Reading a Fed Statement: What Actually Changed Since Last Time

Most people read a Fed statement the way they would read a news article: top to bottom, once, looking for the point. That is the wrong way to read it, and it is wrong in a specific and fixable way. The statement is not written to be read fresh. It is written to be compared against the previous one — it keeps the same skeleton meeting after meeting so that the handful of deliberate edits stand out. Read on its own, it says almost nothing. Read against last time, it says whatever the committee wanted to say.

This piece is about how to do that comparison properly: what to line up, which parts of the document actually carry signal, and — the part that matters most for anyone tempted to trade it — why a changed word is not an edge on its own, and what it is good for instead.

Why the document has a stable shape

The statement is short, and it is short on purpose. Every sentence in it has been argued over by a committee that knows the text will be parsed word by word, and the structure is held constant from meeting to meeting precisely so that a change is unambiguous. When a paragraph keeps its position and one adjective inside it moves, nobody has to wonder whether the shift was intentional. It was.

That constancy is what makes the diff readable, and it is also why reading the new statement in isolation feels so uninformative. Most of the words in front of you are there because they were there last time. The information is in the delta, and if you have not got last time's text open, you cannot see it.

The mechanical part: set up the comparison first

Do this before the release, not after. The statements are published on the Federal Reserve's own site and the previous one is available at any time — there is no reason to be hunting for it in the minute you most need it. Have the prior text saved in a plain-text form you can paste into a comparison tool, and have that tool open. Any diff tool works; this is the same operation as comparing two versions of a file.

Two practical notes on the setup, both of which will otherwise cost you the first few minutes:

• Strip the formatting. Copying from a web page brings line breaks and non-breaking spaces with it, and a naive diff will report those as changes. You want a word-level comparison of the prose, not a character-level comparison of the markup.
• Keep the last several statements, not just the last one. The most useful readings come from seeing a phrase move across three or four meetings, and you cannot reconstruct that after the fact if you only ever kept the most recent.

News organisations publish redline comparisons of their own within minutes, and they are genuinely useful as a cross-check. They are not a substitute for doing it yourself, for two reasons: they arrive after the market has already moved, and they are edited — somebody has decided which changes are worth highlighting, and that decision is exactly the judgement you are trying to develop.

Which parts of the statement actually carry signal

Not every change is equally meaningful, and treating them as if they were is the most common way to over-read the document. The statement's sections do different jobs, and they should be weighted differently.

The economic assessment

The opening passage describes how the committee sees recent activity, the labour market and inflation. This is where the characterising adjectives live, and it is the section people quote most. It is also, on its own, the weakest signal in the document — largely because it is describing data that everybody has already seen. The committee is not telling you something new about the economy here; it is telling you how it has chosen to frame data you could have read yourself.

That framing still matters, but for an indirect reason: it is the justification the committee is building for what it intends to do later. A downgraded description of the labour market is not itself news about the labour market. It is news about which argument the committee is preparing to make.

The policy decision and the guidance around it

This is the part that matters most, and it is where a small edit does the most work. The decision itself is usually the least surprising thing in the statement — by the time the meeting happens, the market has generally worked out what the committee will do. What has not been settled is what happens next, and that is carried in the sentence or two around the decision: what the committee says it is watching, what it says would have to change, and how firmly it commits to anything at all.

Pay particular attention to changes in the strength of a commitment rather than its direction. Language that softens from a definite statement into a conditional one, or that acquires a qualifier it did not have before, is doing something deliberate. So is the disappearance of a sentence. A removed line is a change, and it is the one a top-to-bottom read is most likely to miss entirely, because there is nothing on the page to notice.

The vote and any dissents

The statement records who voted for the decision and who did not, and a dissent names the dissenter and what they would have preferred instead. This is the most concrete information in the whole document — it is a fact about the committee rather than a characterisation of the economy — and it is routinely skipped by people reading for the headline.

A dissent tells you something the prose is designed not to: that the decision was contested, and in which direction. A committee that is unanimous today and split next time has told you something about its own stability, regardless of what its adjectives did.

What counts as noise

The corollary of taking small edits seriously is that you have to be disciplined about which small edits you take seriously, or you will find meaning in everything. Some changes are genuinely mechanical:

• Dates, meeting references and the calendar housekeeping that changes every time by definition.
• Membership changes — the voting roster rotates, and a new name in the list is not a signal about policy.
• Standing boilerplate about the committee's objectives, which is reaffirmed rather than rewritten.
• Sentence reordering that leaves the meaning intact. This one is worth flagging because a word-level diff will light it up dramatically, and it is usually nothing.

The test that keeps this honest is to ask what the alternative wording would have been. If a change had no plausible alternative — if the date could only have been that date — it carries no information. If somebody in the room could have argued for a different word and lost, it does.

The thing that undoes all of it thirty minutes later

Here is the part that most changes how you should use any of this. The statement is not the end of the event. The chair's press conference follows shortly after, and it regularly reframes, softens or effectively contradicts the impression the statement created. A careful diff can be entirely correct about what changed on the page and entirely wrong about what the day meant, because the person who wrote the page then spent an hour explaining it.

This is not an occasional accident. It is closer to a design feature: the statement is a committee document constrained by consensus, and the press conference is where nuance that could not survive that consensus gets added back. The practical consequence is that the market's first move on the statement and its settled move after the conference are two different things, and the first is not a preview of the second.

If you have read the earlier piece on this blog about why the first candle after a release lies, this is that same problem in a slower and more legible form. There, the reversal came from mechanics — an empty book and a headline that got refined. Here it comes from a human being taking questions. The lesson is the same one and it is worth stating plainly: the initial reaction to a Fed statement is a reaction to a partial document.

Why a changed word is not a trade

It is tempting to conclude that if you can spot the meaningful edit, you can act on it. You cannot, at least not in the way that framing implies, and it is worth being blunt about why.

The diff is not scarce. The statement is machine-readable, the previous one is known in advance, and comparing them is exactly the kind of task that is trivial to automate. Whatever the text changed, that change was parsed and acted on before a person could finish reading the first paragraph. There is no version of a human reading race that ends with you in front. Any strategy whose premise is 'notice the change quickly' is competing in the one dimension where the competition is strongest.

There is a second reason, and it is the more interesting one. Even a correctly identified change has an ambiguous price implication, because the market does not trade the change — it trades the change against what it already expected. A softening that everybody anticipated can move price in the opposite direction to the one the words suggest, simply because the anticipated version was already in the price and the actual version was milder than the anticipation. Reading the document tells you what it says. It does not tell you what was expected of it, and the second quantity is the one that sets the reaction.

What the diff is genuinely good for

All of which sounds like an argument for not bothering. It is not. The diff is valuable — just not as a signal generator on the day.

Its real use is cumulative. A single statement compared to a single predecessor is a data point with a lot of noise around it. The same comparison kept across a run of meetings turns into something much more useful: a record of the direction the committee's language has been travelling, at what pace, and whether the press conferences have been reinforcing that direction or repeatedly walking it back. That is a picture nobody can construct in the minutes after a release, and it is not in competition with anything faster, because it is not about speed at all.

The practical version is a short log kept per meeting. For each one, record the handful of edits you judged meaningful and why, what the vote looked like, what the press conference did to the impression, and — this is the entry people skip and the one that does the teaching — how price actually behaved over the following days, not the following minutes. Keep it for a few meetings and the entries start correcting each other. You will find that some changes you flagged as important turned out to matter to nobody, and that is the most useful thing the log can tell you.

It also makes you a better reader of the next one. Knowing that a particular qualifier has been softening for three meetings changes what the fourth appearance means. That context is not available to a first-time reader of a single statement, and it is the closest thing to an edge available here — not a faster reaction, a better-informed one.

The honest summary

Diff the statement rather than reading it. Weight the policy language and the vote above the economic description. Discard the mechanical changes on the alternative-wording test. Do not treat the result as a signal, because you are not going to win that race and the sign of the reaction is not determined by the text anyway. Keep the log, wait for the press conference, and let the understanding accumulate — which is a slower answer than most people want and the only one that survives contact with how this event actually trades.

FAQ

Should I just read the redline a news outlet publishes?

Read it, but after doing your own. The published redlines are accurate and fast, and they are a good check on whether you missed something. The problem with using one as your only source is that it comes pre-judged: the changes an editor chose to highlight are the ones they thought mattered, and developing your own sense of what matters is the entire point of the exercise. There is also a subtler cost — reading somebody else's summary is much easier than doing the comparison, so in practice it becomes the only thing you do, and the cumulative log never gets built.

How much does the projections material change this?

At the meetings that include the committee's projections, quite a lot — and it is worth knowing which meetings those are in advance, because it changes the character of the event. The projections are a separate document with a separate kind of information in it, and they routinely dominate the statement's own wording on the day they appear. If you are keeping the log described above, note which meetings had them; comparing a projection meeting to a non-projection one without flagging the difference will produce conclusions that do not hold.

Does this apply to other central banks?

The method does; the specifics do not transfer. Other central banks publish statements with their own stable structures, their own conventions about what gets said where, and their own relationship between the written statement and whatever press event follows it. Some put more weight on the accompanying materials than on the statement itself; some publish a vote breakdown, some do not. The diffing procedure is the same operation in each case, but the judgement about which section carries the signal has to be rebuilt for each institution, and assuming it carries over is a reliable way to misread one.

I trade gold. Is the Fed statement the single most important thing on my calendar?

It is unambiguously in the top handful, for the obvious reason that gold pays no yield and is priced against the alternatives that do. But 'most important' is a less useful question than it sounds. Deliberately not ranking these, because a ranking would imply a precision that varies with what the market currently cares about — in some periods inflation prints dominate everything, in others they pass with a shrug. The stable advice is the calendar-first routine covered elsewhere on this blog: know which events are scheduled, size and position around them rather than reacting to them, and let the log tell you over time which ones actually move your instrument.

If I cannot trade the release, what should I do during it?

Read, log, and manage what you already have — which is not a consolation prize. The decisions that matter around a scheduled event are mostly made before it: whether to be in a position at all, at what size, and where the risk sits if the move goes the wrong way. Once the statement is out, the useful work is observational. Watch what the market did with a change you thought was meaningful, note where you disagreed with it, and wait for the press conference before concluding anything. That is a genuinely productive half hour, and it compounds in a way that a rushed trade into a thin book does not.