Gold on news days
Three scheduled US releases move gold harder than anything else on the calendar: the jobs report, the inflation print, and the Fed’s rate decision. Plenty of sites will tell you that. This page shows you the part they cannot - what a specific rules-based strategy actually did on each of those dates, every leg, winners and losers.
Date taken from the official BLS schedule, not estimated. What NFP does to gold.
News days vs ordinary days
liveBoth averages are per day that actually traded, over the same window, so the two sides are counted the same way. Releases where nothing closed are excluded from the news average for exactly that reason - the ordinary-day figure can only contain days that traded, so including empty releases on one side only would stack the comparison. Break-evens are excluded from both win rates, as everywhere else on this site. Every figure on this page recomputes from the live trade log and refreshes on its own about once a minute.
Every verified release, day by day
Tap a day to see every leg that closed on it.
CPI12 August 20260 legs0p
Source: the Bureau of Labor Statistics public data API, and the New York Fed for the policy rate - the primary publishers, read directly. BLS revises its estimates, so a figure here is the current official reading for that month rather than the number that printed on the morning. Actual figures only: analyst consensus is a commercial product we hold no licence for, so no “vs expected” is shown rather than one being guessed.
No position closed on this date.
NFP7 August 20261 leg+544.6p
Source: the Bureau of Labor Statistics public data API, and the New York Fed for the policy rate - the primary publishers, read directly. BLS revises its estimates, so a figure here is the current official reading for that month rather than the number that printed on the morning. Actual figures only: analyst consensus is a commercial product we hold no licence for, so no “vs expected” is shown rather than one being guessed.
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
FOMC29 July 20263 legs-997.3p
Source: the Bureau of Labor Statistics public data API, and the New York Fed for the policy rate - the primary publishers, read directly. BLS revises its estimates, so a figure here is the current official reading for that month rather than the number that printed on the morning. Actual figures only: analyst consensus is a commercial product we hold no licence for, so no “vs expected” is shown rather than one being guessed.
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
CPI14 July 20262 legs+1,553.4p
Source: the Bureau of Labor Statistics public data API, and the New York Fed for the policy rate - the primary publishers, read directly. BLS revises its estimates, so a figure here is the current official reading for that month rather than the number that printed on the morning. Actual figures only: analyst consensus is a commercial product we hold no licence for, so no “vs expected” is shown rather than one being guessed.
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
NFP2 July 20261 leg+376.9p
Source: the Bureau of Labor Statistics public data API, and the New York Fed for the policy rate - the primary publishers, read directly. BLS revises its estimates, so a figure here is the current official reading for that month rather than the number that printed on the morning. Actual figures only: analyst consensus is a commercial product we hold no licence for, so no “vs expected” is shown rather than one being guessed.
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
FOMC17 June 20264 legs+1,783.7p
Source: the Bureau of Labor Statistics public data API, and the New York Fed for the policy rate - the primary publishers, read directly. BLS revises its estimates, so a figure here is the current official reading for that month rather than the number that printed on the morning. Actual figures only: analyst consensus is a commercial product we hold no licence for, so no “vs expected” is shown rather than one being guessed.
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
CPI10 June 20266 legs+160.2p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
NFP5 June 20261 leg+372.4p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
CPI12 May 20262 legs-104.3p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
NFP8 May 20260 legs0p
No position closed on this date.
FOMC29 April 20262 legs-493.8p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
CPI10 April 20260 legs0p
No position closed on this date.
NFP3 April 20260 legs0p
No position closed on this date.
FOMC18 March 20264 legs-1,121.5p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
CPI11 March 20264 legs-1,038.7p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
NFP6 March 20264 legs-1,282.2p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
CPI13 February 20261 leg+430p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
NFP11 February 20264 legs-1,696.9p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
FOMC28 January 202614 legs+2,529.5p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
CPI13 January 20261 leg0p
This is the span between the highest and lowest price our own orders actually transacted at that day - not the day’s true high and low, which we do not hold a price feed for and will not estimate.
NFP9 January 20260 legs0p
No position closed on this date.
This is why we tell members to trade light on news days
Reduce-risk reminders go out before every high-impact release, and have done since long before this page existed. The numbers above are the measurement that backs them: a news day has averaged +63.5p against +723p on an ordinary day, at a lower win rate (47.9% against 61.8%).
The reason is mechanical rather than mysterious. A scheduled release removes liquidity for a few minutes and widens the spread, so price can jump straight through a level instead of trading at it. A stop does not guarantee a fill at the price it was set - in that window it fills at whatever is there. Every part of a rules-based system that assumes an orderly market is weakest exactly then.
Smaller size is the whole adjustment. The rules do not change on a news day and no trade is skipped; what changes is how much a bad fill can cost you. Position size is the only lever that works on a risk you cannot forecast.
General information about how we run the strategy, not personal advice or a recommendation about your own position size.
By release type
| Release | Releases | Traded | Net | Avg / traded day | Win / Loss / Break even | Win rate |
|---|---|---|---|---|---|---|
| NFP | 8 | 5 | -1,685.2p | -337p | 3 / 6 / 2 | 33.3% |
| CPI | 8 | 6 | +1,000.6p | +166.8p | 7 / 6 / 3 | 53.8% |
| FOMC | 5 | 5 | +1,700.6p | +340.1p | 13 / 13 / 1 | 50% |
Read the counts before the percentages. A win rate computed from two or three closed legs is not a win rate, it is a coin toss with a decimal point - which is why the number of releases and the number that actually traded are printed next to it rather than left off.
Upcoming releases
How to read this page
The sample is small, deliberately. Covers 21 verified releases since 2026-01-09. Only dates confirmed against the official BLS and Federal Reserve schedules are counted, so the sample grows by a handful of days each month rather than being back-filled from an estimate.
A handful of releases proves nothing about the next one. What it does give you is something checkable: named dates, named trades, and pip figures you can hold against a chart. That is a different and much lower claim than “our strategy performs well on news”, and it is the only one this page makes.
No expectations are shown. Analyst consensus figures are commercial products from Bloomberg and Reuters surveys, and we hold no licence for them. Rather than guess the number everyone actually trades against, the page shows the released actual and stops there.