Every quarter the same thing happens. The dot plot drops, someone screenshots it, and within about ninety seconds there is a chart floating around showing the median dot moved up or down by a quarter point, with a caption implying the Fed just told you what rates are doing next year. Then the market does something that seems to contradict the whole thing, and everyone acts surprised. The dot plot is one of the most misread pieces of information the Fed publishes, and most of the misreading comes from treating it as a plan when it was never built to be one.
I find it genuinely useful, but only once you understand what you are actually looking at. So here is how I read it, and more importantly, the ways I have watched people misread it and pay for it.
What a dot actually is
The dot plot lives inside the Summary of Economic Projections, the SEP, which the Fed releases roughly four times a year alongside the meeting statement. The SEP is a collection of forecasts from every participant on the committee, both the voting members and the regional presidents who are in the room. That last part matters and I will come back to it.
Each dot represents one participant's view of where the appropriate midpoint of the policy rate should be at the end of a given year, and then further out at what they call the longer run. So you get a column of dots for this year, a column for next year, one for the year after, and a longer-run column that is basically their guess at the neutral rate once everything settles.
The thing to burn into your head is the word appropriate. A dot is not a prediction of what the participant thinks will happen to the economy and therefore where rates land. It is where that person thinks rates should be if their own economic forecast comes true. Every dot is conditional on the forecaster's own view of growth, unemployment, and inflation, which are the other tables in the SEP that nobody screenshots. Two participants can have the same dot for completely different reasons, and one participant can hold a dot that assumes a world none of the others are forecasting.
Why the median gets the headline and the distribution carries the information
The number you see quoted is almost always the median dot, because the median is easy to point at and it collapses nineteen or so opinions into one clean figure. The problem is that the median throws away almost everything that is actually informative.
What I look at first is the spread. A column where every dot is bunched within a quarter or half point of each other tells you the committee broadly agrees, and the median means something close to consensus. A column where the dots are smeared across a full two percentage points tells you the committee has no idea, and the median is an accident of where the middle person happened to land. Those are two completely different situations that produce the exact same headline number.
A few things I actually pull out of the distribution:
- The width. Tight cluster means conviction. Wide scatter means the committee is split and the median is fragile, so small data surprises can swing it hard next time.
- Which tail is fat. If the dots skew so that a handful sit well above the median while the rest bunch below, the risk is asymmetric. The median can hold steady while the balance of the committee is quietly leaning one way.
- The gap between the near-term dots and the longer-run dot. That gap is the committee telling you how far they think policy is from neutral, which frames whether they see themselves as restrictive, easy, or roughly balanced.
- Movement of the whole cloud, not just the middle. Sometimes the median is unchanged but the entire distribution shifted, because dots moved on both sides of the median in a way that cancels at the middle. The median lies about that. The cloud does not.
You cannot get any of this from the one number in the headline. You have to look at the actual chart and count.
Forecasts, not commitments
Here is the misread that costs people the most. A dot is a snapshot of one person's opinion on one afternoon, conditional on everything they knew that day. It is not a vote, it is not a promise, and no participant is bound by their previous dot. When the next SEP comes out and the dots have moved, that is not the Fed breaking a commitment. That is forecasters updating on new data, which is the entire point of a forecast.
The committee itself has been fairly blunt that the dots are not a plan and should not be read as the committee's collective intention. There is no mechanism by which the median dot becomes policy. Policy comes out of the meetings, one decision at a time, driven by the data that actually arrives. The dots are the committee thinking out loud about what they would do if the world behaves the way they currently expect, and the world rarely does.
This is why reacting hard to a dot shift is usually a mistake. If the median moves from, say, two cuts penciled in for next year to one, people treat it as the Fed removing a cut, like something was taken off the table. Nothing was taken off any table. The median forecaster revised an opinion about a hypothetical. The path can revise right back the next quarter if inflation cools or the labor market cracks, and historically it moves around a lot.
How to weigh the SEP against the market
The dots are one forecast. The rates market is another, and the market is putting real money behind its view every day, repricing continuously as data comes in, while the dots update roughly four times a year and go stale almost immediately. So when the two disagree, I do not automatically side with the Fed.
The market and the dots answer slightly different questions. The dots say where the committee thinks rates should go if their forecast holds. The market prices the probability-weighted average of every path, including the ugly ones where something breaks and the Fed cuts fast. That is a real reason the market path often sits below the dots. The market is pricing in tail risk that no single dot represents, because a dot is a point estimate and the market is a distribution.
The practical workflow I use when a fresh SEP lands:
- Read the median for context, then immediately ignore it and look at the actual scatter. Note how tight or wide each year's column is.
- Check the other SEP tables. If the dots moved, find out whether the growth, unemployment, or inflation forecasts moved with them. A dot shift that lines up with a real forecast change is more durable than one that does not.
- Pull up where the rates market is pricing the same horizon and measure the gap. A large, persistent gap between dots and market is information, not a mistake to be corrected. It usually means the market sees risks the committee is not putting in a point estimate.
- Ask what would have to be true for the dots to be right versus the market. Which side is assuming inflation is stickier, which side is assuming the labor market holds. Then watch the incoming data as a scoreboard for that specific disagreement.
- Remember who is in the room. The dot cloud includes non-voting presidents, so the distribution can lean more hawkish or dovish than the people who will actually cast votes at the next meeting.
None of this gives you a trade by itself. What it gives you is a way to stop getting whipsawed by a headline that says the Fed just did something it did not do. Most quarters the honest read of a dot shift is that a handful of forecasters nudged their opinions and the median caught it, and the durable signal is whatever the data does over the following months, not the dots themselves. When I catch myself about to react to a median moving a quarter point, the useful question is usually just whether the underlying economic forecasts moved too, and whether the market already priced it before I finished reading the chart.