---
title: How to Read an Economic Release
description: A rates trader's guide to reading a data release: actual vs forecast vs prior, the sigma surprise score, revisions, and how the miss moves the curve.
source: Helious
canonical: https://helious.io/guides/how-to-read-an-economic-release
---

# How to Read an Economic Release

Actual vs forecast vs prior, the sigma surprise, and how the miss actually moves the curve.

**The 30-second read:** a data print is never traded on its level. It's traded on the *surprise*, actual minus the consensus forecast, judged against how noisy that series usually is. Standardize the miss into a sigma ([surprise z-score](https://helious.io/learn/surprise-z-score)) so a 0.1 on CPI and a 90k beat on payrolls sit on the same ruler. Then check the [prior](https://helious.io/calendar) for revisions and read the internals before you fade the first tick. Hot data lifts yields; soft data rallies bonds, and the front end moves most when the print shifts the [Fed funds](https://helious.io/learn/fed-funds-rate) path.

## 1. The three numbers: actual, forecast, prior

Every release lands as a triplet. The **actual** is what the agency reported. The **forecast** is the survey consensus: the median economist estimate from the Bloomberg or Reuters panel, not an official number, and it is what the market has already priced. The **prior** is last period's actual, and it is rarely frozen: [payrolls](https://helious.io/data/nonfarm-payrolls) revises the previous two months, and [GDP](https://helious.io/data/gdp) and [retail sales](https://helious.io/data/retail-sales) get routinely rewritten. The comparison that matters is actual vs forecast, because the forecast is the market's positioning.

Know the exact cut you're reading before the number crosses. [CPI](https://helious.io/data/cpi) and [core CPI](https://helious.io/data/core-cpi) come as both month-over-month and year-over-year; the market trades the m/m core to two decimals even though the print rounds to one: a 0.34 and a 0.26 both show as 0.3 but are a different world for the [PCE](https://helious.io/data/core-pce) handoff. Headline vs core, m/m vs y/y, SA vs NSA: mismatch the cut and you'll misread the whole tape.

## 2. The surprise is what trades, not the level

A 4.1% [unemployment rate](https://helious.io/data/unemployment-rate) is neither bullish nor bearish in isolation. It's bullish bonds if consensus was 3.9% and bearish if it was 4.3%. The market has already discounted the forecast, so price only moves on the residual: actual minus expected. This is why a 'strong' number can rally the front end (it beat a whisper of something even stronger) and a 'weak' number can sell off. Trade the delta, not the headline adjective the wire puts on it.

Watch for the whisper: the buy-side's real expectation, which can drift off the published consensus in the days into a print, especially after a hot [jobless claims](https://helious.io/data/jobless-claims) run or a soft [ISM](https://helious.io/data/ism-services). If positioning is already leaning one way, an in-line number can still move the market as the leaners get squeezed. The published consensus is your anchor, but the pain trade tells you where the risk is.

## 3. Standardize the miss: the sigma surprise

Raw misses aren't comparable across series: a 0.1 pp beat on core CPI and a 70k beat on payrolls mean nothing side by side until you scale them. The [surprise z-score](https://helious.io/learn/surprise-z-score) does exactly that: take the miss (actual minus consensus) and divide by the standard deviation of that series' own historical surprises. The result is in sigmas. A +2σ CPI print is a genuine two-standard-deviation event, rare, and it should move rates hard; a +0.4σ is inside the noise and often round-trips within the hour.

This is the single most useful number on a release, because it puts every indicator on one ruler and separates signal from a technically-a-beat headline. Helious computes the sigma surprise on every release the moment it crosses, captured seconds after the agency posts, so you can see instantly whether a [core PCE](https://helious.io/data/core-pce) or [JOLTS](https://helious.io/data/jolts) print is a real tail event or statistical wallpaper, without doing the arithmetic mid-tape.

## 4. Don't trust the headline: revisions and internals

The headline is the bait; the story is underneath. On [nonfarm payrolls](https://helious.io/data/nonfarm-payrolls), a big beat that comes with a two-month downward revision that nets out the gain is not a strong report. The bond market frequently trades the net of headline-plus-revision, not the top line. Unemployment comes from the household survey while payrolls come from the establishment survey, so the two can diverge; average hourly earnings and the participation rate carry as much wage-inflation signal as the job count itself.

On inflation, the composition decides the Fed read. [Core CPI](https://helious.io/data/core-cpi) is ex food and energy, but the desk zeroes in on shelter/OER (the biggest, laggiest weight) and on 'supercore', core services ex housing, which is the [FOMC](https://helious.io/fomc)'s preferred gauge of sticky domestic inflation. A hot headline built on a one-off used-car or airfare spike is a fade; a hot supercore is the print that repriced the [dot plot](https://helious.io/learn/dot-plot). Read the internals before you commit size.

## 5. Map the surprise to the curve

Direction first: an upside growth or inflation surprise is bearish Treasuries: yields up, prices down, because it pushes the Fed toward higher-for-longer; a downside surprise rallies bonds. Where on the curve it bites depends on what the data speaks to. Data that moves the near-term policy path, CPI, payrolls, [core PCE](https://helious.io/data/core-pce), hits the front end hardest, because 2-year yields are essentially a strip of expected [Fed funds](https://helious.io/learn/fed-funds-rate). That's a bear-flattener on a hot print, a bull-steepener on a soft one.

Size the move in [basis points](https://helious.io/learn/basis-points) per sigma and keep a rough map: a clean +1σ CPI surprise can be worth several bp in [2-year](https://helious.io/auctions/2-year-note) yield and a repricing of the SOFR strip; a growth print like retail sales or ISM leans more on the belly and long end via the term-premium and real-rate channel. The [2s10s](https://helious.io/learn/2s10s-spread) reaction tells you whether the market read the surprise as a policy story (front end leads) or a growth/supply story (long end leads).

## 6. Not every release is tradable

Tier your [calendar](https://helious.io/calendar) by how reliably a series generates sigma. Tier one: CPI, nonfarm payrolls, core PCE, and increasingly [ISM services](https://helious.io/data/ism-services), routinely produce multi-bp moves and deserve pre-positioning. Tier two: [retail sales](https://helious.io/data/retail-sales), [JOLTS](https://helious.io/data/jolts), [ISM manufacturing](https://helious.io/data/ism-manufacturing), weekly [jobless claims](https://helious.io/data/jobless-claims), move rates only on a real surprise or when they confirm a nascent theme. [PPI](https://helious.io/data/ppi) matters chiefly for the components that feed the PCE calculation two weeks later.

Context is a multiplier. The same +1σ CPI is worth far more into a live [FOMC](https://helious.io/fomc) meeting, at a data-dependent turning point in the cycle, or when the market is already leaning hard one way. A first-tier miss in a quiet regime can be a non-event; a second-tier print can dominate the day if it breaks the current narrative. Regime, not just the number, sets the beta.

## 7. Trade the release on Helious

Helious captures each release seconds after it posts and renders the full breakdown: actual vs forecast vs prior with the standardized sigma surprise, the instant it crosses, so you're reading signal instead of refreshing a government page. The [economic calendar](https://helious.io/calendar) sequences what's coming with consensus attached; the live [news feed](https://helious.io/) buckets the reaction into DATA and MOVING tabs with a market-impact slider, and the server-side move detector arms per-symbol alerts with regime-aware thresholds so a real repricing pings you and the noise doesn't.

Pair the data read with the supply read: an auction into a hot CPI is a different animal than one into a soft print. See [how to read a Treasury auction](https://helious.io/guides/how-to-read-a-treasury-auction) for the other half of the front-end story, browse more [guides](https://helious.io/guides), or watch the next release score live on the [desk](https://helious.io/).

## Questions

### What do actual, forecast, and prior mean on an economic release?

Actual is the number the agency just reported. Forecast is the survey consensus: the median economist estimate the market has already priced in. Prior is last period's reading, which is often revised. The market moves on actual minus forecast (the surprise), not on the level, because the forecast was already discounted.

### What is a sigma surprise or surprise z-score?

It's the release's miss standardized into standard deviations: take actual minus consensus, then divide by the standard deviation of that series' historical surprises. It puts every indicator on one ruler, so a CPI beat and a payrolls beat are directly comparable. Roughly, ±1σ is a real move, ±2σ is a rare tail event that should reprice rates hard, and anything under about ±0.5σ is usually noise.

### Does a strong jobs report raise or lower bond yields?

A stronger-than-expected jobs report is bearish Treasuries: yields rise, prices fall: because it pushes the Fed toward higher-for-longer policy. The 2-year and front end react most, since they price the expected Fed funds path. But check the revisions: a headline beat offset by big downward revisions to prior months can net out and blunt or reverse the move.

### Why do revisions to the prior matter more than the headline sometimes?

Because the bond market frequently trades the net of the current headline plus the revision to earlier months. A 50k payrolls beat paired with an 80k downward revision is a net loss of jobs versus prior expectations, so the market can rally even though the headline 'beat.' Series like payrolls, GDP, and retail sales are revised routinely, so the prior is a moving target.

### Which US economic releases move the rates market the most?

CPI, nonfarm payrolls, and core PCE are tier-one and reliably move the curve several basis points on a real surprise, with ISM services close behind. Retail sales, JOLTS, ISM manufacturing, and jobless claims are tier-two: they move rates on a genuine sigma surprise or when they confirm an emerging theme. A print's impact also scales with context: the same miss is worth far more into an FOMC meeting or a data-dependent turning point.