How we measure the delay between a headline and the market's reaction

Definitions

A full worked example

On Mar 23, 2026, the trump/iran de-escalation post crossed publicly on Truth Social. Here is the complete, exact timeline MoveSurge captured:

The running record

Over the last 30 scheduled releases we measured, the median gap between the official release second and our published headline was 1.79 seconds. The fastest was 0.3s and the slowest 18.26s. Every observation in that sample is in the table below, slow ones included. Sample runs 2026-10-13 to 2026-10-27.

Added automatically, newest first. The official column is the release time the economic calendar carried before the print; the published column is the timestamp stored on our own event, to the millisecond. Releases without a fixed public second — speeches, tentative dates, anything released into a window — are not measured, and a row stamped exactly on the scheduled second is discarded rather than counted as a zero-second win.

SEC filings: from EDGAR to our filings page

Our filings lane reads EDGAR's list of new filings every second. It stores three times for each filing. Accepted is the second EDGAR accepted it, printed on the filing's EDGAR index page. Seen is the first time our lane found it on EDGAR's public list. Stored is when it was ready on our filings page and in our API.

Over the last 5583 filings the lane caught live since Monday, October 26, 2026, the median time from Seen to Stored was 79 ms. Nine in ten took 718 ms or less. The median gap from EDGAR acceptance to our first detection was 31 s. That includes EDGAR publication time and our detection delay. The lane targets one poll per second; network delays and retries can make detection take longer. Seen-to-Stored measures our processing only, not source-to-browser delivery or speed against another service.

Newest first, added automatically. To check a row, open the filing on EDGAR and compare its Accepted time. Filings the lane found only later are left out. That happens after a restart or in the nightly check against EDGAR's daily index, and their Seen time says nothing about speed.

What this one case does and doesn't prove

It proves MoveSurge's pipeline is capable of capturing a public post and a named instrument's reaction to sub-second and low-single-second precision, with exact, independently verifiable timestamps, on at least this one occasion. It does not, on its own, prove a general win rate against any specific competitor across all breaking news -- that would require a standing, published, repeatable benchmark comparing simultaneous timestamped captures across multiple providers on the same events, over an extended period.

What a fuller benchmark would need

• An explicit comparison period and a defined, non-cherry-picked set of qualifying events.
• A clear, published definition of "market-moving" applied consistently across every event in the sample.
• A named timestamp source for each competitor being compared, with a stated position on whether that timestamp reflects publication, API delivery, browser rendering, or end-user receipt -- these are meaningfully different things.
• A documented treatment for ties, edited headlines, and duplicate or re-published stories.
• Downloadable sample data so the comparison can be independently checked, not just quoted.
• An explicit statement of limitations and any conflicts of interest in how the comparison was run.