Where to Get Reliable Economic News as an Investor

Helious Desk · 28 July 2026 · 11 min read

You get more economic news than you can read, and less of it holds up than you think. Access is not the problem. The agencies publish for free, wire copy is everywhere, and someone has charted everything. The problem is that most of what reaches you has been rewritten twice. It lands late, it skips the number that mattered, and the story around it was written before the data was out.

These are the ten questions we get asked most about sourcing economic news: where the numbers come from, how fast you genuinely need them, how to tell a good print from a bad one in about fifteen seconds, and how to stop a headline talking you into a trade the bond market never agreed with.

We build Helious, a live markets desk for rates and equities, so we have a view. We have tried to keep this useful whether or not you ever open it.

What this post answers

• What makes an economic news source reliable?
• Where does economic data actually come from?
• Are free economic news sources reliable enough to invest on?
• How fast does economic news need to be?
• How do I tell whether a report was good or bad?
• Which economic releases actually move markets?
• Where can I follow what the Federal Reserve is actually saying?
• How do I check the market's reaction instead of trusting the headline?
• Do Treasury auctions count as economic news?
• How do I build a news routine without drowning in noise?

1. What makes an economic news source reliable?

Reliable is not the same as famous. Plenty of well-known outlets get the number right and are still useless to an investor, because being right does not tell you what to do about it. What matters is duller than the name on the masthead.

Where the number came from

Can you trace it back to the agency that produced it in one step? A good source names the release, the body that published it, and the reference period. "Inflation came in hot" is not sourcing. "June CPI, released by the Bureau of Labor Statistics at 8:30am ET, 0.4% month over month against 0.3% expected" is.

When it was captured

Not when the article went up. When the number was captured. Those are two different times, and retail traders lose money in the gap between them.

What it is being measured against

A number on its own means nothing. You need it against the consensus forecast, against the prior after revisions, and against how noisy that series usually runs. Without all three you cannot say whether 0.4% is a shock or a rounding error. That last comparison is the one almost nobody gives you, and the surprise z-score page explains why it does most of the work.

One more thing, and it is about character rather than data. Does the source admit when it does not have something? A feed that shows a stale number without saying so, or invents a forecast to fill a gap, is worse than no feed at all, because you will act on it with confidence. We publish how everything is captured and scored on our methodology page, and a bit on who we are on the about page. Hold every source you use to the same standard.

2. Where does economic data actually come from?

Almost every economic headline you will read this year is downstream of about six organisations. Journalists add narrative, not data. Once you know the map you can go upstream whenever you want to.

The Bureau of Labor Statistics publishes most of its numbers at 8:30am ET: CPI and core CPI, PPI, nonfarm payrolls and the unemployment rate, with JOLTS later in the morning at 10:00am ET. The Bureau of Economic Analysis publishes GDP in three successive estimates, and PCE with core PCE, which is the measure the Fed actually targets. The Census Bureau covers the physical side of the economy: retail sales, durable goods, housing starts, building permits, new home sales and the trade balance.

The Department of Labor puts out weekly jobless claims every Thursday at 8:30am ET, which is the highest frequency read on the labour market anyone publishes. The Federal Reserve puts out industrial production at 9:15am ET and runs the whole FOMC apparatus. The rest is private and academic: ISM manufacturing and ISM services at 10:00am ET, the S&P Global PMIs, the Conference Board's consumer confidence, Michigan sentiment, ADP employment, the realtors' existing home sales, and the Atlanta Fed's GDPNow tracker.

The primary source is free and it is public, and nobody gets the number before the agency publishes it. People still pay for a feed, for a different reason: they do not want twenty tabs open at 8:30am, and they want to know what the number meant before the market has finished repricing.

Every series above has a page on the terminal with its full print history, the surprise on each print, and what the curve did afterwards. The data hub is the index, and the release calendar tells you what is due and when.

3. Are free economic news sources reliable enough to invest on?

For the number itself, usually yes. For the read, no.

Free wires are usually accurate about what printed. What they get wrong is everything wrapped around the number, and the wrapping is the part you trade on.

The forecast is the first thing to go missing. If you do not know what was expected, you cannot know whether a print was a surprise, and only the surprise moves anything. The cut they lead with is usually the least useful one as well. Headline CPI includes food and energy, which is exactly the part the Fed looks through, while the bond market cares about core CPI and core PCE. Month over month and year over year tell different stories from the same release, and outlets pick whichever is more dramatic.

Revisions get buried too. Payrolls rewrites the previous two months on every release, so a 90k beat sitting on top of a 120k downward revision is a weak report wearing a strong headline. Then there is the clock. Free coverage is optimised for readership, which means a writer, an editor and a delay. Perfectly reasonable journalism. Just not a data feed.

For most investors the answer is a split. Use free primary sources as ground truth, because they are the ground truth. Do not use free commentary as your read. Our live news feed is public if you want to see the difference, and every item on it carries its capture time.

4. How fast does economic news need to be?

Be honest about your horizon, because the answer changes completely.

If you rebalance quarterly, twenty minutes late is fine. A day late is often fine. What you need is accuracy and context, and you should spend nothing on speed.

If you are positioned into a print, the first ninety seconds are the trade. Treasury yields reprice in the first second, the equity futures follow, and by the time an article exists the move you are reading about has already happened. React to it then and you are trading the retracement without knowing that is what you are doing.

The uncomfortable middle is the swing trader who believes they are in the first camp and behaves like the second. If you find yourself reaching for the phone when a number lands, you are in the second camp. What you need then is capture measured in seconds with the context already attached, not a push alert carrying a headline.

Time your own source

Next time a tier one number lands, write down the exact second your source displays it. Compare that against the scheduled release time. If the gap runs to minutes, what you are reading is history. We aim to capture within seconds of the print, which is also why the calendar carries the exact scheduled time for every release instead of a vague window. When speed is about hearing rather than reading, as it is during an FOMC press conference, there is a live audio squawk instead.

5. How do I tell whether a report was good or bad?

Most people skip this part and end up collecting reassurance instead of reading data.

How big the surprise was

The level does not matter. The surprise does. But you cannot compare a raw surprise across releases: a 0.1 miss on CPI is enormous and a 20k miss on payrolls is noise. Divide the surprise by the standard deviation of that series' own past surprises and you get one number that works everywhere. That is the surprise z-score. It scores a print the second it lands instead of leaving it to be argued over for an hour. If the statistics are unfamiliar, the standard deviation page explains it without the maths.

What happened to the prior

Always read the revision line before you read the headline. Payrolls rewrites two months of history every time, GDP arrives in three estimates that can disagree materially, and retail sales is revised routinely and cheerfully.

What the internals say

Look at average hourly earnings and the participation rate inside payrolls, shelter and core services inside CPI, new orders and prices paid inside the ISM surveys. The headline is a summary written by a statistician, not a trader, and a strong headline with a soft interior gets faded within the hour.

One habit matters more than any of that. Before the print, decide what number would change your mind. If you cannot name one, you are not reading data. There is a full walkthrough with a worked example in our guide on how to read an economic release.

6. Which economic releases actually move markets?

Rank them by how much they typically move the front end of the Treasury curve. That is the cleanest measure of whether a print changed anyone's mind.

The four that stop everything

Clear the diary for CPI and core CPI, nonfarm payrolls with the unemployment rate, the FOMC decision with its statement and press conference, and core PCE. That is tier one.

The ones that matter only when they surprise

Retail sales and ISM services sit in tier two, along with PPI, which also feeds the PCE estimate, and GDP with JOLTS. Weekly jobless claims are there too, and get promoted to tier one the moment the labour market becomes the story.

Everything else

ADP employment, the S&P Global PMIs, ISM manufacturing, durable goods, industrial production, housing starts, building permits, existing home sales, new home sales, consumer confidence, Michigan sentiment, the trade balance and GDPNow are texture rather than trade. Useful for building a picture. Rarely worth a position on its own.

That ranking moves with the regime. When inflation is the fight, CPI is the only meeting that matters and the rest is filler. When the labour market cracks, a Thursday claims number can move the front end more than a CPI print did the week before. So do not trust a static list. Watch which release actually moves the 2-year yield and let the market rank them for you. Each series page keeps that history, so you can see the pattern rather than guess at it.

7. Where can I follow what the Federal Reserve is actually saying?

The Fed publishes everything itself, so you would think that settles the question. It does not. The volume is the problem. Eight meetings a year, each with a statement and a press conference. Four sets of economic projections including the dot plot. Minutes three weeks after each meeting. And dozens of speeches a month from nineteen participants, of whom only twelve vote in any given year.

Reading all of it is not a strategy. Start with who is speaking and whether they vote this year, because a non-voting regional president making news is often just making news. Then place them on the hawk to dove range against their own past comments rather than against the committee. A dove sounding slightly less dovish is a bigger signal than a hawk being hawkish again.

Read the statement as a diff against the previous one and look only for what changed. A single word moving is often the whole story. And when the dot plot lands, everyone quotes the median while the dispersion carries more information. A tight cluster and a wide scatter around the same median mean very different things for how much the next print can move rates.

The Fed hub tracks the roster with each speaker's lean and voting status, the FOMC hub carries the meeting itself, and there is a full guide to reading the statement and dot plot alongside a page on how the FED SPEAK tab scores tone. Two definitions are worth having in front of you before any of it: the fed funds rate and the dot plot itself. If you want to know where overnight money is genuinely clearing rather than where the target says it should, SOFR is the number to watch.

8. How do I check the market's reaction instead of trusting the headline?

Let the price tell you what the number meant. It is the most useful habit in this post and it costs nothing.

The Treasury market is the scoreboard for economic news. It has no editor and no narrative. Millions of dollars vote on every print within a second of it landing, and the result is public. When a number crosses, look at the tape before you read a word of commentary.

The 2-year yield is the Fed path, distilled. If a headline calls inflation hot and the 2-year has not moved, either the market disagrees with the headline or it had already priced it, and both mean the same thing for you: there is no trade there. The 10-year yield carries growth and inflation over the longer horizon, plus term premium, the extra yield investors want for holding duration, which moves for reasons that have nothing to do with today's data. Then the curve, usually 2s10s, where direction matters far more than level. Bear flattening after a hot print says the market repriced the Fed. Bear steepening says it repriced inflation or supply. Those are two entirely different trades wearing the same headline.

Read all of it in basis points and judge the size against the day's normal range, because 4 basis points on a quiet Tuesday is a far bigger event than 4 basis points on FOMC day. Basis points, the 2s10s spread and duration are the words to learn first, and yield curve inversion is the next layer down.

The live curve sits on the rates page, and our guide on how to read the yield curve walks a real move from print to repricing. If you would rather have one number telling you whether the tape is confirming a story or fighting it, that is what the momentum score exists for.

9. Do Treasury auctions count as economic news?

Yes, and most investors ignore them, which is part of why they are worth watching.

An auction is not a survey, a forecast or an opinion. It is a live vote with real money on whether the world will fund the United States at that yield, on that day. Auctions run several times a month across bills, notes and bonds, with results around 1:00pm ET. Three numbers carry almost the whole story.

The tail is the auction's high yield against the when-issued yield trading in the seconds before the deadline. A positive tail means the auction had to concede yield to clear, which is weak. A negative one is a stop-through, which is strong. Bid to cover is judged against its own trailing average for that tenor, never against a rule of thumb someone repeated on television. And the bidder split tells you who turned up. Indirect bidders are the proxy for foreign and real-money demand, direct bidders are domestic non-dealers, and primary dealers take whatever is left over. A heavy dealer take is a soft auction whatever the cover ratio says.

A tailing 10-year note or 30-year bond can move the long end more than a tier two data print, and a run of soft auctions is a slow-burning term premium story that never reaches a front page. What a result tells you depends on which part of the market is being funded.

The front and belly, meaning the 2-year, 3-year, 5-year and 7-year notes, is where the Fed path gets funded. A soft result there says the market is not yet comfortable with the cuts it has already priced. The long end is the 10-year note, 20-year bond and 30-year bond, where term premium shows up first and a badly received long auction drags equities with it more often than people expect.

The bills are the 4-week, 8-week, 13-week, 17-week, 26-week and 52-week, along with the 2-year floating rate note that reprices off them. Read those as a funding and money-market signal rather than a duration one, and watch them closely around quarter end and any debt-ceiling deadline. That leaves the inflation-linked corner, the 5-year, 10-year and 30-year TIPS, where demand is a cleaner read on what real money expects from inflation than any sentiment survey you will see quoted.

Start at the auctions hub for results as they land, or the Treasury auction tails dataset if you want the history in one place. The step-by-step version is in how to read a Treasury auction, with when-issued trading and the quarterly refunding announcement for the layers underneath, plus a reference page on how each field is scored.

10. How do I build a news routine without drowning in noise?

Not by reading more. By deciding in advance what you will read, and letting everything else go past.

Fifteen minutes a day

The night before, check tomorrow on the calendar and mark anything tier one. If there is nothing, you have permission to ignore the news entirely tomorrow, and most days there is nothing. Fifteen minutes before a print, know the consensus, and know what number would change your mind. When it lands, read the surprise rather than the headline, and check the revision line before you form a view. Five minutes later, look at the 2-year and the curve. Did the market agree with the story you just read? If not, trust the market. At the end of the day, one read of what actually drove the tape, not twelve versions of it.

What to cut

Cut anything with no number in it, and any source that cannot tell you the consensus forecast. Price commentary written after a move to explain the move goes too, because it is always available and never predictive. So do alerts on instruments you would never trade. Arm the handful you would actually act on and let the rest stay silent.

That shape is roughly how the terminal is built, because it is how a desk works: a live feed bucketed by what each item actually is rather than who published it, alerts you arm yourself on thresholds you choose, and a written read of the session at the end. You can also rearrange the panels so the three things you check are the three things you see. The help centre covers the mechanics and the FAQ covers most of the rest.

The stack we would build

Take the primary agencies as your truth. BLS, BEA, Census, the Department of Labor, the Federal Reserve and the Treasury are free, public, and upstream of every headline you will ever read. Put a calendar over the top of them so that nothing ever surprises you simply by existing.

Then add one scored feed that timestamps the capture and attaches the forecast, the revision and the size of the surprise, because that is the part that is genuinely hard to assemble yourself. Use the Treasury market as your check on every headline you read. And keep a glossary and a set of guides to hand, so the vocabulary stops being the thing standing between you and the read.

Where to go next

To see it working, open the news feed, the data hub, the auctions hub and the rates page. The vocabulary lives in the glossary and the trading guides, and the committee lives on the Fed hub and the FOMC hub.

Setting it up means subscribing and turning on squawk audio, with pages for what to do if alerts are not firing or a number looks stale. Changing plan is on the cancel page, and yes, cancelling really is two clicks.

If you would rather talk to people than read, there is the community board and the public roadmap, where you can suggest and vote on what gets built next. If you want to work with us, there is the affiliate programme and how it pays. Anything else, get in touch.