Back
Live · Macro Playbook · Clear

When macro moves, metals move. Here's the historical playbook.

HY credit spreads, yield-curve inversions, housing, jobless claims, oil shocks — the rules that have reliably preceded every recession since 1976. Every card shows you the current value, the level that flips it, and what gold, silver, GDX and GDXJ have historically done while it was at warn or alert.

10 rules · 6 recessionsUpdated daily · 05:00 UTCAs of 2026-06-12
Composite risk level
10.5
/ 100
Clear
Triggered1
Warning0
Watch0
Clear6
When the composite last crossed 40…

A recession followed within 16 months (median). Gold rallied +12.8% over the next year.

Median lead
16mo
34 prior matches
Gold fwd 12m
+12.8%
median
SPX fwd 12m
+12.3%
median
Composite · vs NBER recessions
Composite scoreNBER recessionLatest: clear

Triggered rules

1

Oil +50% / 6m (Hamilton)

Alert

WTI +50% over 6 months. 10/11 post-war recessions associated with an oil shock. Binary supply-shock amplifier.

Value
+58.8%
Target
≥ Alert
Lead
~6mo
Triggered 2026-04-29 (44d)
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Warn+0.5%-4.4%+0.2%+11.1%
Alert-0.1%-0.4%+0.2%+11.1%
Oil supply shock — WTI up 50% in six months. Historically this is the worst moment to add to miners.
AVOID juniors: GDXJ -17% median over 6 months, 94% of windows negative. Silver -2.5%. Gold roughly flat. Equities surprisingly resilient (+12%).

Clear

9

HY OAS > 600 / 800 bps

Clear

High-yield credit spread. > 600 bps = warn, > 800 bps = triggered. Zero false positives since 1996 when combined with curve inversion.

Value
278 bp
Target
> 800 bps
Lead
~2mo
Hit rate 100%
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Warn+5.0%+2.2%+0.2%+11.1%
Alert+9.7%+6.6%+0.2%+11.1%

10Y-3M curve regime

Clear

10-year minus 3-month Treasury spread. Rule fires (status 3) while the curve is inverted or during the 12 months after un-inversion — the classic Estrella-Mishkin peak-risk window. Lead: 12-18 months.

Value
0.70
Target
≥ Clear
Lead
~14mo
Hit rate 89%
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Warn+8.0%+9.8%+0.2%+11.1%
Alert+8.5%+3.9%+0.2%+11.1%

10Y-2Y curve regime (steepener-aware)

Clear

10Y-2Y spread with bull/bear steepener decomposition. Status 3 during inversion (unchanged). In the 12mo post-uninversion window we classify the steepener type via 60-day Δ on each leg: bull steepener (2Y falling faster — Fed cuts being priced in) keeps status 3 (40% historical recession-within-12mo rate); bear steepener (10Y rising faster — healthy growth expectations) downgrades to status 1 (only 14% rate); flattening also status 1; neutral status 2. Empirical lift: bull-steepener filter takes precision from 27.5% (baseline uninversion) to 40% (1985-present).

Value
0.39
Target
≥ Clear
Lead
~10mo
Hit rate 100%
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Warn-2.6%-3.5%+0.2%+11.1%
Alert+6.1%+1.7%+0.2%+11.1%

Housing Starts YoY < -20%

Clear

Housing starts YoY change. Double-digit decline preceded 8 of 9 recessions since 1960. Lead: 6-12 months.

Value
+9.5%
Target
≥ Clear
Lead
~9mo
Hit rate 89%
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Warn+1.2%0.0%+0.2%+11.1%
Alert+1.2%+1.1%+0.2%+11.1%

Initial Claims 4wk MA YoY > +20%

Clear

Initial jobless claims 4-week MA YoY. Reliably coincident, weak lead. Best used as a confirming signal alongside curve + credit. **Now superseded in the composite by the No Hire, No Fire Index** — kept here for the per-rule timeline only.

Value
-7.9%
Target
≥ Clear
Lead
~1mo
Hit rate 90%
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Warn+0.9%-5.3%+0.2%+11.1%
Alert+4.5%+0.7%+0.2%+11.1%
Tactical signal — flags miner-specific risk, excluded from composite.

Sovereign bond stress

Clear

Counts four sovereign-debt stress sub-signals: (1) 10Y rate shock ≥ +60 bps/60d while bear-steepening, (2) long-bond index ≥ 12% below its 1y high (duration crash), (3) gold outrunning bonds by ≥ 25pp over 90d (fiat-hedge rotation), (4) ≥ 2 of Japan/UK/Germany 10Y up ≥ 50 bps over 6m (global selloff). 1 = watch, 2 = warn, 2+ incl. fiat-hedge = alert. Excluded from the recession composite — bond stress episodes (1994, 2013, 2022) historically did NOT predict US recessions (2/9 hit rate).

Value
0.0
Target
≥ Clear
Lead
~6mo
Hit rate 22%
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Warn+2.9%+4.8%+0.2%+11.1%
Alert-3.4%-16.7%+0.2%+11.1%
Tactical signal — flags miner-specific risk, excluded from composite.

Oil shock × credit widening

Clear

Oil shock fires + HY OAS is widening (20d Δ ≥ +30 bps) AND HY OAS level ≥ 350 bps. Empirically distinguishes 'oil shock that survives' (HY tight) from 'oil shock that kills miners' (HY widening). 4/4 GDX killer events since 2007 had this combo; 3/3 survivors did not. Excluded from recession composite (it's a miner-tactical signal, not a recession-risk signal).

Value
58.8
Target
≥ Clear
Lead
~0mo
Hit rate 100%
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Warn+7.5%-1.8%+0.2%+11.1%
Alert-3.6%+6.6%-14.2%+7.3%
Tactical signal — flags miner-specific risk, excluded from composite.

SLOOS C&I tightening

Clear

Senior Loan Officer Opinion Survey: net % of large/mid-market banks tightening C&I lending standards. ≥ 10% = watch, ≥ 20% = warn, ≥ 20% for 2+ consecutive quarters OR ≥ 35% = alert. Quarterly since 1990-Q2. Catches credit-led recessions (1990, 2001, 2008); does not catch pure supply-shock recessions (COVID). Confirming signal, not primary.

Value
8.1
Target
≥ Clear
Lead
~3mo
Hit rate 75%
Fwd 6-month median returnWhile rule is active
AuAgGDXGDXJSPX
Alert+4.5%+1.5%+0.2%+11.1%

No Hire, No Fire Index

Clear

Labor-market composite (NFP 6mo trend, continuing claims trend, yield curve, 10Y yield trend) inspired by George Gammon's framework. Backtested AUC 0.92, avg 5.3-month lead, zero false positives across 2001/2008/2020. Replaces the standalone claims_yoy + sahm rules in the composite.

Value
69.4
Target
≥ Clear
Lead
~5mo
Hit rate 92%
After the last 34 ≥ 40 readings

Gold wins, SPX mixed, miners volatile.

Forward returns measured from the day the composite first crossed 40 (events at least 6 months apart).

First ≥ 40Rules triggeredMonths → recessionGold fwd 12mSPX fwd 12mGDX fwd 12m
1978-09-29115 mo+90.3%+5.9%0.0%
1979-03-2819 mo+112.4%-3.8%0.0%
1979-09-2413 mo+86.2%+18.1%0.0%
1980-03-24115 mo+0.4%+35.6%0.0%
1980-09-2219 mo-35.8%-10.5%0.0%
1981-03-2313 mo-37.6%-16.3%0.0%
1981-09-210105 mo-4.6%+6.5%0.0%
1982-03-22199 mo+27.0%+33.6%0.0%
1982-09-20193 mo-3.5%+38.1%0.0%
1989-03-03116 mo+4.1%+14.6%0.0%
1989-08-30210 mo+6.0%-9.1%0.0%
1990-02-2624 mo-11.3%+10.4%0.0%
1990-08-272126 mo-8.1%+22.3%0.0%
2000-03-16111 mo-8.3%-21.1%0.0%
2000-09-1226 mo+3.1%-29.9%0.0%
2001-03-12281 mo+9.4%-1.2%0.0%
2001-09-10375 mo+10.6%-16.7%0.0%
2003-02-07158 mo+12.1%+37.4%0.0%
2006-04-19119 mo+12.8%+12.3%+10.3%
2006-10-16214 mo+28.0%+12.4%+31.2%
2007-04-1628 mo+37.4%-9.1%+14.5%
2007-10-1512 mo+11.6%-35.6%-43.0%
2008-04-141142 mo-4.0%-36.6%-29.8%
2008-10-131136 mo+27.1%+7.0%+83.5%
2009-04-131130 mo+29.9%+39.4%+41.4%
2009-10-120124 mo+29.0%+8.7%+18.4%
2020-03-162— no NBER yet+14.2%+66.1%+48.9%
2020-09-141— no NBER yet-8.1%+31.3%-24.6%
2022-10-181— no NBER yet+21.1%+16.0%+24.7%
2023-04-172— no NBER yet+17.7%+21.7%-3.7%
2023-10-162— no NBER yet+39.0%+33.0%+40.3%
2024-05-012— no NBER yet+40.7%+11.7%+41.7%
2024-10-282— no NBER yet+41.6%+18.3%+71.7%
2025-04-282— no NBER yet+38.6%+29.1%+79.1%
Median16 mo+12.8%+12.3%0.0%

Forward returns measured from first cross of 40 · events ≥ 6 months apart · median across 34 events