It's the credit cycle.

The ~2.1-year volatility cycle we found using astrology as a frequency detector is the credit-volatility cycle. We can now see it directly, measure it, and explain what drives it.

Here’s how we know.

We bandpass-filtered every major asset class to the same 600–1,000 day frequency band and cross-correlated them with the S&P 500 volatility cycle we isolated yesterday. If the equity volatility cycle is driven by something — monetary policy, credit conditions, commodity demand, risk appetite — the driver should show up at the same frequency, either leading or lagging.

One correlation came back at 0.975.

The smoking gun

High-yield corporate bond volatility is correlated with S&P 500 volatility at r = 0.975 in the ~2-year frequency band.

That is not a typo. The correlation between the ~2-year component of high-yield bond volatility and the ~2-year component of equity volatility is 0.975 out of a maximum of 1.000. They are, for practical purposes, the same cycle.

The lag is -25 days — HYG volatility trails S&P volatility by about one month. Equities lead credit, which is the standard story: equity markets are forward-looking, credit markets adjust to realized conditions.

The full cascade

Here’s every asset we tested, sorted by correlation strength:

Reading this table tells a complete macro story, and it reads like a textbook credit cycle:

Phase 1 — short rates rise (T-Bill leads by ~23 months). The Fed tightens. Money becomes expensive.

Phase 2 — equity returns fall (leads vol by ~3 months). The stock market, being forward-looking, prices in the tightening before volatility actually arrives.

Phase 3 — equity and credit volatility spike (concurrent, the cycle peak). Investment-grade and high-yield bond volatility move in lockstep with equity volatility. This is the crisis phase — forced selling, margin calls, spread widening.

Phase 4 — high-yield bonds sell off (lag ~3.5 months). Credit markets adjust. Spreads widen. Weaker borrowers get squeezed.

Phase 5 — Treasury yields fall (lag ~2 months). Flight to safety. Capital flows from risk assets to government bonds.

Phase 6 — yield curve flattens (lag ~8 months). The long end rallies while the front end remains elevated, compressing the spread.

Phase 7 — oil prices fall (lag ~10 months). Real economy impact. Demand destruction hits commodity markets.

Phase 8 — gold falls (lag ~19 months). The risk-off trade unwinds. Deflationary forces dominate.

Then the cycle resets. The Fed eases, credit conditions loosen, risk appetite returns, and the 2-year clock starts again.

Why ~2 years?

The natural question: why does this cycle have a period near 2 years?

Monetary policy transmission. The standard estimate for the lag between a Fed rate change and its full impact on the real economy is 12–24 months. A tightening cycle takes roughly 12 months to peak, followed by 12 months of easing to trough — total period approximately 24 months, or 730 trading days. Our measured period of 788 trading days (~2.15 calendar years) is right in this range.

The cycle strengthened post-2000 (amplitude doubled) because post-2000 monetary policy became more aggressive and more globally coordinated. The Greenspan put, the Bernanke QE era, and the post-COVID policy cycle all compressed the feedback loop between central bank action and market response. More leverage in the system means tighter coupling between rate changes and volatility responses.

The period also wanders — from 734 days in some decades to 938 days in others. This is consistent with an endogenous economic cycle rather than an exogenous astronomical one. Economic cycles don’t have fixed periods. They respond to policy, shocks, and structural changes. A planetary orbit would be metronomic. The credit cycle is not.

Why Mars found it

Mars orbits the Sun in 687 days (sidereal) but returns to the same position relative to Earth in 780 days (synodic). The synodic period is what matters for zodiac sign positions as seen from Earth.

The credit-volatility cycle has a mean period of 788 days. Mars’s synodic period is 780 days. The difference is 8 days out of 780 — a 1% mismatch. This is close enough that when you bin market data by Mars’s zodiac sign (12 bins of ~65 days each), you get phase bins that approximately track the credit cycle.

“Mars in Capricorn” is not a celestial influence. It’s a label for “we are approximately 270–300 degrees into a ~780-day clock.” The credit cycle happens to have a similar period. During any given decade, the phase alignment between Mars’s orbit and the credit cycle drifts slowly — sometimes in sync (making “Mars in Capricorn” look predictive), sometimes out of sync (making it disappear).

This is why the signal replicated across international markets (all four share the same credit cycle) but failed the pre-2000 temporal split (the cycle was weaker and at a different period). It’s why the Rayleigh test showed weak phase-locking (R = 0.067 — similar frequencies, not causal coupling). And it’s why the Mars sign at cycle peaks was uniformly distributed (the phases drift relative to each other).

The astrology was never the signal. The credit cycle was the signal. Mars was the clock that happened to tick at almost the right rate.

What this means

For the astrology hypothesis: dead, thoroughly. The entire chain from “Saturn in Taurus predicts influence” to “Mars in Capricorn predicts volatility” has been traced to its mundane explanation. Every apparent astrological signal in this project was either a statistical artifact (slow-planet aliasing), a trivial rediscovery (September-October seasonality), or a coincidence of orbital mechanics with an economic cycle (the credit-volatility cycle).

For the macro synthesis engine: the project found something real. A ~2-year credit-volatility cycle, visible in the spectral decomposition of equity volatility, correlated at r > 0.9 with credit market indicators, and consistent with the monetary policy transmission lag. Whether this is novel depends on how you look at it — the credit cycle itself is well-known, but its precise ~2-year periodicity in the post-2000 era and its spectral characterization may not be widely reported in this form.

For trading: the cycle is tradeable in principle. The S&P return cycle leads the volatility cycle by ~70 days. T-Bill rates lead by ~23 months. Investment-grade bond volatility is concurrent. These lead-lag relationships could inform a systematic strategy: when the return cycle turns negative, expect volatility 3 months later; when short rates peak, expect the volatility peak ~23 months later.

Whether it works in practice depends on whether the cycle persists out of sample — and whether 8.4% of variance is enough to extract alpha net of transaction costs. That’s the next project.

The arc

Eight posts ago we set out to test astrology. The methodology was designed to be honest: find patterns, then kill them.

The pattern-killing worked. Saturn in Taurus died on Nobel replication. Slow-planet aliasing died on the fast-planet restriction. The September-October effect was explained by the calendar. Mars in Capricorn died when we bandpass-filtered the credit cycle and watched it drift out of phase with Mars’s orbit.

What survived is not astrology. It’s a macro credit cycle with a ~2-year period, driven by monetary policy, visible in equity volatility, credit spreads, bond prices, commodities, and currencies. The zodiac found it because zodiac signs are phase bins, and Mars’s orbit happens to tick at 780 days, which is 8 days short of the cycle’s mean period.

We used the wrong tool and found the right thing. That’s the whole story.

The methodology

Cross-correlation analysis. Each market proxy downloaded via yfinance (daily data, 1990–2026 where available). Proxies interpolated to S&P 500 trading dates. Both series bandpass-filtered to 600–1,000 day band (3rd-order Butterworth, zero-phase). Pearson cross-correlation computed at lags from -500 to +500 trading days in steps of 5 days. Peak correlation and lag reported.

Proxies tested. 10Y Treasury Yield (^TNX), 13W T-Bill Rate (^IRX), yield curve slope (10Y–13W), HYG (iShares High Yield Corporate Bond ETF), LQD (iShares Investment Grade Corporate Bond ETF), GLD (SPDR Gold), DX-Y.NYB (US Dollar Index), CL=F (WTI Crude Oil Futures). For each proxy, both price level and realized volatility (21-day rolling) were tested.

S&P 500 return cycle. 63-day rolling log return, bandpass-filtered to the same 600–1,000 day band.

Data limitations. HYG available from 2007 (19 years). LQD from 2002. GLD from 2004. These are relatively short for a 2-year cycle — roughly 8–10 full cycles. The correlations are high but the effective sample size is small. The Treasury yields (from 1990) give more cycles and still show significant correlations.