US Macro Pane

Is US data running hot or cold?

Twelve official US series, each scored against its own recent trend and blended into one daily composite. Above zero = data coming in hotter than trend; below = colder. This is a trend-deviation index, not a consensus-surprise index — see the methodology below. Updated daily.
Macro Surprise Composite
Ten years of macro surprise

Weekly snapshots of the daily composite. The shaded band marks the ±1 “normal” zone. Click the legend to toggle the Growth and Inflation sub-indices.

What’s driving it — series breakdown

One row per input series: the latest observation, its “pulse” (the transform scored, e.g. 3-month annualised change), the trend it was compared against, the surprise z-score, and the staleness weight it carries in today’s composite. Click a header to sort.

Series Observation Pulse Trend Surprise z ▾ Weight
Global macro tracker — market-implied, daily

What markets themselves are pricing: curve slopes, real yields, breakevens, credit, the dollar and commodities. Each row shows the current level, 1-month change and where the level sits in its 1-year range.

Methodology

Trend deviation, not consensus surprise

Indices like Citi’s Economic Surprise Index measure data against economist forecasts. This meter has no forecast feed: each series is measured against its own trailing 12-observation trend, scaled by its trailing 5-year dispersion (minimum 3 years, shorter series excluded), clipped to ±3. “Hot” therefore means hotter than its own recent trend, not “beat expectations”. The two usually agree directionally but are not the same number.

Publication-lag honesty

FRED provides observation dates, not release dates. Each series only enters the composite after its typical publication delay (measured from the FRED observation date, which for monthly series is the first day of the reference month): payrolls & unemployment +35d, weekly claims +5d, CPI +43d, retail sales +45d, industrial production +45d, factory orders +65d, Michigan sentiment +28d, Philly Fed +18d, Empire State +14d, housing starts +47d. The ~10-year backfill walks forward using only observations available at each point in time under these constants — an approximation of real-time availability.

Revisions are ignored

FRED serves latest-revised values; first-print values aren’t recoverable from this feed. The backfilled history is therefore modestly cleaner than what a real-time observer would have seen.

Weighting and sub-indices

Each series carries a staleness weight of 0.5(days since observation / 45), so fresh data dominates. Composite = weighted mean of the surprise z-scores. Growth = labor + consumer + industry + housing; Inflation = the two CPI series; Macro momentum = 63-day change in the Growth sub-index. Zones: Hot ≥ +1, Warm ≥ +0.3, Neutral, Cool ≤ −0.3, Cold ≤ −1.

Substitutions & limitations

The high-yield spread (BAMLH0A0HYM2) is capped to roughly three years of history on FRED’s keyless endpoint due to ICE licensing — enough for its 1-year percentile, not for long history. The copper/gold ratio uses front-month futures (HG=F / the site’s existing gold series) and is shown ×1000 for readability. Oil is WTI front-month (CL=F).

Not investment advice. Educational information only — see the disclaimer.