What a Central Bank Speech Calendar Is Actually Worth

Between two policy meetings, the officials who set rates give speeches. They open conferences, take questions at universities, sit for interviews, address bankers' associations. A serious economic calendar lists every one of these, and a trader who has just discovered that page tends to react in one of two ways: treat every entry as an event and end up flinching at nothing all week, or conclude that none of it matters and stop looking.

Both are wrong, and they are wrong for the same reason — they treat the speeches as a single category. They are not. A small number of these appearances can move the price of gold in the minute they happen; the large majority cannot move anything at all, and you can usually tell which is which before the speaker opens their mouth. This piece is about how to sort them, and about what the sorting is actually good for, which is not what most people expect.

Why the speeches exist at all

A previous piece here argued that a rate decision is mostly not news, because the market has worked out the answer well before the day. That raises an obvious question: how did it work the answer out? The speeches are a large part of how. Committees do not like surprising markets on decision day — a surprise is a sign the communication failed — so the expectation is moved gradually, in public, in the weeks beforehand. The meeting is where the expectation gets settled. The speeches are where it drifts.

That framing tells you immediately what you are looking for. A speech matters when it changes what people expect the committee to do next. It does not matter when it repeats what the speaker has already said, when the speaker does not get a say, or when the topic is not policy. Most entries on the calendar fail all three tests at once.

The three questions

Before an appearance, three things decide whether it is capable of moving anything. None of them requires an opinion about the economy, and all three can be answered from the calendar entry itself plus a minute of checking.

1. Who is speaking?

Not every official on the calendar has a vote, and not everyone with a vote carries the same weight. The chair speaks for the committee and is read as the committee's position. Other members speak for themselves, and how closely the market listens to any one of them depends on whether they are currently seen as representative of where the committee is heading or as a known outlier at one end of it. An official whose views are already fully understood — who has said the same thing at every appearance for months — is close to unable to surprise anyone, whatever their formal standing.

The practical version: the chair always qualifies. Voting members qualify. Non-voters and permanent outliers usually do not, unless something in the second question changes that.

2. What is the venue, and is it about policy?

A speech titled with the words 'monetary policy' or 'the economic outlook' at a central banking conference is a different object from the same person's remarks at a community development event or a university commencement. Officials know exactly which of their appearances are policy appearances, and they say policy things in the ones that are. The calendar usually gives you the title of the speech; read it, because it is doing more work than it looks like it is.

The other half of the venue question is whether there will be unscripted questions, which is the part people miss and which the next section is entirely about.

3. Is there anything left to say?

This is the question that decides most cases and the one nobody asks. A speech can only surprise if the market's expectation is currently contested. When the committee's next move is genuinely uncertain — the data has been mixed, the members are visibly disagreeing, the implied probabilities are near a coin flip — a policy speech by a voting member is a live event, because the speaker has room to shift the balance. When the next move is settled and every member has said the same thing for weeks, the same speaker at the same podium is a non-event, because there is no room left for them to move.

The consequence is that the identical calendar entry is worth watching in one month and safely ignorable in the next. Any list of 'speakers who matter' that stays fixed is describing something that changes.

The prepared text is not where the move comes from

Here is a specific mechanic worth understanding, because it explains a shape you will otherwise find baffling: a scheduled appearance passes quietly for twenty minutes and then produces a sharp move out of nowhere.

Prepared remarks are frequently released to the wires at the moment the speech begins, and sometimes slightly before. If there is anything market-relevant in the text, it is therefore consumed in the first seconds — the reaction happens against the text, not against the speaking. What is not released in advance is the question-and-answer session afterwards, where the speaker is answering a journalist or a moderator without a script. That is where a phrase gets used that was not in the text, and that is where the headline comes from.

So the useful thing to know about an appearance is not only when it starts, but whether it has a Q&A attached and roughly when that begins. A calendar entry that says 'speech' and one that says 'speech and Q&A' or 'moderated discussion' are describing different risk profiles for the same person on the same day.

The blackout is the most useful entry on the page

For a period before each policy meeting, officials stop making public comments on policy. This is a published, scheduled rule rather than a courtesy, and it produces the one genuinely reliable structural fact available on this whole subject.

Inside that window, the expectation for the upcoming meeting cannot be moved by an official voice, because there are no official voices. It can only be moved by data. That is worth knowing for two reasons. It tells you what a quiet stretch means — the quiet is structural rather than a sign that interest has faded — and it tells you that any release landing in that window carries more weight than it would otherwise, because there is nobody available to contextualise it afterwards and the market has to reprice on the number alone.

It also produces a small, real edge in the other direction: the appearances immediately before the blackout starts tend to be the ones officials use to leave the market where they want it. Those are worth more attention than the same speaker three weeks earlier.

What the calendar is actually for

Now the uncomfortable part. Having sorted the speeches properly, you do not have a source of trades. You have something less exciting and considerably more useful, and it is worth being precise about what it is.

• Attribution. When gold moves sharply and nothing has been released, the first question is whether something was scheduled that you did not know about. Most of the time the answer is yes and it was a speaker. Knowing that is the difference between recording an event and recording a mystery, and a log full of mysteries teaches you nothing.
• Not being caught mid-size. The purpose of knowing an appearance is coming is largely so that a position taken twenty minutes beforehand is taken deliberately rather than in ignorance. That is a risk decision, not a trade idea.
• Reading the quiet correctly. A week with no scheduled voices because of a blackout is a different environment from a week with no scheduled voices by coincidence, and they call for different expectations about what will move things.
• Building a personal map. Over time, your own log will tell you which names have historically preceded movement in the instrument you actually trade. That finding is yours and does not transfer — it depends on your instrument, your session, and the period. It is also the only version of this that is worth anything, because it is measured rather than assumed.

What it is not for: predicting direction. Nothing above tells you which way a speech will push anything, and any process that claims to has smuggled in a forecast of what the official will say.

A five-minute weekly pass

The whole thing fits into one short session at the start of the week, and it degrades gracefully — done badly it still works better than not doing it.

• Open the calendar and filter to central bank speakers for the coming week. Write the week's appearances down in one list with their times converted into your own clock, because a time you have to convert in the moment is a time you will get wrong.
• Against each one, answer the three questions in a word each: who, policy or not, contested or settled. Most entries will collect a 'no' on the first pass and can be crossed out immediately.
• Mark which of the survivors have a Q&A, and note when that part begins if the calendar gives it.
• Note whether the week falls inside a blackout window. If it does, expect the list to be nearly empty and treat scheduled data as carrying more than usual.
• At the end of the week, go back and write what actually happened next to what you marked. This is the step that everybody skips and the only step that makes the exercise cumulative.

That last line is the same discipline the rest of this series keeps arriving at, and it is not a coincidence. A calendar read forwards is a list of times. A calendar read backwards, against what your instrument actually did, is the only way to find out which entries were ever worth your attention — and the answer will be a much shorter list than the page suggests.

Common questions

How much does a speech typically move gold?

Refusing to put a number on this, because any number would be invented. It varies by speaker, by how contested the outlook is at the time, by what else is happening that session, and by the period — a figure that was true one year is misleading the next. If you want a number, it has to be one you measured on your own instrument and your own session, which is precisely what the weekly log above is for. Be suspicious of anyone who offers you an average here; producing one requires choosing a sample, and the choice does all the work.

Which officials should I actually watch?

Deliberately not giving a list, because a fixed list is the wrong shape for the answer. Who matters depends on who currently has a vote, who is seen as sitting near the middle of the committee rather than at a known extreme, and whether the next decision is contested at all. Those change. What is stable is the method: chair always, voting members when the outlook is genuinely uncertain, everyone else rarely. Apply that each week rather than memorising names.

Is a paid speech-alert service worth it?

The schedule itself is public and free — central banks publish their own speaking calendars, and every serious economic calendar aggregates them. What a paid service is really selling is either speed on the headline or a summary of what was said, and both deserve a hard question: speed only helps if you had already decided what you would do with each possible outcome, and a summary written by somebody else is exactly the thing this series keeps arguing you should be producing yourself. Start with the free calendar and the log. If, after a couple of months of that, you can point to a specific decision you would have made differently with faster information, you will then know what you are buying and why.

The calendar shows the speech in a different time zone every week. Is that a bug?

Almost always it is daylight saving, moving on different dates in different countries, so the gap between your clock and the speaker's changes for a few weeks each year. This is a boring detail that costs people real money twice a year, because a session that was reliably in your afternoon quietly becomes an hour earlier. Convert the time into your own clock each week rather than relying on a remembered offset, which is why that instruction is in the procedure above rather than left implicit.

If most speeches do not matter, why not just ignore the whole calendar?

Because the cost of the pass is a few minutes and the cost of the alternative is attributing a scheduled move to something imaginary. The value here is almost entirely defensive, and defensive value is easy to underrate precisely because when it works nothing happens. The honest summary is that this is a small piece of process that prevents a category of confusion, not a source of trades — and it is worth doing on exactly those terms.