Release times, revisions, consensus and the whisper
Lesson 12 · about 12 min
This lesson pulls the calendar together: when things come out, how much they usually move, why the number you traded may not be the number that ends up in the history books, and how to tell what the market actually expects.
The monthly US schedule
Times are Eastern. Dates are typical; check the calendar on the site at /news/calendar or the publishing agency for the exact schedule, which shifts for holidays and, occasionally, for government shutdowns that delay releases.
| Week of month | Release | Day | Time |
|---|---|---|---|
| Week 1 | ISM Manufacturing | 1st business day | 10:00 |
| Week 1 | JOLTS | Early in the week | 10:00 |
| Week 1 | ADP | Wednesday before NFP | 8:15 |
| Week 1 | ISM Services | 3rd business day | 10:00 |
| Week 1 | Jobs report (NFP) | First Friday | 8:30 |
| Week 2 | CPI | Roughly 10th-15th | 8:30 |
| Week 2 | PPI | Around CPI | 8:30 |
| Week 2 | Michigan sentiment (prelim) | Friday | 10:00 |
| Week 3 | Retail sales | Mid-month | 8:30 |
| Week 3 | Industrial production | Mid-month | 9:15 |
| Week 3 | Housing starts | Around the 17th-19th | 8:30 |
| Week 3-4 | Existing home sales | Around the 20th-25th | 10:00 |
| Week 4 | Flash PMIs | Around the 22nd-24th | 9:45 |
| Week 4 | Durable goods | Around the 26th | 8:30 |
| Week 4 | GDP (advance / second / third) | End of month, quarterly | 8:30 |
| Week 4 | PCE, personal income and spending | Last week | 8:30 |
| Every week | Initial jobless claims | Thursday | 8:30 |
| 8 times a year | FOMC decision | Wednesday | 14:00 |
The 8:30 slot is where the liquidity is thinnest and the moves are fastest, because it is an hour before the cash equity open. The 10:00 releases land in a fully open market and tend to produce smaller, cleaner moves for the same surprise.
Typical reactions, with caveats
The table below gives rough, order-of-magnitude first-thirty-minute ranges for a one-sigma surprise in a regime where the market is actively trading that release. These are not forecasts. In quiet regimes the same surprise can produce a quarter of these moves; in stressed regimes, double. Measure your own instrument over the last six months before you trust any of it.
| Release | ES (index points, approx.) | 10-year yield (bp) | EUR/USD (pips) | Gold ($/oz) | Notes |
|---|---|---|---|---|---|
| CPI | 30-80 | 5-15 | 40-80 | 15-40 | The largest data event in inflation-focused regimes; 2022-2023 saw 2%+ ES days |
| NFP | 20-60 | 5-12 | 30-60 | 10-30 | Mixed prints reverse often |
| FOMC decision plus presser | 30-100 | 5-20 | 40-100 | 15-50 | Two waves; net move by day two is a better measure |
| ISM Manufacturing | 10-30 | 3-8 | 15-40 | 5-15 | Larger when at the 50 boundary |
| ISM Services | 10-30 | 3-8 | 15-40 | 5-15 | Prices paid can matter more than headline |
| Retail sales | 10-25 | 2-6 | 10-30 | 5-10 | Control group drives it |
| PCE | 5-20 | 2-5 | 10-25 | 5-10 | Usually well-telegraphed by CPI and PPI |
| GDP advance | 5-25 | 2-6 | 10-30 | 5-10 | GDPNow reduces surprise |
| Jobless claims | 2-10 | 1-3 | 5-15 | 2-5 | Grows into a major event when the trend turns |
| Housing releases | 0-5 | 0-2 | 0-10 | 0-5 | Homebuilders and lumber react more than indices |
Three caveats to repeat to yourself: regime, positioning, and clustering. Regime is Module 1. Positioning means a market that is heavily long going into a hawkish print falls further than the surprise alone implies. Clustering means that when CPI and FOMC land in the same week, each event's move is conditioned by the other.
Key idea: Release times, not just release contents, shape the move. 8:30 releases hit a thin pre-market; 10:00 releases hit a full one. The size of the reaction depends on regime and positioning at least as much as on the surprise.
Revisions
Almost every series in the table is revised, and some are revised heavily. NFP has two monthly revisions plus an annual benchmark. GDP has three estimates plus annual and comprehensive revisions years later. Retail sales, durable goods and housing starts are all revised the following month. CPI is essentially not revised (seasonal factors are recalculated each February, which slightly alters recent months' seasonally adjusted numbers, but the not-seasonally-adjusted index stands).
The trading implication: you trade the first print, and the first print is the worst estimate. That is fine for event trades and dangerous for narrative-building. "Payrolls have been strong for six months" can become "payrolls were revised down 400k" without any new data. Hold macro conclusions loosely until the revisions are in.
Consensus
Consensus is the median or mean of forecasts submitted by economists to a data vendor (Bloomberg, Reuters and others each compile their own, and they differ slightly). It is collected over the week or two before the release. That lag matters: a consensus collected before a very weak ADP print still reflects pre-ADP thinking.
Along with the median, look at the range and the number of forecasters. A wide range means real uncertainty; the move on the print will be larger for the same surprise because more participants are being forced to update.
The whisper
The whisper is what the desk actually expects, as distinct from the published consensus. It has no official source; it is inferred. Three ways to estimate it:
- Recent related data. If claims, ADP and the ISM employment indices all deteriorated after the consensus was collected, the whisper is below consensus.
- Price action into the release. If the 2-year yield has fallen 10bp over the last three days with no news, the market has partly priced a soft print. The whisper is soft, and a soft print will disappoint the people who expected a very soft one.
- Options pricing. The implied move in ES or in rates options into the event tells you how big a surprise is being insured against, though not which direction.
There is no point trying to be precise. The question is only: is the market leaning above or below consensus, and how much has it already moved? A print that matches consensus will rally if the whisper was lower and sell off if the whisper was higher.
Try it: For the next CPI release, write down three days in advance: consensus, the forecast range, the 2-year yield, and your estimate of the whisper (above, at, or below consensus) with one sentence of reasoning. After the print, record the actual, the 2-year move at 8:35, and whether the reaction matched your whisper call.
Recap
- 8:30 ET releases hit a thin pre-market and move fast; 10:00 releases hit a full market and move cleaner.
- Typical first-thirty-minute reactions are order-of-magnitude guides only; regime, positioning and event clustering change them by multiples.
- Most series are revised; you trade the first print, but build narratives only after revisions.
- Consensus is a lagged median of economist forecasts; the range and its collection date matter.
- The whisper is inferred from recent related data, pre-release price action and options pricing.
- The calendar is a prior, not a promise; holidays and shutdowns move it.
See it drawn
Original diagrams for the ideas on this page. Illustrative, not real market data.