate bands map drawdown percent to gate state, and gate state selects the row. There is no operator input in that path. This matters most exactly when it feels worst — a trader two good days into a recovery wants the Growth row back, and the arithmetic says Buffer. The panel's answer is that rows change when the equity curve says so, not when confidence returns first. Confidence recovering before capital is precisely the sequence that produces round-trip drawdowns.

• Row selection input: drawdown % from equity peak. Nothing else.
• Row changes take effect at the next cycle boundary — never mid-cycle.
• An operator who disagrees with the row logs the disagreement and follows the row anyway; the log is where the system learns.

The key idea

The rows let the system compress capital without corrupting evidence.

Every governance system faces the same tension: respond to damage without destroying the information needed to diagnose it. The pool-budget rows are MARS's answer. Damage compresses the row. Evidence still selects the tier. The two signals travel separately through the entire stack, and every module downstream gets to read both truths instead of one blended, unrecoverable average.