SeriesMagCorrOverlapThrough
These move opposite each other about 50% of the time
Near-zero inverse
Weak relationship — not much actionable signal here beyond directional colour.
US CPI
CPIAUCSL
Building Permits
PERMIT
What to Watch
Drifting from pattern
Recent correlation (+69%) is running 1.5σ away from the long-run average of -1%.
Tighter in drawdowns
Pair moves more tightly when both fall (+77%) than when both rise (+15%) — typical risk-off behaviour.
Regime-dependent
The correlation has flipped sign across time windows — not reliable as a standalone signal.
Time Series
Time-Shifted Correlation
See how correlation changes when one series is offset in time. A taller bar at a non-zero shift means the two move together better when one leads the other — that's a potential lead/lag signal.
Correlation by shift
Click a bar to inspect. Range: -6 to 6 days.
Selected shift
No shift
Correlation at this shift
-1%
No shift — both series at their actual dates.
794 overlapping points at this shift
Time series with Building Permits at original dates
Indexed to 100 at start
Deep Dive
Stability
How the correlation evolves over time. A stable line means the relationship is reliable; large swings signal regime-dependent behavior.
Corr
-1%
Pearson correlation
95% CI
-8% → +6%
Likely range
R²
0.0%
Variance explained
Trend agreement
50.3%
Same-direction moves
Significance
n.s.
Statistical confidence
Data points
794
Deep
Pipeline
Pipeline Summary
794 paired data points survived the daily window.
Raw input
950
795
Normalized
950
795
Prepared
950
795
Aligned
794
794
Invalid removed
Explore
The strongest positive and inverse pairs we’ve saved for each side of this comparison — good jumping-off points if this result raised a question.
Do They Crash Together?
How these series behave when markets are rising, falling, or diverging. A correlation that holds in drawdowns is very different from one that only works in rallies.
Both Rising
+15%
119 periods · Return correlation when both series rose
Both Falling
+77%
7 periods · Return correlation when both series fell
Diverging
-23%
123 periods · Return correlation when series moved apart
Scatter
0
A: 0 / B: 0
Duplicates removed
0
A: 0 / B: 0
Alignment drops
157
A: 156 / B: 1
Series A
CPIAUCSL
FRED · 950 raw → 950 prepared
Series B
PERMIT
FRED · 795 raw → 795 prepared
Sign agreement
100.0%
How often both values share the same sign.
Zero crossings
54
Estimated crossover points between normalized spreads.
Slope
-0.0550
Linear regression slope.
Intercept
1382.1290
Linear regression intercept.
Saved 2 days ago · ID: fred:cpiaucsl|fred:permit|5Y
PERMIT
Building Permits
Strongest positive