MEASURED · 8 FED DECISIONS
How much does gold move on Fed (FOMC) decision days?
At the 8 Fed rate decisions from September 2025 to July 2026, gold’s 2:00 p.m. ET minute (18:00 or 19:00 GMT) moved a median $16.36 from high to low, about 8 times the same minute on an ordinary weekday. It did not stop there. By 3:29, an hour into the Chair’s press conference, price had traveled a median $49.69 from where it stood at 1:59, and never less than $29.40.
The hour before the decision was no quieter than usual, and from 3:30 to 4:29 gold still ranged about 3.2 times as much as on an ordinary afternoon. Eight decisions is a small sample (n = 8).
Which days count, and at what time
The dates come from the Fed’s FOMC meeting calendar. Eight meetings ended inside our bar archive: September, October and December 2025, each a quarter-point cut, and January, March, April, June and July 2026, each a hold. Every statement page reads “For release at 2:00 p.m.”, and the Fed’s January 2026 notice says the Chair’s news conference “will start at 2:30 p.m.” An August 22, 2025 notation vote had no policy statement, so it is not counted.
2:00 p.m. ET is 18:00 GMT under US daylight time and 19:00 GMT in winter, so the December and January decisions sit at 19:00. The bar clock on this Exness feed is GMT, and the decision days confirm it: the 2:00 ET minute was more than twice the archive’s median minute at all 8 decisions, against 16.6% of ordinary 2:00 minutes.
Every decision, measured
| Decision (2:00 ET) | Fed action | First minute | vs ordinary 2:00 | Reach in 15 min | Reach by 3:29 |
|---|---|---|---|---|---|
| Sep 17, 2025 | Cut 0.25 pt, projections | $18.70 | 9.1× | $35.38 | $40.84 |
| Oct 29, 2025 | Cut 0.25 pt | $7.78 | 3.8× | $16.31 | $52.01 |
| Dec 10, 2025 | Cut 0.25 pt, projections | $20.48 | 9.9× | $25.43 | $44.31 |
| Jan 28, 2026 | Hold | $14.01 | 6.8× | $14.25 | $72.16 |
| Mar 18, 2026 | Hold, projections | $11.57 | 5.6× | $9.96 | $47.37 |
| Apr 29, 2026 | Hold | $9.11 | 4.4× | $22.05 | $29.40 |
| Jun 17, 2026 | Hold, projections | $42.24 | 20.5× | $103.22 | $153.92 |
| Jul 29, 2026 | Hold | $42.46 | 20.6× | $42.26 | $72.61 |
First minute: high minus low of the 2:00 ET bar. Ordinary 2:00: the median of the same minute on 235 other weekdays ($2.06). Reach: the farthest price got, up or down, from the 1:59 close, from 2:00 to 2:14 and from 2:00 to 3:29 ET. Projections: the meeting published the Summary of Economic Projections. The dates link to each statement.
8 of the 8 first minutes beat the busiest tenth of ordinary 2:00 minutes ($4.31 or more). The largest was $42.46 on July 29, 2026. The two largest were both holds: a decision to change nothing is not a quiet afternoon. 3 cuts and 5 holds are too few to compare one kind with the other.
The statement, then the press conference
Each part of a Fed afternoon, against the same clock window on ordinary weekdays:
| Window (ET) | Decision days (n = 8) | Ordinary (n = 235) | Ratio |
|---|---|---|---|
| 1:00 to 1:59, before | $14.03 | $14.14 | 0.99× |
| 2:00 to 2:29, statement | $31.80 | $9.43 | 3.4× |
| 2:30 to 3:29, press conference | $51.85 | $13.01 | 4.0× |
| 3:30 to 4:29, after | $35.07 | $10.86 | 3.2× |
Median of each day’s high minus low in the window. The windows are not the same length, so compare each row with its ordinary days, not the rows with each other.
The single biggest minute usually came with the statement: the largest one-minute range between 2:00 and 3:59 fell at 2:00 or 2:01 at 6 of 8 decisions, and inside the press conference (2:35 and 2:36) at the other 2. On ordinary weekdays 2:00 was the largest minute of those two hours 3.8% of the time.
But the range kept building after 2:30. The press-conference hour ran 4 times an ordinary 2:30 to 3:29, and all 8 beat the busiest tenth of ordinary afternoons ($31.39). At 6 of 8 decisions it moved the same way as the half hour after the statement. On January 28, 2026 price had gone almost nowhere (−$0.09) by 2:29, then moved +$71.04 in the press-conference hour. The hour before the decision was neither quieter nor busier than usual.
How far price runs from the 1:59 price
What a position held into the decision has to survive: the farthest move from the 1:59 close.
| From the 1:59 close | Fed decisions (n = 8) | Ordinary 2:00 (n = 235) |
|---|---|---|
| Median reach by 2:14 | $23.74 | $5.41 |
| More than $10 by 2:14 | 7 of 8 | 18.3% |
| More than $20 by 2:14 | 5 of 8 | 4.3% |
| More than $5 both ways by 2:14 | 5 of 8 | 4.7% |
| Median reach by 3:29 | $49.69 | $12.47 |
| More than $40 by 3:29 | 7 of 8 | 7.7% |
The smallest 15-minute reach was $9.96 (March 18, 2026), and even that day had run $47.37 by 3:29. The largest was $103.22 within 15 minutes and $153.92 by 3:29 on June 17, 2026. That is not one bad bar: the separate five-minute download gives the same 2:00 to 2:14 range on all 8 days. At 5 decisions price went more than $5 each way within 15 minutes, so a buy stop and a sell stop $5 either side of the 1:59 price would both have filled.
Does the first move keep going?
Measured from the first minute’s open, the direction held an hour later at 7 of 8 decisions, against 53.9% of random minutes. That test has a head start: price has to undo the whole first-minute move, open to close, before it can fail. The fairer test starts at the first minute’s close, the earliest anyone could act.
| After the first minute | Fed decisions | Random minutes | Ordinary 2:00 |
|---|---|---|---|
| 5 minutes later | 3 of 8 | 49.3% | 55.3% |
| 15 minutes later | 6 of 8 | 49.2% | 52.3% |
| 60 minutes later | 7 of 8 | 49.4% | 51.5% |
This is the one result here that does not look like a coin flip, and it is thin. An hour after the first minute closed, price had kept going its way at 7 of 8 decisions. Across all 8, the median move in the first minute's direction was $20.98. The 95% interval on 7 of 8 runs from 52.9% to 97.8%. If every decision carried on at the ordinary-afternoon rate of 51.5%, 7 or more of 8 would turn up 4.2% of the time, and this study tested three horizons. In the first five minutes it went the other way at 5 of 8, a median $3.71 against. On the same test our CPI study found 6 of 11 and the NFP study 5 of 10.
Read it as a question to test again after the next eight meetings, not as an entry rule. A bot that followed the first minute at 2:01 would first have sat through that five-minute pullback, with fills that bar data cannot show.
The spread: no Fed effect in this data
MT5 stores one spread value per one-minute bar, so the widest tick can be wider. From 1:55 to 2:10 p.m. ET the widest stored spread was 160 points ($0.16), the archive median, at all 4 decisions to January 2026, and 260 to 280 at all 4 from March. Ordinary afternoons broke the same way: that window topped 160 points on 0.9% of other weekdays before February 2026 and 98.3% after. As in our CPI release study, that is this account’s spreads changing, not the Fed.
The minutes are a much smaller event
The Fed releases the minutes of each meeting three weeks after the decision, also at 2:00 p.m. (October 8, 2025 release). On the 8 minutes days in the window the 2:00 minute moved a median $4.37, about 2.1 times an ordinary 2:00 minute, at most $7.81, and price reached a median $8.61 within 15 minutes. It was never the largest minute of its day. If your news filter treats the minutes like a decision, it is pausing for far less.
What to do with this if you run a bot
Put the decision in your platform’s clock. On a GMT server the statement is 18:00 in summer and 19:00 in winter; on a GMT+2 or GMT+3 server, add the offset of the day. The statement comes on the second day of each two-day meeting on the FOMC calendar.
Stop opening trades by 1:55 p.m. ET and stay out until at least 3:30. The longest elevated run from 2:00 was 40 minutes, but the press-conference hour ran 4.0 times an ordinary 2:30 to 3:29, and every decision had moved price at least $29.40 from its 1:59 level by 3:29. A half-hour pause, the kind that covers CPI or NFP, ends just as the press conference starts.
Treat the rest of the afternoon as a different market. From 3:30 to 4:29 gold still ranged 3.2 times as much as in the same hour on an ordinary day. A bot that trades after 3:30 on a Fed day needs stops and size set for that range, or it should sit out the rest of the US afternoon.
Decide about open positions by 1:59. 7 of 8 decisions moved price more than $40 by 3:29. A stop sets roughly where the trade ends, and a fast minute can fill it worse. If you hold through, size for $153.92, not the $49.69 median.
Check the log, not the settings. On the next Fed day, confirm in your EA’s log that it stopped opening trades by 1:55 ET in server time and did not restart at 2:30. A news filter that reads the calendar in the wrong time zone pauses at the wrong hour. That goes for any bot, the Tech Kick Trading Bot included.
Trading from the US? Read whether US traders can use MT5 trading bots first: the instrument measured here is a spot gold CFD.
Method and limits
356,223 one-minute XAUUSDm bars from an Exness MetaTrader 5 feed, August 1, 2025 to August 4, 2026. The twice-the-median rule for “elevated” (the median minute is $1.80), both continuation tests and the 30,000-minute random control are those of our NFP release study. The ordinary control is the same 2:00 p.m. ET minute on 235 other weekdays, leaving out decision and minutes days; 11 weekdays with gaps in the bars were dropped.
8 events is a small sample. One symbol, one broker, one year, a spot CFD rather than futures. What the market expected from each decision is not measured, so nothing here says why a decision moved gold up or down. Every figure comes from one script, research-gold-fomc-decision.py, run on the same bar archive as our other studies.
Common questions
What time is the Fed decision in GMT?
The statement comes out at 2:00 p.m. Eastern: 18:00 GMT while the US is on daylight saving time, 19:00 GMT in winter. The Chair's press conference starts at 2:30 p.m. ET. In the year measured, US clocks went back on November 2, 2025 and forward on March 8, 2026. On a broker server at GMT+2 or GMT+3, add that offset.
Is gold quiet before a Fed decision?
Not in this data. From 1:00 to 1:59 p.m. ET gold ranged a median $14.03 on decision days and $14.14 on ordinary weekdays. The busiest hour before a decision, $57.70, came on January 28, 2026.
Does gold go up or down when the Fed cuts rates?
Either. The first minute fell at 2 of the 3 cuts measured and rose at 1; across all 8 decisions it was 4 up and 4 down. Direction depends on the decision against what the market expected, which this study does not measure.
Should a trading bot trade the Fed decision?
Nothing in this sample says it should. The first minute ranged from $7.78 to $42.46 with no way to know which in advance, and the press conference kept the range at about 4 times normal for another hour. The data supports a rule about exposure: flat, or sized for a $153.92 move, by 1:59 p.m. ET, and no new entries until at least 3:30.