Event Explorer

Pick a U.S. economic release and a market, and see how much it actually moved on the day it printed — measured against a normal day, with a significance test attached. Direction is shown honestly: it is almost always unpredictable.

How to use it

• Period — all years back to 2000, or one decade at a time. This is often where the interesting part is: CPI barely moved the 10-year in the 2000s (×1.09) or the 2010s (×1.05), but does in the 2020s (×1.26). The same release can matter in one regime and not another.
• Economic release — the eight scheduled U.S. releases we track, from CPI to the FOMC decision. Seven of them go back to 2000. The FOMC is the exception: the Federal Reserve does not publish its decision dates through the same feed as the statistical agencies, so ours are entered by hand from the Fed's calendar and currently start in 2023. Pick an earlier era with FOMC selected and the tool will say so rather than show you a number built on nothing.
• Market — Treasury yields across the curve (3-month to 30-year), the Federal Reserve's broad dollar index, and WTI or Brent crude. These are all U.S. government public-domain series; we deliberately do not include copyrighted indices such as the S&P 500 or proprietary gauges like the VIX.
• Holding period — release day, three trading days, or five. Watching a robust release-day effect fade at three and five days is the clearest way to see that these reactions are same-day events.

What the readings mean

• Size of move — the average absolute move around that release divided by the average on an ordinary window. "×2.0" means twice a normal day. The badge tells you whether it is Robust (survives multiple-testing correction), Nominal only (looks significant alone but fails once you account for testing many combinations), or No effect.
• Direction — whether the signed move is predictable. Nothing in the grid survives multiple-testing correction on direction. A handful of combinations do clear p<0.05 on their own — CPI against the dollar, the FOMC against the 5-, 10- and 30-year — but once you account for testing every release against every market, none of them holds. That is not a bug in the tool; it is the finding.
• Sample — how many releases the estimate is built on. Combinations flagged Low sample should be treated as indicative only.

Things worth trying

• Jobs report × 2Y, release day — the largest robust reaction anywhere in this tool: ×1.62 across all years, and ×1.67 if you narrow the period to the 2000s. Nothing else in the grid gets above ×1.7.
• CPI × 10Y, then switch from release day to 3 days to 5 days — ×1.12 becomes ×0.99 and then ×0.97, and the Robust badge disappears. The reaction is a same-day event.
• PCE × USD — ×0.95. The Fed's preferred inflation gauge moves the dollar slightly less than a normal day, because CPI has already told the story.
• Any release × 3M — the front of the curve is calm on every release day (×0.81 to ×1.16, none of it robust), because it is anchored to the current policy rate rather than to data.

The numbers here run lower than the ones in the individual guides — ×1.62 for jobs × 2Y instead of the guide's ×2.35. Both are correct; they measure different things, and the reason is in Method and sources below.