CPI at 08:30 — you knew that. What no calendar publishes is what gold did in the fifteen minutes after the last two hundred of them. One of those numbers is worth sizing around. The other is close enough to a coin flip that acting on it is guessing.
TEST WHAT YOU KNOW →Every calendar on the internet publishes the same three things: the time, the consensus, and an impact icon somebody assigned by reputation. None of them go back afterwards and measure what the price did.
So "high impact" means whatever a vendor decided it means. Crude oil inventories carries a red flag almost everywhere. For gold it barely registers. Meanwhile FOMC minutes — a release with no consensus figure at all, nothing to be surprised by — is among the largest movers on the calendar.
That gap is the entire product. We measure every red-flag release against the window that followed it, publish the result whether or not it flatters us, and let the flags become an output of the data rather than an opinion about which events sound important.
And here is the part that should decide whether you trust us. Our own glossary states that Change of Character — a concept this entire methodology rests on — measured 51.6% on n=403, and calls it a coin flip. That is on the public site, not buried in a disclaimer. A platform that only shows you its wins is telling you something about itself.
Start with the thing that makes gold different from everything else on a screen. It pays you nothing. A bond pays a coupon. A stock pays a dividend or retains earnings. Cash in a money market pays overnight. Gold sits there.
So the question is never "is gold going up." It is what did you give up to hold it — and what you gave up is the yield you could have earned instead, adjusted for what inflation will do to that yield. That is the real interest rate.
Now the useful part. Every release on this page matters only insofar as it moves one of those two terms. Not because it is important-sounding. Not because a calendar gave it three stars. Because it changed the nominal rate the market expects, or it changed what inflation is expected to do to that rate.
Once you hold that, a thing that looks like a contradiction stops being one. Gold routinely falls on a high inflation print. If CPI came in hot but below what economists forecast, inflation expectations come down — real rates rise — and the mechanism runs the opposite way to the headline. The number went up. The expectation went down. The expectation is what was priced.
Which is the second thing worth internalising: markets price the expectation, not the number. By the time a release lands, the consensus is already in the price. The only new information is the gap between what arrived and what was expected — the surprise — and that is the single most useful field on any calendar.
Why we keep repeating this: almost every mistake people make around economic data comes from trading the number instead of the deviation. It is the same error in eight different costumes, and each tab below shows you the costume it wears for that release.
08:30 ET · monthly, around the 12th · RED FLAG
CPI is the most-watched read on the inflation half of the real-rate equation. A hotter print raises the odds the Federal Reserve keeps policy tight, which lifts real yields and raises the cost of holding an asset that pays nothing.
Gold frequently rises on a hot CPI. If the print is hot but below what economists forecast, expectations fall even though inflation rose — and the mechanism runs the other way. The surprise, not the number.
Core CPI, which strips food and energy, often moves price more than headline. The Fed weights it more heavily because it is less noisy.
08:30 ET · first Friday of the month · RED FLAG
The Federal Reserve has a dual mandate — employment and prices. Payrolls is the clearest monthly read on labour market slack. A strong number says the economy can absorb higher rates; a weak one pulls forward expectations of cuts.
Payrolls arrives with three numbers at once — the headline count, the unemployment rate and average hourly earnings. They regularly disagree, and which one the market decides to trade is not knowable in advance.
Revisions to the prior two months are published alongside. A strong headline paired with heavy downward revisions is a materially different report from the headline alone.
14:00 ET · eight times a year · RED FLAG
By the meeting date the decision itself is typically priced with high confidence. What is not priced is the framing — how close the vote was, what conditions would change it, and how the risks are characterised.
No consensus forecast exists for a text release, so there is no surprise to compute. Yet minutes routinely produce large moves. A release with nothing to be surprised by can still reprice the curve — which is why "surprise" and "impact" are separate ideas.
The dot plot at quarterly meetings, and any change in the characterisation of risks. A single reworded sentence can move rate expectations more than the decision.
08:30 ET · monthly, near CPI · RED FLAG
PPI measures prices at the producer level, upstream of the consumer. It is treated as a partial leading indicator for CPI, so a surprise here shifts expectations for next month's consumer print.
Pass-through from producer to consumer prices is incomplete and slow. PPI is a weaker signal than CPI, and market reaction is correspondingly smaller and less consistent.
When PPI lands the same week as CPI, the second release usually dominates — the first is often faded once the second confirms or contradicts it.
08:30 ET · monthly, end of month · RED FLAG
Core PCE is the measure the Federal Reserve targets explicitly. It differs from CPI in weighting and in how it handles substitution, and it typically runs cooler.
It arrives weeks after CPI covering the same period, so much of its content is already inferable. The reaction is often muted precisely because the market has already done the arithmetic.
The gap between Core PCE and Core CPI. A widening divergence changes how much weight to place on the earlier CPI print next month.
08:30 ET · with payrolls
Published alongside payrolls from a separate household survey. Where the two disagree, the market has to choose which to believe, and that choice is not predictable.
The rate can rise for a good reason — more people entering the labour force — or a bad one. The headline number does not distinguish them; the participation rate does.
Participation rate alongside the headline. A rate rising on higher participation is a very different signal from one rising on job losses.
08:30 ET · monthly, mid-month
Consumer spending is roughly two-thirds of US output, so retail sales is a fast read on whether growth is holding up — which feeds directly into how much room the Fed has.
The headline includes autos and fuel, both volatile and both driven by factors unrelated to consumer strength. The control group is the number that matters and it often tells a different story.
The control group, which excludes autos, fuel, building materials and food services. It feeds GDP directly.
08:30 ET · quarterly, three estimates
The broadest measure of output. It sets the backdrop against which every other release is interpreted, though it arrives too late to be news on its own.
Published three times for the same quarter — advance, second and third estimates. The advance reading moves price most because it contains the most new information; later revisions rarely register.
The advance estimate. By the second revision the market has usually moved on.
Loading…
Each row is a completed fifteen-minute window, measured from a closed candle after the release. Published once the window closes and not before — a partial window reported as a complete one is the one thing this record cannot contain.
This table writes itself. Every red-flag release is measured the moment its window closes and appended here automatically. No row is added by hand, none is removed, and the ones that contradict us stay — a release where the surprise pointed one way and gold went the other is the most instructive row on the page, and deleting it would make every other row worthless.
Six questions, and every one explains itself the moment you answer — right or wrong. The score is not the point; the explanation is. Each question is a real situation where the intuitive answer and the correct one come apart, which is exactly where money is lost.
Every red-flag release gets measured and posted. The daily structure read is free and updates all day at today's read.
◆ @ONLYGOLDFANSHQEntry, invalidation, targets and risk-to-reward — published before the open, graded against the close, wins and losses alike. On release days, a preview thirty minutes ahead and a measured result once the window closes.
SEE TODAY'S LEVEL — $149/MOMeasured results come from the engine — a completed window off a closed candle, written at the moment each release resolves. Nothing on this page is typed by hand.
Where a figure has not been measured, this page says so rather than estimating. An unmeasured number is indistinguishable from a measured one to a reader, which is why it is not published until it exists.
New to the vocabulary? The full glossary defines every concept the engine tracks — including where our own measurements found a term performed no better than chance.