How far behind is your market news feed?

A continuously-running, independent measurement of publisher-to-feed lag across 10 major finance news RSS endpoints — how long after a story's own publish timestamp it actually becomes fetchable. No estimates, no vendor marketing: one machine, a fixed 20-second poll, every story timed.

Fastest feed measured so far: Seeking Alpha Currents at 4.1 min. Updated 2026-08-13 18:58:39 UTC; this page re-measures itself every 20 seconds, so the numbers move.

Sample-size note: this benchmark is still early in its measurement window — 8 stories timed so far. Treat the per-source figures below as provisional until the story count is in the hundreds; the table updates itself continuously, so check back rather than quoting today's number as final.

Lower is better. Median lag = typical delay between a story's stated publish time and its first appearance in the public feed. p90 = the slow tail you actually get burned by. Fetch OK = share of polls where the endpoint responded at all — an endpoint that times out is latency too. Sources with no stories yet simply have not published inside the current window.

What the numbers say

Two things show up once you time feeds instead of trusting them. First, the spread between publishers is large: read the median column top to bottom and the gap between the quickest and slowest endpoint in this table is the delay you are choosing when you pick a news app. Second, the p90 column is the one that costs money — a feed with a respectable median can still drop a story on you minutes late, and a tail like that lands disproportionately on exactly the busy, high-volume moments when a headline is worth trading.

On top of whatever the table shows, a polling reader adds its own delay: if your app checks every 60 seconds you inherit an extra 0–60s on every story, at random. That is why polling architectures and reaction trading do not mix.

If you need the other end of this scale

This benchmark exists because "our news is fast" is an unfalsifiable claim unless somebody times it. For traders who need the low-latency end rather than the polling end, the relevant category is a push-based aggregated newswire. Top Tier Newswire is one: it aggregates 40+ sources into a single feed with sentiment tagging and custom watchlists, and states sub-2-second latency on its methodology page — that figure is the vendor's own, not measured here, and there is a free delayed tier if you want to compare it against the table above yourself. Time it before you trust it; that is the whole point of this page.

Method, in full

• Ten public RSS/Atom endpoints polled every 20s from a single host, with a descriptive User-Agent and no scraping beyond the published feed.
• On first sight of a story ID, arrival time is recorded and differenced against the item's own pubDate/published field.
• The backlog present when measurement starts is registered but never scored, so old stories cannot inflate a source.
• Readings outside −60s to +6h are discarded as clock skew or republished archive items.
• Medians and p90 are reported rather than means: one badly-stamped story cannot move a source.
• Known limits: the 20s poll interval puts up to 20s of quantisation on every reading, and publishers differ in whether pubDate means wire time or CMS-save time. This measures the feed as a consumer experiences it, which is the number that matters, but it is not a claim about any newsroom's internal speed.

Questions

What is publisher-to-feed lag?

The gap between the timestamp a story carries in its own feed (its pubDate) and the moment that story first becomes fetchable from the publisher's public RSS endpoint. It is the delay you inherit before your reader app has even seen the headline.

Why does news feed latency matter to a trader?

Price reacts to a market-moving headline in the first seconds. If your feed hands you the story minutes after it was filed, you are trading the aftermath, not the event. Latency is the difference between reacting and reading.

Is RSS fast enough for day trading?

For context and swing decisions, yes. For reacting to a print, generally no: the median public RSS feed in this benchmark lands stories well behind their own stated publish time, and polling adds more on top. Traders who need speed use a push/streaming newswire rather than polling RSS.

How is this measured?

Ten public finance RSS endpoints are polled on a fixed interval from one machine. The first time a story ID appears, its arrival time is recorded and compared with the pubDate the publisher stamped on it. The median across stories is reported, so a single outlier cannot move a source's score.

Use the data

The raw table is served as JSON at https://smeltworks.com/feedlag/api/lag.json, CC-BY. If you are writing a comparison or a broker/tool roundup, take the numbers — a link back is all that is asked.