Use cases & endpoints
FOMC dot plot history — every projection vintage
The dot plot is a distribution, not a forecast, and the way it moves between rounds carries more information than any single median. Participants are anonymous and unranked, dots are not commitments, and a round where the median holds still while the distribution shifts underneath it is one of the more interesting things the committee produces.
The Summary of Economic Projections that contains it is published at four meetings a year, as a document rather than a dataset, in a layout that has changed over time. Reconstructing a history means collecting each release and normalizing it without flattening the distribution into a single number.
Hindcast keeps every SEP round as its own vintage with the full distribution intact, so a question about how projections changed between two rounds is answerable directly — including the cases where the median stayed put and the dots underneath did not.
hindcast — session
previewhindcast> compare two projection rounds for the same year-end
→ tool: cb_projections { bank:"FED", release:["...","..."], diff:true }
✓ both distributions with the median and the dot movement
source: FOMC Summary of Economic Projections, both vintages retained
What's covered
- Every SEP round retained as a separate vintage
- Full distribution kept, not just the median
- Projections by horizon including the longer-run value
- Release date attached to each round
- Layout changes across years normalized without losing detail
- Linked to the statement and minutes from the same meeting
Join the waitlist — this page's dataset is already selected.
Questions
How often is the SEP published?
At four of the year's scheduled meetings rather than all of them, which is why the projection history is a quarterly series while the statement history is not — a detail that trips up anyone trying to align the two.
Can dots be attributed to individual officials?
No, and no dataset should claim otherwise. The projections are published anonymously; anyone matching dots to names is inferring, and that inference does not belong in a data layer.
Related
Join the waitlist. Vote the roadmap. First in, first served.