Early Warning Indicator¶

Monitor — instantiates Overshoot-Crash Load Management

Watches leading precursors — accelerating growth, rising variance, slowing recovery, thinning reserves — that flag an approaching crash while there is still time to act.

By the time a stock's level looks dangerous, the cheap options are usually gone. Early Warning Indicator watches the system's dynamics instead of its level — the tell-tale changes that arrive before a crash: growth that is still accelerating, rising variance and jitter, a lengthening time to recover from small shocks, and reserves that are quietly thinning. Its defining bet is that a system losing resilience announces it in these leading precursors well ahead of any threshold being visibly breached, so a signal can fire while drawdown, diversion, or capacity protection are still feasible. That is what separates it from the archetype's dashboards: it is not an operational picture of where things stand now, but a tripwire tuned to catch the turn.

Example¶

A lender is riding a fast credit boom. The loan book looks healthy on every level metric — delinquencies are low, capital ratios fine — so a level-only view sees no problem. The Early Warning Indicator watches the derivatives instead: month-over-month loan growth is still accelerating; the dispersion of borrower quality in new originations is widening (rising variance); the time for a stressed borrower to cure a missed payment is creeping up (slowing recovery); and the liquidity buffer, as a share of the book, is thinning. None of these is a loss yet. Together they say the book is losing resilience.

The indicator fires an amber signal perhaps a couple of quarters before losses would show on the level metrics — long enough that the lender can still tighten origination and rebuild the buffer without a fire-sale. Read alone, each precursor is noise; read as a set against a resilient baseline, they are the shape of a bust forming.

How it works¶

Track precursors, not just levels. Compute rate-of-change, acceleration, variance, and autocorrelation on the stock and on the reserves that will have to absorb its unwind.
Watch recovery time. Measure how long the system takes to bounce back from small perturbations; a lengthening recovery is a classic loss-of-resilience tell.
Compare to a resilient baseline. A precursor matters only relative to the system's own calm-state behavior, so the baseline definition is part of the instrument.
Fire leading, accept some false alarms. The signal is deliberately early and therefore probabilistic — its job is lead time, not certainty.

Tuning parameters¶

Sensitivity / trip threshold — how strong a precursor must be before firing. Lower catches the turn earlier but cries wolf more; the setting encodes how costly a missed crash is versus a false alarm.
Indicator set — which precursors are watched (acceleration, variance, recovery time, reserve depletion). Broader sets are more robust but noisier and harder to interpret.
Lookback window — how much history defines "normal" and how quickly the baseline adapts. Too short chases noise; too long misses a fast turn.
Lead-time target — how far ahead of the crash the signal aims to fire, traded directly against confidence.

When it helps, and when it misleads¶

Its strength is buying time — the one thing the archetype most needs and most easily loses — by catching an accelerating stock that will cross a nominally safe band before level-only monitoring reacts.

Its failure modes are the failure modes of any leading indicator. It produces false positives and can be argued away as noise precisely when it matters; it can miss abrupt, noise-driven collapses that arrive with no precursor at all; and repeated near-misses breed alarm fatigue. The classic misuse is tuning it to stay quiet — raising thresholds until it never fires — so no one has to authorize the costly early action it would trigger. The named phenomenon it leans on, critical slowing down, is real but not universal,^[1] so the discipline is to wire the signal to a pre-committed action band and response rather than leaving each firing to be re-debated on the day.

How it implements the components¶

Early Warning Indicator realizes the archetype's leading-detection layer — the signals that flag an approaching problem:

early_warning_signal — its primary output: the precursor-based tripwire (acceleration, variance, recovery time, reserve depletion) that fires ahead of a visible crash.
saturation_signal — the composite level-plus-rate-plus-stress reading that shows the stock entering the regime where further growth increases instability.

It only signals; it neither aggregates the full operational picture nor holds the action-band status — that is Saturation Dashboard — and it does not track the post-peak tail, which is Post-Crash Residual-Load Dashboard. It also trips a response but does not enforce it: capping the input is Threshold-Triggered Input Cap.

Instantiates: Overshoot-Crash Load Management — this monitor supplies the lead time the rest of the intervention depends on.
Consumes: Growth-and-Crash Stock-Flow Model supplies the resilient-baseline and capacity reference the precursors are measured against.
Sibling mechanisms: Growth-and-Crash Stock-Flow Model · Saturation Dashboard · Post-Crash Residual-Load Dashboard · Threshold-Triggered Input Cap · Controlled Drawdown Schedule · Sink Capacity Audit · Cohort Staggering · Clearance Pathway Enhancement · Source Reduction Program · Secondary-Capacity Reserve Activation · Staged Harvesting or Decommissioning · Hotspot Containment and Removal · Reentry Gate Review

References¶

[1] Critical slowing down — as a system approaches a tipping point it recovers more slowly from perturbations, showing rising autocorrelation and variance. It is a genuine, widely-studied early-warning signature in ecology and other dynamical systems, but it can be absent when a transition is abrupt or noise-driven, which is why the indicator is treated as a probabilistic tripwire, not a guarantee. ↩