A Primer on Liquidity Cycles
What Drives Markets Isn't Price — It's Liquidity
Every major asset rally and crash in the last two decades has coincided with large swings in one variable: the rate of change in global liquidity. Not the level — the delta.
When central banks expand balance sheets and M2 grows, risk assets go up. When they tighten, risk assets come down. It's not complicated. But it's also not well-quantified by most market participants, which is where our system comes in.
The M2 Signal
M2 money supply is the broadest measure of money in circulation that the Fed publishes regularly since M3 was discontinued in 2006. Our system tracks it through the FRED API (M2SL series) and computes:
- Year-over-year growth rate
- 3-month rolling momentum
- Deviation from 10-year trend
- Rate of change of the rate of change (acceleration)
Why Acceleration Matters
Markets are forward-looking. They don't wait for M2 to peak — they react to the
second derivative. When M2 growth is still positive but decelerating,
risk assets have already begun to weaken. Our m2-growth-regime-12m slot
captures exactly this dynamic.
The Yield Curve Connection
| Regime | M2 Growth | Yield Curve | Asset Impact |
|---|---|---|---|
| Expansion | > 6% YoY | Steepening | Risk-on |
| Peak | Decelerating from peak | Flat | Rotation begins |
| Contraction | < 2% YoY | Inverting | Risk-off |
| Trough | Negative but improving | Re-steepening | Early risk-on |
Illustrative cutoffs; fitted thresholds live in methodology.
What Our System Does With This
The Liquidity Cycle Analysis pipeline ingests M2 data alongside a broad FRED panel, engineers features that capture these regime transitions, and produces conformal prediction intervals with empirically audited coverage (3/12 slots on-target as of the 2026-03-05 audit).
The key insight: you don't need to predict where BTC will be in 12 months. You need to predict which liquidity regime we'll be in, and set your intervals accordingly.
The numbers mean something. The liquidity cycle is the number that matters most.
Published from MacroSynchronicity Labs — Facility 4.2