Start with the complete inflation measure
Our example is the Polymarket question: Will Core CPI MoM be 0.3% in September? Its retained description specifies September 2026, CPI-U excluding food and energy, the seasonally adjusted change from the preceding month, and the official BLS news release at one-decimal precision. The short title does not contain all those requirements.
Headline CPI covers all items. Core CPI excludes food and energy. MoM compares with the preceding month; YoY compares with the same month a year earlier. Seasonal adjustment is another independent choice. Replacing a seasonally adjusted monthly figure with an unadjusted one changes the measurement even if both numbers happen to coincide.
Write the required measure down before opening a news story: core CPI-U, seasonally adjusted, one-month percent change, September 2026, exactly 0.3%. This prevents a headline or search snippet from choosing the answer for you. CPI-W, chained CPI and PCE are also different series.
Use Table A's row and column together
The fixed HowSure Check preserves a BLS summary read on October 14, 2026, at 22:12 UTC. Its September figures illustrate four different answers in one release:
- Core monthly CPI: 0.3%, seasonally adjusted.
- Headline monthly CPI: 0.6%, seasonally adjusted.
- Core 12-month CPI: 2.6%, unadjusted.
- Headline 12-month CPI: 3.7%, unadjusted.
Check the number against the narrative
For this contract, find the core row, labeled All items less food and energy, and the September column under the seasonally adjusted monthly heading. The annual column sits beside the monthly columns; reading across without checking the header can turn 0.3% into 2.6%. The headline row would instead give 0.6% for the month.
The retained opening narrative and Table A both report core monthly inflation of 0.3%. That matches the selected 0.3% outcome numerically. Keep the row, column headers and narrative with the result. If they disagree, stop the comparison and reconcile the source rather than selecting whichever number fits the market title.
September data can be published in October
September is the reference month in this example. The contract names October 14, 2026, at 8:30 AM ET as its scheduled release, and the retained BLS report prints that release label. The October publication does not turn the observation into October CPI.
Keep three dates distinct: the period measured, the report's printed publication label and when your system actually read it. Here the saved read occurred hours after the printed time. Neither a catalogue closing timestamp nor a successful later download proves the exact first-publication contents.
Exactly 0.3% and 0.6% or more use different comparisons
The September question asks for equality: the qualifying one-decimal figure must be 0.3%. It does not mean at least 0.3%. By contrast, a separate July 2026 contract asks whether Core CPI MoM is 0.6% or more. Its boundary includes exactly 0.6%.
For a hypothetical qualifying release, 0.6% satisfies that inclusive threshold, 0.7% also satisfies it, and 0.5% does not. These are illustrations of the operator, not observations about July. The July terms were retained in HowSure's October 15 production reading.
Both examples specify the reported one-decimal figure. Do not reconstruct extra precision from today's index database or invent a range around an equality outcome. HowSure recognizes bounded core monthly CPI terms, including this 0.6%-or-more form; that does not establish support for every inflation contract or threshold.
A current summary and a historical release answer different questions
The BLS archive index points to a current-summary URL that is reused for later releases. Check its period on every read. For an older month, follow the matching archive entry; a valid publisher URL showing another month is still the wrong evidence.
BLS warns that archived figures can differ from later revised data. Its technical notes also describe revisions to seasonally adjusted data for up to five years. Preserve the version you used instead of replacing it with a newer database value. The example's terms do not explicitly settle how later corrections or revisions should be treated. A retained archive or same-day download alone does not prove unchanged original publication bytes.
Unavailable evidence is not a negative outcome
These terms contain a delayed-release fallback: wait until the scheduled next CPI report, then use the latest previous month with data if the required figures remain absent. That requires evidence about the deadline and actual nonpublication. An API timeout or missing archive link cannot trigger it.
Keep the reported value, numerical comparison and final venue resolution separate. The fixed September Check establishes a retained 0.3% reading; it leaves original-release fidelity, later-correction treatment and settlement unverified. Its source review was prepared with AI assistance and is not independent editorial certification.
Make the next read reproducible
This public request selects the same exact contract and explicitly requests CPI release evidence. It needs no key or paid model. A later response can differ or be unavailable; save its identity, source times, version, gaps and limitations. It does not overwrite the fixed Check or establish settlement.
curl --fail-with-body --show-error --proto '=https' --max-time 30 --get \
'https://www.howsure.com/api/free/v1/research/evidence' \
--data-urlencode 'market=will-core-cpi-mom-be-0pt3-in-september-2026' \
--data-urlencode 'include_cpi_release=true' \
-o howsure-september-cpi.jsonSources & verification
- September Core CPI Check — fixed October 14, 2026 source review
- Exact September 0.3% Polymarket contract
- BLS CPI current summary, Table A and technical notes — checked October 16, 2026
- BLS CPI archived releases and revision caution — checked October 16, 2026
- Separate July 0.6%-or-more Polymarket contract
Sources and instructions checked 16 October 2026. Live responses can change. Editorial method and limitations.