---
title: Surprise Z-Score
description: Surprise z-score explained: how standardizing beats and misses makes CPI, NFP, and retail sales comparable, with thresholds and a worked example.
source: Helious
canonical: https://helious.io/learn/surprise-z-score
---

# Surprise Z-Score

A **surprise z-score** standardizes an economic data surprise: take the actual print minus the consensus forecast, then divide by the historical standard deviation of that indicator’s surprises. The result says how unusual the miss or beat is *for that specific series*, in units every release shares.

Raw surprises are incomparable. Is payrolls missing by 60k a bigger deal than core CPI missing by 0.1pp? Z-scores answer that. If NFP surprises have a standard deviation near 70k, a 60k miss is about −0.9σ, notable but routine. If core CPI surprises have a standard deviation near 0.07pp, a 0.1pp miss is roughly −1.4σ, genuinely rare. The CPI miss is the bigger event despite the smaller-looking number.

- |z|  2: an outlier that can reprice the Fed path on its own.

**Worked example:** Consensus expects retail sales at +0.3% m/m; the print is +0.9%. Historical surprise standard deviation for the series is 0.4pp, so z = (0.9 − 0.3) / 0.4 = +1.5. Helious flags it as a strong beat, and the 2-year cheapens 6bp as the market trims cut odds, proportionate to a 1.5σ event.

## Questions

### What is a sigma surprise on an economic release?

The gap between the actual print and the forecast, divided by how much that series normally surprises by. It converts every release onto one scale, so a 2-sigma CPI surprise and a 2-sigma payrolls surprise are comparable shocks even though one is in percent and the other is in thousands of jobs.

### Why not just look at actual versus forecast?

Because the same absolute miss means different things on different series. Payrolls routinely miss by 50,000 and nobody blinks; CPI missing by 0.2 percentage points is a large shock. Standardising is the only way to say which of two prints on the same morning actually mattered.

### How big is a big surprise?

Under 0.25 sigma reads as on the screws. Around 1 sigma is a real surprise. Two or more moves the market. Helious computes the figure the moment a release prints and publishes it with the direction on [the data hub](https://helious.io/data); the rules are in [the methodology](https://helious.io/methodology).

## Live data

Helious measures this and publishes the current figures at [https://helious.io/learn/surprise-z-score](https://helious.io/learn/surprise-z-score), and through the get_auctions and get_rates_and_curve tools on its MCP server at https://helious.io/mcp.