Central-bank minutes add detail after a policy decision: the discussion, risks and reasons behind it. Learn to distinguish that summary from a transcript, identify which information is new and match it to the meeting and publication dates.

What the minutes actually are

FOMC minutes summarize discussion and record decisions, attendees and formal votes. They are not a verbatim transcript; the fuller FOMC transcript arrives after about five years. The ECB’s monetary-policy account reviews conditions and discussion in an unattributed form. Do not infer which named official is behind an anonymous group or treat minutes as a new rate decision.

Compare the document with the same meeting’s statement and conference to identify added explanation, then with the previous account to examine changes between meetings. These are different reference points. Some formal facts are already public; additional reasoning is neither automatically new to the market nor a prediction of the next decision. How that policy feeds through to the currency is something I unpack in a piece on how the Fed decision moves the dollar.

Fed, ECB and Bank of England publication practices

This comparison covers monetary-policy documents at three institutions, not all central banks or their other committees. Confirm the actual issue’s release date, time and timezone. Meeting date and publication date are different; one fixed converted local hour is not a year-round schedule.

  • FOMC minutes: the regular-meeting rule is three weeks after the policy decision. Named formal votes may already be in that decision’s statement. Confirm the actual release timestamp rather than treating eight scheduled meetings as an exhaustive rule for every possible meeting.
  • ECB monetary-policy accounts are normally published about four weeks after monetary-policy meetings of the Governing Council. Not every Governing Council meeting is a monetary-policy meeting. Confirm the publication date in the ECB calendar; the account does not supply a named vote breakdown.
  • Bank of England MPC: Monetary Policy Summary and Minutes are published together with the announced decision. Check the vote split and named positions in that package. Do not create a separate three-weeks-later UK event from the Fed’s rule or describe a summary as the entire debate.

The takeaway is simple: you build a minutes calendar separately for each bank whose currency you follow. The most reliable dates are published by the institutions themselves, well in advance. How to weave those dates into a wider market-watching routine is something I cover in a guide to using the economic calendar.

What to look for in the text

Identify whether a passage reports staff analysis, participant views or a formal policy action. Record its topic, horizon and conditions. Mentioning an inflation risk does not itself constitute a vote for a hike, and considering a scenario does not mean adopting it. Read the surrounding argument, including passages that challenge your first interpretation.

Keep formal votes separate from anonymous discussion groups. In a hypothetical committee, three dissents for a hike differ from three for a cut, but one vote split cannot show that a wing is growing without a comparable earlier split and unchanged membership. Unanimity does not prove identical reasoning or confidence.

Classify the information before interpreting it: our fictional notes
Different claims need different earlier reference material
PassageEarlier informationClassification
Two voters preferred a cutAlready in the decision statementRepeated formal fact
Several participants discuss delayed-response riskNo corresponding discussion found in the earlier materialAdditional reasoning; no invented vote count
Easing possible if price pressure weakensCondition requires future dataConditional scenario, not a cut date
New inflation data after the meetingNot available during the meetingSeparate later information

This is a method illustration, not a real meeting summary or a measured surprise. Keep the source passage and earlier comparison. Not finding a fact in your notes does not establish that nobody in the market knew it.

Mentions do not count distinct people

Hypothetical model: four people discuss prices and three discuss employment. In case A, prices involve A, B, C, D; employment involves C, D, E. Two people are in both groups, giving five distinct people despite seven topic memberships. The letters are invented identifiers, not FOMC member labels.

Scroll sideways to see the full diagram

Two topics with shared participantsThe prices group has A B C D; employment has C D E. Count shared C D once: 4+3-2=5 people. In a separate case, disjoint A B C D and E F G give 7.Case A: prices (4) and employment (3)A BC DEShared C and D: count each person once.4 + 3 - 2 = 5 people: A, B, C, D, ECase B: A B C D / E F G = 7 people7 topic memberships need not be 7 people.
The circles overlap where the same people discuss both topics. The second case has equally large groups but no overlap.

Without knowing the overlap, adding group sizes does not establish the number of distinct people. For these two groups the union can contain four to seven people, if the wider population permits every case. Even correctly counting people making arguments does not establish the formal vote split. In minutes, preserve the topic and stated group type rather than inventing names.

Assumptions: Exact counts and identifiers are fictional. Both groups are known in A; they are disjoint in B. We do not reconstruct anonymous passages from actual minutes or a vote. One person can discuss more than one risk. Official material: Fed: what FOMC minutes contain.

What was known then, and what is known now

The July 30, 2025 FOMC statement already named Michelle Bowman and Christopher Waller as dissents preferring a quarter-percentage-point cut. The July 29–30 meeting minutes were released on August 20, 21 days after the decision. They repeated the formal vote and expanded the reasoning. Repeating those names revealed zero new dissents; examine what extra explanation was actually added. No currency-price response is measured here.

A participant majority does not determine the voting subset

Our fictional discussion has 19 participants: 12 with a vote and seven without. Ten prefer a future policy option. These counts are not assigned to a real meeting.

Same ten preferences, different voting subsets
Preference groupScenario AScenario B
Voting participants supporting option3 of 1210 of 12
Nonvoting participants supporting option7 of 70 of 7
All participants supporting option10 of 1910 of 19

Both scenarios have a majority of all participants, but A has only three supporters among twelve voters while B has ten. The overall count cannot recover the subset. A future preference also is not a formal vote on today’s action. Preserve whether the source says participants, members, staff or a recorded vote; do not translate several into our fictional ten.

Place the meeting, subsequent releases and account publication on separate timeline entries. Do not attribute later information to the participants, but consider it when judging whether an earlier argument remains relevant today. A hawkish interpretation does not determine exchange-rate direction. Compare communication with its prior context; I break it down in a piece on Fed chair public remarks. For the wider picture of how fundamental analysis links central-bank policy to currencies, see our fundamental analysis section.

Your next step before the next minutes

  1. Record actual publication dates. Take the Fed and ECB issue dates, times and zones from their official schedules and distinguish them from the meeting dates. Keep the Bank of England’s joint package separate from those later releases. A calendar label does not establish the size of a currency reaction.
  2. Read one set of minutes calmly, after the fact. Take the most recent FOMC minutes and go through them once, looking only for the four things: the vote split, the balance-of-risks language, the hints about the next move, and the degree of agreement. The exercise shows you where the real information sits in the text.
  3. Use two reference points. Compare the account with the same meeting’s statement and conference, then with the previous account. Record each changed argument and its conditions, with a source passage. Read nearby qualifications; one changed word alone is not proof of a policy turn or a trade signal.
  4. Separate observation from trading. Practice identifying added information without opening a position. If researching a strategy, predefine the observation window, execution rules and costs. A reaction chart cannot show that you could trade at its pre-publication price after learning the document. Follow an existing exposure plan and check actual account conditions.