thesisPrototype

Government data

Initial claims, week ending Sep. 12, 2026

What will the U.S. Department of Labor first report as the advance seasonally adjusted initial unemployment insurance claims count, series ICSA, for the week ending 2026-09-12?

Forecast

207k
191k–223k
80% prediction interval
gpt-5.5Run record ↗

Cumulative probability

Chance that the outcome is at or below each value.

0%25%50%75%100%167k207k247kForecast value

Shaded band: 80% interval (191k–223k). Dashed line: point estimate (207k).

Derived from the point estimate and 80% interval; the agent did not report a full distribution.

Observed outcomeinside 80% interval
actual
198k
forecast
207k with 80% interval [191k, 223k]
error
-9k · absolute 9k
cdf score
CRPS 5.49 · PIT 0.28
source
dol_eta Unemployment Insurance Weekly Claims (advance)

DOL ETA UI Weekly Claims news release, advance seasonally adjusted figure for the week ending 2026-09-12, read from FRED ICSA (advance vintage) as the cell's resolver names.

Analysis

§US initial claims first-print forecast for week ending 2026-09-12

Framing and exact resolver: this forecasts the U.S. Department of Labor Employment and Training Administration advance seasonally adjusted initial claims series, ICSA, for week ending 2026-09-12. All anchors use the same seasonally adjusted initial-claims variant, not not-seasonally-adjusted claims, continued claims, or all-program weeks claimed. DOL is the substantive agency source; ALFRED is the registered mechanical advance-vintage mirror.

▸ Reported tool use: official.lookupmodel report
Inspected DOL ETA news release listings for recent Unemployment Insurance Weekly Claims Reports.
↳ Fetched DOL first-print seasonally adjusted initial claims in thousands: 2026-08-29 = 206, 2026-08-22 = 203, 2026-08-15 = 206, 2026-08-08 = 209, 2026-08-01 = 199, 2026-07-25 = 197, 2026-07-18 = 187, 2026-07-11 = 208, 2026-07-04 = 215, 2026-06-27 = 215, 2026-06-20 = 215, 2026-06-13 = 226.
▸ Reported tool use: official.lookupmodel report
Checked the DOL Office of Unemployment Insurance archive publication schedule and the FRED release calendar for the weekly claims release timing.
↳ Fetched schedule numbers: DOL says the UI Weekly Claims News Release is published Thursday at 8:30 AM EST and lists Wednesday 2026-11-25 as the only 2026 non-Thursday exception; FRED's release calendar lists Unemployment Insurance Weekly Claims Report on Thursday 2026-09-17 at 7:30 AM Central and Thursday 2026-09-24 at 7:30 AM Central.
▸ Reported tool use: alfred.lookupmodel report
Checked the ALFRED/FRED ICSA series page for the registered mirror identity and latest revised observation context.
↳ Fetched ALFRED/FRED ICSA mirror values in persons: 2026-08-29 = 206,000, 2026-08-22 = 204,000, 2026-08-15 = 207,000, 2026-08-08 = 212,000, 2026-08-01 = 200,000; series updated 2026-09-03 at 7:34 AM CDT.
▸ Reported tool use: official.lookupmodel report
Checked BLS Employment Situation for August 2026 as contemporaneous labor-market context.
↳ Fetched BLS labor context: August 2026 nonfarm payroll employment increased by 162,000, unemployment rate was unchanged at 4.1 percent, unemployed persons were 7.0 million, and June plus July payroll revisions were +55,000.

Base rate/reference class: the recent official-source reference class is twelve DOL first-print seasonally adjusted ICSA weekly values from 2026-06-13 through 2026-08-29: 226, 215, 215, 215, 208, 187, 197, 199, 209, 206, 203, and 206 thousand. The base rate is tightly centered near 207 thousand; the full-sample mean is 207.17, the median is 207, and the latest four first prints average 206.

Level, momentum, one-off, and policy-mechanism effects: the level starts from 206 thousand for week ending 2026-08-29. Momentum is flat after 209, 206, 203, and 206, while the July 18 trough at 187 looks like a one-off low rather than the new center. The August payroll gain of 162,000 and unchanged 4.1 percent unemployment argue against an imminent layoff breakout, but claims often stay low in a slow-hiring environment. I found no policy mechanism in the checked public releases that should mechanically shift regular initial claims for the Sep. 12 week.

Prior/update/interval: prior model is two-week persistence around the latest DOL first print and recent reference-class center; historical sample is the twelve fetched DOL first prints from 2026-06-13 through 2026-08-29. Adjustment components are +1 thousand toward the 207 thousand median, 0 thousand from flat recent momentum, 0 thousand from the strong August jobs report because it mainly confirms low layoffs rather than reducing claims below the current range, and 0 thousand for policy effects; point = 206 + 1 = 207. Interval method uses realized dispersion of successive weekly changes in the fetched first-print history: -11, 0, 0, -7, -21, +10, +2, +10, -3, -3, +3 thousand; sample sigma = 8.98 thousand. One-week 80% half-width = 1.28*sigma = 11.49 thousand; with two unreleased weeks from the latest 2026-08-29 print to the target 2026-09-12, scale by sqrt(2): 11.49*1.414 = 16.25 thousand. Final implied bounds = 207 +/- 16.25 = [190.75, 223.25], rounded to [191, 223].

Counter-considerations: upside risk is a real post-Labor-Day layoff pickup, a seasonal-adjustment miss, or a claims catch-up after the low July/August prints, which would land above the interval if the first print exceeds 223 thousand. Downside risk is another unusually quiet filing week like 2026-07-18, which would land below the interval if the first print is under 191 thousand. An outside the interval outcome would most likely require either a renewed sub-190 low or a return to the June high-220s.

Review disposition: accepted the reviewer finding that no required fixes were needed. Kept the registered ALFRED advance-vintage resolver and ledger resolutionDate distinction, and retained the explicit 12-week first-print volatility basis because it satisfies the required source grounding and interval arithmetic.

Key drivers

  • Latest DOL first print was 206 thousand for week ending 2026-08-29
  • Recent first-print four-week average is 206 thousand
  • August employment report was stronger than expected with unemployment unchanged at 4.1 percent
  • Claims remain in a low-layoff low-200-thousand regime
  • Two unreleased weeks create moderate horizon and Labor Day seasonal-adjustment risk

Tool evidence

This run has no captured tool responses. “Reported tool use” in the analysis is the model’s account, not an archived tool response.

Sources and resolution

Official source
U.S. Department of Labor ETA UI Weekly Claims News Release, preserved through ALFRED ICSA advance vintage
Resolved
September 29, 2026
Resolution rule
Resolve through the registered ALFRED graph CSV source binding for sourceSeriesId ICSA, field ICSA, releasePolicy advance_vintage, selecting the first advance vintage corresponding to the first official U.S. Department of Labor ETA seasonally adjusted initial claims print for the week ending 2026-09-12. Multiply persons by 0.001 to express thousands and round to the nearest 1 thousand. Use the first print only; ignore later DOL revisions, FRED/ALFRED vintage changes after the first available advance vintage, and benchmark updates. The official DOL weekly release schedule points to Thursday 2026-09-17 for the agency first print, while the registered ledger target preserves resolutionDate 2026-09-29 and expected ALFRED window 2026-09-25 through 2026-09-29.

Forecast history

Select a version to read its estimate and analysis.

Forecast versions, estimates, and intervals
VersionDateEstimate80% intervalCRPS
gpt-5.5 · selectedSep 7, 2026207k191k–223k5.49
persistence.last_printSep 7, 2026203k193k–213k3.21
Run details

The analysis is the model’s written report. Tool-use descriptions in that report are model claims; the activity artifacts contain the execution record.

thesis.analyst · gpt-5.5 · fast · v2.5.11

pre-submit review · completed

Draft is publication-ready against the rubric, with resolver, prior, update, interval, tails, and JSON fields coherent with the registered target.

  • info optional_suggestion: Consider shortening the resolver note so the agency first-print date and ledger resolutionDate distinction is easier to scan.
  • info optional_suggestion: The interval method is acceptable; a larger historical sample could be mentioned if available, but the current explicit 12-week volatility basis is sufficient.

disposition not applicable: Review disposition: accepted the reviewer finding that no required fixes were needed. Kept the registered ALFRED advance-vintage resolver and ledger resolutionDate distinction, and retained the explicit 12-week first-print volatility basis because it satisfies the required source grounding and interval arithmetic.

disposition not applicable: Review disposition: accepted the reviewer finding that no required fixes were needed. Kept the registered ALFRED advance-vintage resolver and ledger resolutionDate distinction, and retained the explicit 12-week first-print volatility basis because it satisfies the required source grounding and interval arithmetic.

Activity artifacts

Complete original trace

§US initial claims first-print forecast for week ending 2026-09-12

Framing and exact resolver: this forecasts the U.S. Department of Labor Employment and Training Administration advance seasonally adjusted initial claims series, ICSA, for week ending 2026-09-12. All anchors use the same seasonally adjusted initial-claims variant, not not-seasonally-adjusted claims, continued claims, or all-program weeks claimed. DOL is the substantive agency source; ALFRED is the registered mechanical advance-vintage mirror.

▸ Reported tool use: official.lookupmodel report
Inspected DOL ETA news release listings for recent Unemployment Insurance Weekly Claims Reports.
↳ Fetched DOL first-print seasonally adjusted initial claims in thousands: 2026-08-29 = 206, 2026-08-22 = 203, 2026-08-15 = 206, 2026-08-08 = 209, 2026-08-01 = 199, 2026-07-25 = 197, 2026-07-18 = 187, 2026-07-11 = 208, 2026-07-04 = 215, 2026-06-27 = 215, 2026-06-20 = 215, 2026-06-13 = 226.
▸ Reported tool use: official.lookupmodel report
Checked the DOL Office of Unemployment Insurance archive publication schedule and the FRED release calendar for the weekly claims release timing.
↳ Fetched schedule numbers: DOL says the UI Weekly Claims News Release is published Thursday at 8:30 AM EST and lists Wednesday 2026-11-25 as the only 2026 non-Thursday exception; FRED's release calendar lists Unemployment Insurance Weekly Claims Report on Thursday 2026-09-17 at 7:30 AM Central and Thursday 2026-09-24 at 7:30 AM Central.
▸ Reported tool use: alfred.lookupmodel report
Checked the ALFRED/FRED ICSA series page for the registered mirror identity and latest revised observation context.
↳ Fetched ALFRED/FRED ICSA mirror values in persons: 2026-08-29 = 206,000, 2026-08-22 = 204,000, 2026-08-15 = 207,000, 2026-08-08 = 212,000, 2026-08-01 = 200,000; series updated 2026-09-03 at 7:34 AM CDT.
▸ Reported tool use: official.lookupmodel report
Checked BLS Employment Situation for August 2026 as contemporaneous labor-market context.
↳ Fetched BLS labor context: August 2026 nonfarm payroll employment increased by 162,000, unemployment rate was unchanged at 4.1 percent, unemployed persons were 7.0 million, and June plus July payroll revisions were +55,000.

Base rate/reference class: the recent official-source reference class is twelve DOL first-print seasonally adjusted ICSA weekly values from 2026-06-13 through 2026-08-29: 226, 215, 215, 215, 208, 187, 197, 199, 209, 206, 203, and 206 thousand. The base rate is tightly centered near 207 thousand; the full-sample mean is 207.17, the median is 207, and the latest four first prints average 206.

Level, momentum, one-off, and policy-mechanism effects: the level starts from 206 thousand for week ending 2026-08-29. Momentum is flat after 209, 206, 203, and 206, while the July 18 trough at 187 looks like a one-off low rather than the new center. The August payroll gain of 162,000 and unchanged 4.1 percent unemployment argue against an imminent layoff breakout, but claims often stay low in a slow-hiring environment. I found no policy mechanism in the checked public releases that should mechanically shift regular initial claims for the Sep. 12 week.

Prior/update/interval: prior model is two-week persistence around the latest DOL first print and recent reference-class center; historical sample is the twelve fetched DOL first prints from 2026-06-13 through 2026-08-29. Adjustment components are +1 thousand toward the 207 thousand median, 0 thousand from flat recent momentum, 0 thousand from the strong August jobs report because it mainly confirms low layoffs rather than reducing claims below the current range, and 0 thousand for policy effects; point = 206 + 1 = 207. Interval method uses realized dispersion of successive weekly changes in the fetched first-print history: -11, 0, 0, -7, -21, +10, +2, +10, -3, -3, +3 thousand; sample sigma = 8.98 thousand. One-week 80% half-width = 1.28*sigma = 11.49 thousand; with two unreleased weeks from the latest 2026-08-29 print to the target 2026-09-12, scale by sqrt(2): 11.49*1.414 = 16.25 thousand. Final implied bounds = 207 +/- 16.25 = [190.75, 223.25], rounded to [191, 223].

Counter-considerations: upside risk is a real post-Labor-Day layoff pickup, a seasonal-adjustment miss, or a claims catch-up after the low July/August prints, which would land above the interval if the first print exceeds 223 thousand. Downside risk is another unusually quiet filing week like 2026-07-18, which would land below the interval if the first print is under 191 thousand. An outside the interval outcome would most likely require either a renewed sub-190 low or a return to the June high-220s.

Review disposition: accepted the reviewer finding that no required fixes were needed. Kept the registered ALFRED advance-vintage resolver and ledger resolutionDate distinction, and retained the explicit 12-week first-print volatility basis because it satisfies the required source grounding and interval arithmetic.

calibrated forecast · 80% CI
207k[191k · 223k]
Target metadata

Data point: us.dol.initial_claims.sa.week_2026-09-12

More government data forecasts