A print is never traded on its level. It trades on the surprise, actual minus the consensus forecast, and even that only means something once you know how noisy the series usually is. Standardize the miss into a sigma (surprise z-score) and a 0.1 on CPI sits on the same ruler as a 90k beat on payrolls. Then check the prior 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 path. That is the whole method in 30 seconds.
Every release lands as three numbers. The actual is what the agency reported. The forecast is the survey consensus, the median economist estimate off the Bloomberg or Reuters panel, and while it is nobody's official number it is the one the market has already priced. The prior is last period's actual, and it is rarely frozen. Payrolls revises the previous two months. GDP and retail sales get rewritten as a matter of course. So compare the actual with the forecast. That is where the market is positioned.
Know which cut you are reading before the number crosses. CPI and core CPI both come month over month and year over year. The market trades the m/m core to two decimals even though the print rounds to one, so a 0.34 and a 0.26 both show as 0.3 and they are a different world for the PCE handoff. Headline is not core, m/m is not y/y, and the seasonally adjusted series (SA) is not the raw one (NSA). Read the wrong cut and you misread the whole tape.
A 4.1% unemployment rate is neither bullish nor bearish by itself. It is bullish bonds if the consensus was 3.9% and bearish if the consensus was 4.3%. The forecast is already discounted, so the only thing left to trade is the residual, actual minus expected. That is why a number the wire calls strong can rally the front end, because it beat a whisper of something stronger still, and why a weak one can sell off. Trade the delta, not the adjective.
The whisper is what the buy side actually expects, and in the days into a print it can drift off the published consensus, especially after a hot run of jobless claims or a soft ISM. That is why an in-line number can still move the market. If positioning is leaning one way, the leaners get squeezed. The published consensus is your anchor. The pain trade tells you where the risk sits.
Raw misses do not compare 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 does the scaling. Take the miss, actual minus consensus, and divide it by the standard deviation of that series' own history of surprises, and the answer comes out 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.
That one number does more work than anything else on a release. It puts every indicator on the same scale, and it separates a real move from a technically-a-beat headline. Helious runs it the moment a release crosses, off a capture taken seconds after the agency posts, so you know straight away whether a core PCE or JOLTS print is a tail event or statistical wallpaper, and you are not doing arithmetic in the middle of the tape.
Read down the release, not across the top line. A big beat on nonfarm payrolls that lands with a two-month downward revision netting out the gain is not a strong report, and the bond market will often trade the net of the two. The unemployment rate comes out of the household survey and the payroll count comes out of the establishment survey, which is why the two can point different ways. Average hourly earnings and the participation rate carry as much wage-inflation signal as the job count does.
On inflation, the composition decides how the Fed reads it. Core CPI strips out food and energy, but the desk goes straight to shelter and OER, the biggest weight and the slowest to turn, and then to supercore, meaning core services excluding housing, which is the 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. Read the internals before you commit size.
An upside surprise on growth or inflation is bearish Treasuries, so yields up and prices down, because it pushes the Fed toward higher for longer. A downside surprise rallies bonds. Where on the curve it bites is the second question, and it depends on what the data speaks to. Anything that moves the near-term policy path, so CPI, payrolls and core PCE, hits the front end hardest, because a 2-year yield is not much more than a strip of expected Fed funds. A hot print gives you a bear flattener, short yields rising faster than long ones, and a soft one gives you a bull steepener.
Carry a rough sense of how many basis points a sigma is worth. A clean +1σ CPI surprise can be several bp in the 2-year yield and a repricing of the SOFR strip. A growth print like retail sales or ISM leans on the belly and the long end instead, through term premium and real rates. Watch 2s10s for the read. Front end leading means the market took it as a policy story, long end leading means it took it as growth and supply.
Rank your calendar by how reliably a series throws off a real sigma. CPI, nonfarm payrolls and core PCE are the top tier, and ISM services is more and more one of them. Those are the prints that routinely move the curve several basis points, and the ones worth pre-positioning for. Below them sit retail sales, JOLTS, ISM manufacturing and the weekly jobless claims, which move rates only on a genuine surprise or when they confirm a theme that is already forming. PPI is worth watching mainly for the components that feed the PCE calculation two weeks later.
The regime sets the beta as much as the number does. The same +1σ CPI is worth far more into a live FOMC meeting, at a data-dependent turn in the cycle, or when the market is already leaning hard one way. A top-tier miss in a quiet stretch can be a non-event, and a second-tier print can own the day if it breaks the story the market has been telling itself.
Helious captures each release seconds after it posts and puts the actual, the forecast, the prior and the standardized sigma surprise on screen the instant it crosses, so you are reading the number instead of refreshing a government page. The economic calendar sequences what is coming, with consensus attached. The live [news feed](/) sorts the reaction into DATA and MOVING tabs with a market-impact slider. A server-side move detector arms per-symbol alerts on regime-aware thresholds, so a real repricing pings you while ordinary chop stays quiet.
Read the data alongside the supply. An auction into a hot CPI is a different animal from one into a soft print, and how to read a Treasury auction covers that other half of the front-end story. The rest are in guides, and the next release scores live on the [desk](/).
