US continued claims, week ending Jun. 27 2026

The resolver is the national seasonally adjusted insured unemployment count, also called continued claims, for the week ending June 27, 2026. All anchors use the same DOL/ETA seasonally adjusted insured-unemployment variant, not initial claims, unadjusted claims, or FRED as the resolution source. The point and interval are derived independently from the cited official DOL and BLS evidence, not from any catalog forecast, prior point estimate, prior interval, or prior Thesis run.

Base rate/reference class: the tightest reference class is recent DOL first-release-adjacent seasonally adjusted insured-unemployment levels. The latest five same-variant values rose from 1.771 million to 1.814 million, with adjacent changes of +0.015, +0.014, +0.012, and +0.002 million. A persistence prior therefore starts near 1.814 million but allows a small continued increase.

Level, momentum, one-off, and policy split: the level is not recessionary but is drifting higher. Momentum from May 23 through June 13 was clearly upward, while the latest weekly increase slowed to 0.002 million. Low initial claims reduce upside inflow, but weak payroll growth and downward revisions increase the risk that claimants remain insured longer. I found no policy mechanism in the checked public release context that would create a discrete level break for this week.

Prior/update/interval: persistence prior is latest-value persistence at 1.814 million, with a reference class of the five fetched same-variant DOL values from May 23 through June 20. I add +0.004 million for recent upward continued-claims momentum, +0.003 million for weak hiring/longer-duration risk, and -0.001 million for low initial-claims inflow, giving 1.820 million. Interval method uses realized dispersion of successive same-variant changes; sigma = 0.006 million from changes +0.015, +0.014, +0.012, and +0.002, so 1.28*sigma = 0.008 million. I widen modestly to about 0.010 million because the target week follows a holiday-adjacent labor market period, weak payroll revisions raise duration risk, and the volatility sample is only four weekly changes.

Point calculation: 1.814 latest + 0.004 momentum + 0.003 weak-hiring duration adjustment - 0.001 low-initial-claims inflow adjustment = 1.820 million. Interval calculation: sigma = 0.006 million; 1.28*sigma = 0.008 million; widened half-width = 0.010 million, so 1.820 - 0.010 = 1.810 and 1.820 + 0.010 = 1.830 million.

Counter-consideration: upside risk is that weak June payrolls and benefit-duration pressure keep more claimants on rolls despite low new claims, which would land above the interval if the first print exceeds 1.830 million. Downside risk is that the latest low initial-claims weeks and seasonal adjustment pull insured unemployment back below 1.810 million. An outside the interval move would most likely come from holiday-week seasonal distortion, a state processing backlog, or a sharper-than-visible hiring/layoff turn.

Review disposition: accepted the leakage critique by removing reliance on any prior Thesis run and explicitly grounding the forecast in official DOL/BLS evidence only; accepted optional clarity by naming the July 9, 2026 first-print release and noting the short volatility sample limitation.

Framing and exact resolver: this forecasts the DOL ETA first-print regular-state-program seasonally adjusted insured unemployment series, also called continued claims, for the week ending June 27, 2026. All anchors use the same SA insured-unemployment variant, not NSA continued weeks claimed or all-program continued claims.

Base rate/reference class: the tightest reference class is recent DOL first-release-adjacent weekly changes in seasonally adjusted insured unemployment. A local-random-walk or latest-value persistence prior is appropriate because this is a one-week-ahead level target with substantial weekly noise but no identified policy break.

Prior local Thesis run check: a prior public repository run for the same target used a 1.820 million point with a much tighter 1.810 to 1.830 million interval. I keep the point direction but widen the interval because the prompt requires sizing from realized dispersion of successive changes rather than a rounded hedged band.

Prior/update/interval: persistence prior = latest same-variant level of 1.814 million. Historical sample = last 13 one-week SA insured-unemployment changes ending June 20, 2026 from the DOL table: -45, +22, -1, -32, -18, +18, -5, +14, -14, +15, +14, +12, +2 thousand. Adjustment components: recent upward momentum +0.005 million, low initial-claims inflow -0.001 million, weaker-duration/continuation pressure +0.002 million, no policy mechanism +0.000 million, giving 1.814 + 0.006 = 1.820 million. Interval method uses realized dispersion of those successive changes; sigma = 0.0207 million, and 1.28*sigma = 0.0265 million, so the 80% interval is 1.820 +/- 0.027 = [1.793, 1.847] million.

Counter-considerations: upside risk is that duration pressure and state-level seasonal layoffs keep more claimants on insured unemployment rolls, which would land above the interval if the July 9 first print exceeds 1.847 million. Downside risk is that lower recent initial claims feed through quickly or the prior increase reverses, which would land below the interval if the first print is under 1.793 million. Outside the interval would most likely require a weekly move larger than about 27 thousand away from the 1.820 million point.

Framing and exact resolver: this forecasts the DOL ETA first-print regular-state-program seasonally adjusted insured unemployment series, also called continued claims, for the week ending June 27, 2026. The target is in millions; the DOL release prose reports persons and the historical table reports thousands for the same SA variant, so all anchors here are converted to millions.

Base rate/reference class: the tightest reference class is recent DOL first-release-adjacent seasonally adjusted insured-unemployment levels. The latest five same-variant values rose from 1.771 million to 1.814 million, with adjacent changes of +0.015, +0.014, +0.012, and +0.002 million. A persistence prior therefore starts near 1.814 million but allows a small continued increase.

Level, momentum, one-off, and policy split: the level is elevated relative to spring 2026 but not jumping. Momentum from late May through mid-June was upward, while the latest weekly increase slowed sharply. Low initial claims reduce near-term inflow pressure, but soft payroll growth and downward revisions raise duration risk. I found no policy mechanism in the checked public release context that would create a discrete break for this week.

Prior/update/interval: persistence prior is latest-value persistence at 1.814 million, with a reference class of the five fetched same-variant DOL values from May 23 through June 20. I add +0.004 million for recent upward continued-claims momentum, +0.003 million for weak hiring/longer-duration risk, and -0.001 million for low initial-claims inflow, giving 1.820 million. Interval method uses realized dispersion of successive same-variant changes; sigma = 0.006 million from changes +0.015, +0.014, +0.012, and +0.002, so 1.28*sigma = 0.008 million. I widen modestly to about 0.010 million because the target week is near the July 4 seasonal-adjustment period and because weak payroll revisions raise duration risk.

Point calculation: 1.814 latest + 0.004 momentum + 0.003 weak-hiring duration adjustment - 0.001 low-initial-claims inflow adjustment = 1.820 million. Interval calculation: sigma = 0.006 million; 1.28*sigma = 0.008 million; widened half-width = 0.010 million, so 1.820 - 0.010 = 1.810 and 1.820 + 0.010 = 1.830 million.

Counter-consideration: upside risk is that weak June payrolls and benefit-duration pressure keep more claimants on rolls despite low new claims, which would land above the interval if the first print exceeds 1.830 million. Downside risk is that low initial claims and seasonal adjustment pull insured unemployment below 1.810 million. An outside the interval move would most likely require holiday-week seasonal distortion, a state processing backlog, or a sharper-than-visible hiring or layoff turn.

Framing and exact resolver: this targets the DOL ETA weekly claims release variant labeled seasonally adjusted Insured Unemployment, not unadjusted continued weeks claimed in all programs. The target week is week ending June 27, 2026, first print, converted from thousands to millions.

Base rate and reference class: for one-week-ahead level forecasts in this DOL series, the strongest base rate is persistence from the latest first-print/revised official level, with recent successive weekly changes used to size uncertainty. The last four observed same-variant SA continued-claims changes were +15,000, +14,000, +12,000, and +2,000, so momentum is positive but decelerating.

Prior/update/interval: persistence prior = 1.814 million from the latest DOL SA insured unemployment print; historical sample = 52 weekly same-variant changes from June 28, 2025 through June 20, 2026 in the DOL table; adjustment components = +0.004 million for recent positive continued-claims momentum, -0.001 million for softer same-week initial claims and the slowing +2,000 latest move, giving point 1.814 + 0.004 - 0.001 = 1.817 million; interval method = one-week change dispersion, sigma = 0.02099 million, half-width = 1.28*sigma = 1.28*0.02099 = 0.0269 million, rounded to about 0.027 million; final implied bounds are 1.817 - 0.028 = 1.789 and 1.817 + 0.026 = 1.843 million.

Upside risk is that late-June continuing unemployment reflects delayed separations or benefit-duration persistence not visible in initial claims, which would land above the interval if SA insured unemployment jumps by more than about 29,000 to above 1.843 million. Downside risk is a sharper reversal in continuing claims after the latest initial-claims easing, which would land below the interval if the level falls under 1.789 million. An outside the interval result would most likely come from a state-level reporting swing or seasonal-adjustment miss in the holiday-adjacent summer weeks.

Framing and exact resolver: the target is the DOL ETA regular state programs seasonally adjusted insured unemployment series, also called continued claims, for the week ending June 27, 2026. The resolution is the advance first print in the weekly claims release, in millions of persons, ignoring later revisions.

Base rate/reference class: for a one-week-ahead continued-claims level forecast, the best outside-view anchor is persistence from the latest first-print/revised DOL SA insured unemployment level plus the empirical distribution of recent weekly changes in the same DOL variant. The latest level is 1.814 million, and the last three weekly changes were +0.014, +0.012, and +0.002 million.

Variant control: every anchor above is the regular state programs Seasonally Adjusted Insured Unemployment series from the DOL weekly claims release, not NSA insured unemployment and not continued weeks claimed in all programs.

Prior/update/interval: persistence prior = 1.814 million from the latest DOL same-variant print; historical sample = 52 successive weekly changes from the June 21, 2025 to June 20, 2026 DOL table; adjustment components = +0.004 million for recent upward continued-claims momentum, -0.001 million because initial claims for June 27 were only 215,000 and down 1,000, +0.000 million for policy mechanism because no UI rule change is needed for the first print; point before ladder = 1.817 million. From the fetched history, sigma = 0.020987 million for successive SA insured-unemployment changes, so 1.28*sigma = 0.026863 million. The ladder-implied 80% half-width is about 0.030 million, 1.12x the 1.28*sigma half-width, modestly wider because holiday/seasonal adjustment around late June can move continued claims even when initial claims are flat.

Ladder: P(X <= 1.780) = 0.06; P(X <= 1.790) = 0.10; P(X <= 1.800) = 0.20; P(X <= 1.810) = 0.36; P(X <= 1.815) = 0.46; P(X <= 1.820) = 0.56; P(X <= 1.830) = 0.72; P(X <= 1.840) = 0.84; P(X <= 1.850) = 0.90; P(X <= 1.860) = 0.95; P(X <= 1.875) = 0.98. Linear interpolation gives p10 = 1.790, p50 = 1.817, and p90 = 1.850 million after rounding to 0.001 million.

Counter-considerations: upside risk would be a repeat of broad unadjusted insured-unemployment increases like the +34,778 NSA move in the latest release, which could lift the SA print toward or above 1.850 million. Downside risk would be a reversal in benefit duration after initial claims held at 215,000, which could pull the SA level below 1.790 million. A sudden state reporting distortion or seasonal-factor miss would land outside the interval.

Review disposition: accepted the suggestion to clarify that the July 9 DOL URL is the expected exact release URL and that the DOL archive/latest-release page is the fallback if the URL pattern changes; accepted the comparable-week source concern by removing the uncited 2025 comparable value from historicalContext.

Derived run: the pointwise median of the CDFs of 3 independent thesis.analyst fast rollouts on this target — median prediction sampling per Turtel et al. 2025 (arXiv:2505.17989). No new model call; this run is a deterministic aggregate of the recorded rollouts at 2026-07-08T02:45:44Z, 2026-07-08T02:45:58Z, 2026-07-08T02:46:42Z. Drivers and resolver fields mirror the rollout closest to the median.

Median CDF quantiles: q10 = 1.793, q50 = 1.82, q90 = 1.843. Constituent points [1.82, 1.82, 1.817] with 80% widths [0.054, 0.02, 0.054]; the median interval inherits the central rollout mass rather than averaging tails.