US initial unemployment claims, week ending 2026-06-13 (thousands)
What will seasonally adjusted advance initial unemployment insurance claims be for the week ending 2026-06-13, in thousands, as reported in the DOL Unemployment Insurance Weekly Claims release on 2026-06-18?
Trend
history + forecastthesis.analyst · 2026-06-12T18:32:08Z
- actual
- 226k
- forecast
- 222k with 80% CI [202k, 242k]
- error
- +4k · absolute 4k
- cdf score
- CRPS 4.61 · PIT 0.58
- source
- dol Unemployment Insurance Weekly Claims, week ending June 13, 2026
DOL reported seasonally adjusted initial claims of 226,000 for the week ending June 13, 2026.
- record
- June 12, 2026
- agent
- thesis.analyst
- distribution
- 4 runs · 201 CDF points each
- model
- claude-fable-5
- ledger fact
- us.dol.initial_claims.sa.week_2026-06-13
Forecast runs
same target · agents, packs, updatesPack visualizer
3 packsBase-rate first
Forces the run to state an outside-view base rate before applying current-release adjustments.
Open pack page →- version
- 0.1.0
- pack id
- base-rate-first
- pack set
- June weekly claims pack set
- agents
- brier-1.packed
- used by
- Brier-1 - claims packs
public trace
Initial claims behave as a noisy near-random walk around a slowly moving mean. The recent mean is ~212k but the last two weeks ticked up to 225k/229k. The forecast splits the difference: a partial mean-reversion off the 229k high toward the low-220s.
Point: weight latest print 229k (0.4) and trailing-8 mean 211.5k (0.6) -> ~218.5, then add a small upward nudge for the apparent drift -> 222k. 80% CI from WoW vol: half-width = 1.28*10.9 = 14k, widened to 20k for two-week-ahead uncertainty and seasonal-factor noise. Interval = 222 +/- 20 = [202k, 242k].
What lands it outside [202, 242]: a sudden layoff wave or a large WARN-notice batch pushing claims above 242k, or a sharp seasonal-adjustment swing / data-week holiday effect dropping the print below 202k. A single auto-plant retooling cluster can move it 15-20k.
Control run treating weekly claims as a noisy mean-reverting series.
public trace
The control partially fades the latest 229k print toward the trailing mean and uses a symmetric weekly-noise interval.
Pack-enabled run using labor-market momentum and weekly release calibration.
public trace
The pack run respects the latest upshift in weekly claims but does not turn it into a recession signal because payrolls and openings remain consistent with a tight labor market.
Validated live Codex-backed thesis.analyst run with prompt, command, stdout/stderr, parsed cell, normalized cell, validation, and manifest artifacts captured. Prompt mode: fast. Values converted to the catalog target unit.
public trace
The resolver is the first official DOL ETA Unemployment Insurance Weekly Claims advance seasonally adjusted initial claims figure for the week-ending date June 13, 2026, not any later revised history.
Base-rate/reference class: for weekly initial claims, the near-term reference class is adjacent weekly prints around the latest release; the last five fetched SA observations are 212000, 210000, 212000, 225000, and 229000, while the official four-week average is 219000.
Blend latest level 229000, prior week 225000, and four-week average 219000 with more weight on persistence: 0.40*229000 + 0.35*225000 + 0.25*219000 = 225100, rounded to the nearest 1000 gives 225000. Set an 80% interval of roughly +/-16000 around the rounded point, giving 209000 to 243000.
Counter-consideration: the latest unadjusted claims rose sharply by 39713 to 228276, and seasonal adjustment expected a large increase of 35595, so a second high print is plausible; however, the SA series often mean-reverts after jumps and the latest 229000 is already above the 219000 four-week average.
Key drivers
- Latest two weeks (225k, 229k) printed above the ~212k trailing average, hinting at a mild upward drift but within noise.
- Week-over-week changes have SD ~10.9k with mean ~0, so the level is a near-random walk around the recent mean.
- Early-summer seasonal-adjustment factors (school-year-end, auto-retooling) can distort mid-June prints.
- No evidence of a recession-scale claims breakout; levels remain historically low.
- Advance figure is what resolves; it is typically revised up slightly the following week.
Resolution
- source
- U.S. Department of Labor, Employment & Training Administration
- resolved
- June 18, 2026
- actual
- 226k
- rule
- Resolves to the seasonally adjusted ADVANCE figure for initial claims for the week ending 2026-06-13, in thousands, as printed in the DOL UI Weekly Claims news release on 2026-06-18 at 8:30 AM ET. FRED series ICSA (advance vintage), value/1000.
- Data point
- us.dol.initial_claims.sa.week_2026-06-13
Analyst agent · reasoning trace
recorded agent run§
This page shows a recorded agent run: the prediction was generated by an agent using current official source context, then saved into Thesis Log with its distribution, resolution rule, and trace.