Average gate rank: where the system actually lived this month.

The construction

Gates become ranks, ranks become an average.

Each gate carries a numeric rank, and averaging those ranks across every trading day inside the month produces the metric. The arithmetic is deliberately plain — an AVERAGEIFS over the daily gate-rank column, bounded by the month's start and end dates — and the plainness is a virtue: nothing is weighted, smoothed, or adjusted, so the number is exactly what it claims to be. A month reading 1.3 lived mostly in Growth. A 2.4 sat around Recovery. A 3.7 spent real time between Buffer and Floor. Past 5 and the month was largely conducted in the deep defensive gates, with the arithmetic of recovery running against it the whole way.

What it captures

Duration, not depth — the dimension monthly reviews usually miss.

Max drawdown tells you the worst moment; average gate rank tells you the whole month's residence. Those diverge more than operators expect. A month that spiked briefly into Floor and recovered fast shows an alarming max drawdown and a benign average rank — the account was fine almost all month. A month that never breached Buffer but camped there for twenty sessions shows a modest max drawdown and a materially worse average rank, because deployment was compressed the entire time. The second month is the one that quietly costs compounding, and it's precisely the one a max-drawdown-only review reports as unremarkable. This is the duration dimension from the drawdown-percentile doctrine, measured monthly and made structural.

Why it anchors everything

Three metrics descend from this one.

Average gate rank isn't merely one diagnostic among six — it's the source the others draw on. Compression score is a direct normalization of it. Pressure score carries compression as a weighted component. Structural health carries compression again, at heavier weight. That lineage means the rank's accuracy propagates: a month whose daily gate states were logged incompletely produces a wrong average, and the error travels silently into three downstream scores and from there into the regime classification built on all six. The gate/brake daily record is unglamorous data entry with unusually long reach — which is why the reconciliation discipline treats gate state as a first-class daily field rather than a derived afterthought.

• The rank is an average over trading days, so a month with few sessions is a thinner reading — sample logic applies here too.
• Rank and max drawdown are complements, never substitutes: one measures residence, the other measures the worst instant.
• Every downstream score inherits this metric's data quality. Log gate state daily, not retrospectively.

Reading it across months

The sequence says more than any single value.

One month's average rank is context; a run of them is a trajectory. A sequence drifting upward — 1.4, 2.1, 2.8 — describes a system progressively spending more time compressed, whether or not any single month looked alarming, and that drift is usually visible in the rank series before it registers as a drawdown event. A sequence oscillating hard month to month describes instability the gate volatility index will independently confirm. And a flat sequence low in the range is the quiet signature of a system operating as designed: capital mostly free, deployment mostly authorized, the ladder doing nothing because nothing has required it to.

The key idea

The simplest metric in the monthly layer carries the most weight.

Average gate rank is an average of integers — the least sophisticated calculation in the entire diagnostic stack — and three of the six regime inputs depend on it. That's a common shape in well-built systems: the load-bearing measurement is usually the plain one, and the sophistication lives in what gets built on top. Get the daily gate record right and the whole monthly layer stands on solid ground. Get it wrong and every score above it is confidently mistaken.