How to read the FOMC dot plot:one chart, and the medium-term direction of gold

Every March, June, September and December, the Fed attaches a chart to its rate decision: 19 anonymous dots, each one a participant's view of where the fed funds rate should be over the next few years. This is the dot plot. It is one of the biggest market-moving items on decision day, and one of the most frequently misread charts in macro.

First, the most widespread misunderstanding: the dot plot is neither a forecast nor a commitment. It is 19 individual judgements at one moment in time; six months later the data change and the whole chart can be redrawn. Treat it as "the Fed telling you how many cuts are coming" and everything you derive from it will be skewed.

1. What the dot plot actually is: 19 dots, four times a year

Its formal home is a section of the Summary of Economic Projections (SEP), published only at the March, June, September and December meetings. It sits in the same document as the projections for GDP, unemployment and PCE inflation, and is released at the same moment.

Nineteen people draw the dots: the seven governors plus the twelve regional Fed presidents. One detail is routinely missed — only twelve of them vote, but all nineteen draw dots. Seven of the dots therefore come from people without a vote that year: no say on short-term policy, but plenty of influence on the shape of the chart.

The horizontal axis covers three horizons: the current year-end, the following two years, and the "longer run". The vertical axis is the mid-point of the fed funds target range. One grid step is normally 25 basis points.

Get the timing right as well: the statement is released at 14:00 US Eastern, with the dot plot and SEP at the same moment; the chair's press conference is usually at 14:30. In Beijing time that is 02:00 and 02:30 the next day during daylight time, 03:00 and 03:30 in standard time. This is not a trivial detail — plenty of short-term traders set an alarm for 02:00 and then discover in winter that the market only moves at 03:00.

2. The eight meetings of 2026: which ones have a dot plot

Not every meeting has one. The remaining meeting of 2026 is at the end of October and has no dot plot; the next one is December. The full calendar (source: FOMC calendar on federalreserve.gov, retrieved 2026-10-24):

Beijing times convert from 14:00 US Eastern: during US daylight time (8 March to 1 November 2026) that is 02:00 the next day, in standard time 03:00. January and December fall in standard time, hence 03:00.

The practical value of the table is that it tells you which meetings justify adjusting position size. The four meetings with a dot plot usually reprice the rate path for the next two years; at meetings without one, the move comes mainly from statement wording and the press conference, and is usually smaller in magnitude. The 27–28 October meeting is the latter kind, and the measurement in the next section is a useful comparison.

3. Reading it in three steps: median, dispersion, longer-run neutral rate

Headlines only ever quote one number — "the median shows two cuts this year". But a dot plot carries three layers of genuinely useful information.

Layer one, the median. Sort the 19 dots and take the middle one. It is a proxy for consensus and the number every headline cites. The problem: the market has usually guessed it before publication, so it rarely moves price unless it deviates a long way from expectations.

Layer two, dispersion. The gap between the highest and lowest dot in a given year. If the 2026 dots run from 3.25% to 4.50%, the committee's views span five cuts, and that disagreement destabilises market rate expectations and lifts gold volatility. If the 19 dots huddle together, disagreement is small and the dot plot carries essentially no information.

Layer three, the longer-run neutral rate. The last column of the table. It is not the policy of any particular year but the level at which participants think rates neither stimulate nor restrain the economy. A higher reading means the long-run real rate centre is rising — a slow variable for gold, but a clear one, and bearish; a lower reading the reverse. For medium-term allocators this column matters far more than this year's median.

4. The move is in the gap between the dot plot and market pricing

Before publication the market has already priced a probability for each meeting through fed funds futures — usually labelled FedWatch. So what sets direction is not the absolute level of the dots but the gap between the dots and market pricing.

• Dots more hawkish than the market (participants want fewer cuts, or none): the market immediately marks up short-rate expectations, the dollar and real yields rise together, and gold comes under pressure.
• Dots more dovish than the market: cut expectations are pulled forward, real yields fall, and gold usually firms.
• The two broadly agree: the dot plot generates no move of its own and money rotates into inflation data, employment data or risk sentiment.

That is why the same "two cuts" dot plot sometimes lifts gold 30 dollars and sometimes sinks it 40 — the difference is not the chart, it is what the market had already bet. You do not need a model to measure that gap: in the week before publication, financial media report "the market prices X cuts this year" every day. Write down that X and compare it with the median; the difference is the thing you are trading.

5. From dot plot to gold: one real-yield chain

The transmission is a single sentence: the dot plot changes expectations for the rate path → nominal and real yields adjust → the opportunity cost of holding gold changes → gold reprices. Gold pays nothing, so its competitor is the real return you can actually earn. The formula and current reading are broken down in the dollar index and real yields article: real yield = nominal yield − inflation expectation, and in practice you simply read the 10-year TIPS yield.

A current reading: on 2026-10-21 the 10-year nominal yield was 5.13% (FRED series DGS10) and the 10-year inflation expectation 2.32% (FRED T10YIE, 10-22 reading), a difference of about 2.81%; the 10-year TIPS real yield that day was 2.76% (FRED DFII10). Three numbers cross-checking each other is enough to show the formula works in practice.

But watch the frequency. In the same test, over the last 252 trading days gold's correlation with the dollar index was -0.374 while its correlation with daily changes in the real yield was only -0.118. In other words, the real yield explains medium-term direction, not intraday moves. Anyone expecting "real yields move 10bp, so gold drops 20 dollars" will be slapped around by noise all day. The trading logic of inflation data itself is covered in what gold does on CPI day.

6. Measured: 16 September 2026, what happened in those 60 minutes

The 15–16 September meeting was the third of 2026 with a dot plot, and the statement was released at 18:00 UTC (02:00 Beijing on 17 September). Using the 60-minute and daily bars of the continuous COMEX gold futures contract (Yahoo Finance GC=F), here is what actually happened around the release.

Isolating the release hour (60-minute bars, all times UTC):

Three readings you can use directly:

1. The release hour had a 115.8 dollar range, 2.05 times the whole of the previous day (56.6 dollars). One 60-minute bar equalled two normal days of movement.
2. From the 17:00 close at 4,399.0 to the 19:00 low at 4,273.3, gold fell 125.7 dollars in two hours (-2.86%). Direction was set within the first hour after the statement and then extended.
3. The next day was not calm. The 17 September range was still 128.8 dollars, and price rebounded from the low of 4,273.3 to 4,423.3, a move of +3.51%. The impact of the event does not end in an hour; it carries into the following session.

Sources: Yahoo Finance GC=F (continuous COMEX gold futures) daily and 60-minute bars, retrieved 2026-10-24. Continuous futures carry a basis against spot XAU/USD, so absolute prices differ slightly from the spot quotes elsewhere on this site; the table uses only relative changes within one series and makes no cross-instrument splicing. 16 September fell in US daylight time, so 18:00 UTC is 14:00 US Eastern, matching the statement release.

7. The 27–28 October meeting: no dot plot

The next meeting is 27–28 October (statement at 02:00 Beijing on 29 October, press conference at 02:30). Per the Fed calendar this one publishes no SEP, and therefore no dot plot; the last dot plot of 2026 comes with the 8–9 December meeting.

That does not mean there is no move, but what you trade changes:

• No path repricing. Without the 19 dots, market expectations for the next two years do not shift wholesale because of this chart — that is the difference in magnitude.
• Marginal wording changes get amplified. With no dots to read, the market compares every adjective in the statement with the previous one, and the appearance or removal of phrases like "moderate", "further" or "data dependent" all get traded.
• The press conference Q&A can rewrite direction. In September direction was set in the first hour (see above), but the second hour still carried a 53.6 dollar range, and the back-and-forth during the press conference gets traded too; without a dot plot this segment carries more weight.

One practical time-zone consequence for overseas Chinese readers: a 02:00 Beijing release means Asian-hours traders either stay up or skip the trade. If you are in Singapore or Malaysia (same zone as Beijing) the time is identical; Sydney is +2 hours (04:00), London is 19:00 the previous day, New York 14:00 — the same event, with completely different living costs depending on where you are.

8. How to use it from 60 seconds to 5 minutes: set a bias, not a single trade

The classic mistake in applying the dot plot to short-term trading is "hawkish dots, so short immediately with a 60-second expiry". The data does not support that.

Start with the baseline volatility. Using 475,052 quotes from the site's quote database (Swissquote source) between 2026-09-24 13:45 UTC and 2026-10-24 13:45 UTC, cut into 15-minute segments, 2,105 segments in total:

Sample taken 2026-10-24 21:45 Beijing; source is tick-level Swissquote quotes in the site's quotes table, aggregated into 15-minute buckets, each requiring at least three quotes. There was no FOMC meeting in that 30-day window, so the 02:00 hour reflects "a day with no event".

The table shows something counter-intuitive: on ordinary days the 02:00 hour is one of the quietest of the whole day (median 3.90, half the 22:00 figure). The 115.8 dollars on event day was not "this hour is normally volatile" but "the event dragged it from 3.9 to 115.8". This is exactly where averages mislead — they blend extremes and ordinary values into a meaningless middle.

So use it on two levels:

1. Use the dot plot to set a bias for hours to days. A hawkish dot plot biases you short, but the entry belongs after the event, once volatility has settled, not at 02:00 on the dot.
2. Do not trade 60-second to 5-minute expiries inside the event window. The spread itself may not change (this site's Swissquote feed quotes a fixed spread, measured at 0.658–0.700, see the spread comparison test), but fills will skew — gaps and slippage do not disappear because the spread is fixed. On a 60-second contract with a 1 dollar target, a 0.658 spread already demands an 82.9% win rate just to break even; add one 10 dollar slippage event and the expected value of the trade is outright negative.

In practice, an executable rhythm: no new short-term positions in the 15 minutes before the meeting; observe only between 02:00 and 02:30; after 02:30 wait for the first 15-minute bar to close and check whether it closes outside the pre-event range before deciding direction. For the logic of session selection itself see the three-session measurement; for a live board use the real-time session table. To work out the true cost of one lot inside an event window, use the gold P&L calculator.

9. Frequently asked questions

Q1. How often is the dot plot published, and exactly when?
A. Four times a year, only with the March, June, September and December meetings, alongside the Summary of Economic Projections. Release is 14:00 US Eastern — 02:00 the next day in Beijing during daylight time, 03:00 in standard time. The last dot plot of 2026 comes with the 8–9 December meeting (03:00 Beijing on 10 December).

Q2. Can the dot plot forecast the gold price?
A. No. It gives participants' view of the rate path; the gold price also carries risk sentiment, central bank buying and the dollar. In this site's 2026-10-24 measurement, the correlation between gold and the dollar index over 252 trading days was only -0.374 (R² = 0.140), meaning a single macro variable explains under 15% of daily variance. The dot plot is good for setting a directional bias, not for timing a single entry.

Q3. Which matters more, the median or dispersion?
A. Dispersion. The median is largely priced before publication and generates no move unless it deviates sharply; dispersion shows the committee's internal disagreement, and wider disagreement means an uncertain path and higher gold volatility. Medium-term allocators should also watch the longer-run neutral rate column.

Q4. Does the gold spread widen at the FOMC release?
A. It depends on the quote source. This site's Swissquote feed is fixed-spread: across 721,980 quotes from 2026-10-19 00:00 to 10-24 21:45 Beijing time, the daily median stayed between 0.662 and 0.700 with no systematic widening in event windows. Floating-spread platforms widen visibly at the release. Either way, the real risk is slippage and gaps, not the spread itself.

Q5. Should a 60-second trader wait for the FOMC?
A. Not inside the release window. The 16 September 2026 measurement shows the release hour (02:00–03:00 Beijing) carried a 115.8 dollar range across 60 minutes, 2.05 times the whole previous day; at that magnitude the expiry price of a 60-second contract is decided by slippage and gaps rather than by your directional call.

Quotes on this page are a snapshot at 2026-10-24 21:45 Beijing time (UTC+8) from the XAUAD quote API /api/config, source Swissquote, sampled while the market was closed for the weekend. Historical volatility data comes from the site's quotes table and Yahoo Finance, with retrieval dates noted.