thesisPrototype

Government data

FY2026 DoD prime contract obligations

What will USAspending API v2 report for Department of Defense agency 097 obligations by award category, contracts row, fiscal year 2026, as a registered query snapshot resolving on the first available post-FY2026 closeout snapshot?

Forecast

505
456.5–553.5
80% prediction interval
gpt-5.5Run record ↗

Cumulative probability

Chance that the outcome is at or below each value.

0%25%50%75%100%383.75505626.25Forecast value

Shaded band: 80% interval (456.5–553.5). Dashed line: point estimate (505).

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

Analysis

§FY2026 DoD prime contract obligations

The registered target is the USAspending API v2 agency 097 obligations_by_award_category contracts row for fiscal_year=2026. The unit, slug, dataPointId, and 2026-10-22 outer-bound resolution date are taken from the canonical ledger target. I attempted the required specs.json check; the hosted fetch returned no visible body, so I kept the ledger-provided slug rather than inventing a replacement.

▸ Reported tool use: usaspending.docsmodel report
Fetched USAspending API endpoint index for /api/v2/agency/<TOPTIER_AGENCY_CODE>/obligations_by_award_category/ and target code 097 with fiscal_year=2026.
↳ USAspending API docs list GET /api/v2/agency/<TOPTIER_AGENCY_CODE>/obligations_by_award_category/ as returning obligations by award category within a single fiscal year; target uses TOPTIER_AGENCY_CODE=097 and fiscal_year=2026. The FY2026 live endpoint was used for target identity and endpoint shape, not as an in-year amount input.
▸ Reported tool use: gao.fpds.dashboardmodel report
Fetched GAO Governmentwide Contracting FY2022 dashboard.
↳ GAO FPDS FY2022 defense agencies: services $205B, products $210B, total proxy $415B; source states data are GAO analysis of Federal Procurement Data System data as of May 2023.
▸ Reported tool use: gao.fpds.dashboardmodel report
Fetched GAO Governmentwide Contracting FY2023 and FY2024 dashboards.
↳ GAO FPDS FY2023 defense agencies: services $230B, products $226B, total proxy $456B. GAO FPDS FY2024 defense agencies: services $240B, products $205B, total proxy $445B.
▸ Reported tool use: gao.fpds.dashboardmodel report
Fetched GAO Governmentwide Contracting FY2025 dashboard.
↳ GAO FPDS FY2025 defense agencies: services $238B, products $253B, total proxy $491B; GAO also states governmentwide contract obligations were $793B and other transaction agreements were $18.2B, not included in contract obligations.
▸ Reported tool use: budget.current_signalmodel report
Fetched CBO FY2026 defense budget analysis and CRS FY2026 NDAA funding summary.
↳ CBO reports DoD FY2026 acquisition funding in the budget request was $384B, 18% more than appropriated for 2025; CRS reports enacted FY2026 NDAA DOD-Military authorization total $855.7B versus enacted FY2025 $849.9B, with procurement $161.7B and RDT&E $145.7B.

Base rate / reference class: the recent public FPDS defense-contract proxy totals are FY2022 415, FY2023 456, FY2024 445, FY2025 491 billions. Last-print persistence is 491.0. The simple one-year-change history is +41, -11, +46, so recent realized changes have mean +25.3 and a wide range from -11 to +46. The draft evidence did not include successful FY2022-FY2025 same-endpoint USAspending history fetches, so this proxy is not treated as the resolver series; I carry extra uncertainty for proxy mismatch and first-snapshot timing.

Model candidates under thesis_model_candidate_v1, computed from fetched history: persistence candidate point=491.0, p10=450.6, p50=491.0, p90=531.4, 80% interval 450.6-531.4, 90% interval 439.0-543.0, interval_method=successive-change residual normal, calibration_n=3, train_cutoff=FY2025; mean-change candidate point=516.3, p10=475.9, p50=516.3, p90=556.7, 80% interval 475.9-556.7, calibration_n=3. I select persistence-plus-current-budget-update because the sample is short and FY2026 budget evidence supports only a modest upward move, not the full mean-change rule.

Prior/update/interval: prior=last-print persistence 491.0 from FY2025 GAO FPDS defense services+products proxy; historical sample=FY2022-FY2025 values 415,456,445,491; adjustment components=+14.0 for FY2026 acquisition funding support and mandatory reconciliation availability, shrunk because enacted base procurement/RDT&E is not uniformly above FY2025 and funds are available through 2029; point=491.0+14.0=505.0. For uncertainty, successive changes are +41,-11,+46, mean change 25.3, sigma = 31.6; 80% half-width roughly 1.28*sigma = 40.4, widened by 20% to 48.5 for GAO-proxy-to-USAspending-target mismatch and October registered-snapshot reporting lag, so interval is 505.0-48.5=456.5 to 505.0+48.5=553.5.

Sanity check: an 80% band with half-width about $48.5B would have covered FY2023, FY2024, and FY2025 moves around a persistence prior, while adding room for the target-definition and snapshot-timing uncertainty that is not measured by the three GAO proxy changes alone.

Upside risk: the result would land above the interval if DoD obligates a large share of reconciliation-funded shipbuilding, missiles, and RDT&E-related contract actions before the registered snapshot or if late September closeout awards are reported unusually quickly. Downside risk: the result would land below the interval if FY2026 contract reporting lags past the October snapshot, if procurement execution is delayed, or if large prior-year deobligations hit the contracts row.

Review disposition: accepted the critique that the draft prior used a GAO FPDS proxy rather than same-endpoint USAspending history, so the final trace labels it as a proxy instead of the resolver series, clarifies the FY2026 API fetch was not used as an amount input, and widens the interval 20% for proxy-definition and first-snapshot timing risk. I did not replace the history with unfetched USAspending values.

Key drivers

  • FY2025 defense contract level persistence
  • FY2026 acquisition funding support
  • mandatory reconciliation funding available through 2029
  • USAspending snapshot timing and proxy-definition 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
USAspending API v2, agency 097 obligations by award category
Resolution date
October 22, 2026· outcome not recorded
Resolution rule
Resolve from USAspending API v2 endpoint /api/v2/agency/097/obligations_by_award_category/?fiscal_year=2026, using the results row where category=contracts and field aggregated_amount, multiplied by 1e-09 to convert dollars to billions USD. This is a registered_query_snapshot target for FY2026; later revisions after the registered snapshot are not part of the first-print resolution. Round to one decimal billion USD. The Thesis ledger commits the expected release window 2026-10-15 through 2026-10-22 and the resolutionDate is the outer bound 2026-10-22.
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 · full · v2.5.8

pre-submit review · completed

The draft is publishable after tightening the historical prior source: resolver fields match the ledger and the forecast math is coherent, but the prior relies on a GAO FPDS proxy instead of the same USAspending target series.

  • warning model_prior: The time-series prior uses GAO FPDS defense services+products proxy totals rather than historical USAspending agency 097 obligations_by_award_category contracts rows from the same resolver endpoint.
  • warning interval: The interval volatility is calculated from only three GAO proxy year-to-year changes, so it may understate uncertainty from target-definition mismatch and USAspending snapshot timing.
  • info optional_suggestion: Clarify whether the FY2026 live API fetch was used only to verify endpoint shape or also as an in-year evidence input.
  • info optional_suggestion: Replace the vague note about specs.json returning no visible body with the exact check outcome or remove it if it is not material to the forecast.

disposition accepted: Review disposition: accepted the critique that the draft prior used a GAO FPDS proxy rather than same-endpoint USAspending history, so the final trace labels it as a proxy instead of the resolver series, clarifies the FY2026 API fetch was not used as an amount input, and widens the interval 20% for proxy-definition and first-snapshot timing risk. I did not replace the history with unfetched USAspending values.

disposition accepted: Review disposition: accepted the critique that the draft prior used a GAO FPDS proxy rather than same-endpoint USAspending history, so the final trace labels it as a proxy instead of the resolver series, clarifies the FY2026 API fetch was not used as an amount input, and widens the interval 20% for proxy-definition and first-snapshot timing risk. I did not replace the history with unfetched USAspending values.

disposition not applicable: Review disposition: accepted the critique that the draft prior used a GAO FPDS proxy rather than same-endpoint USAspending history, so the final trace labels it as a proxy instead of the resolver series, clarifies the FY2026 API fetch was not used as an amount input, and widens the interval 20% for proxy-definition and first-snapshot timing risk. I did not replace the history with unfetched USAspending values.

disposition not applicable: Review disposition: accepted the critique that the draft prior used a GAO FPDS proxy rather than same-endpoint USAspending history, so the final trace labels it as a proxy instead of the resolver series, clarifies the FY2026 API fetch was not used as an amount input, and widens the interval 20% for proxy-definition and first-snapshot timing risk. I did not replace the history with unfetched USAspending values.

Activity artifacts

Complete original trace

§FY2026 DoD prime contract obligations

The registered target is the USAspending API v2 agency 097 obligations_by_award_category contracts row for fiscal_year=2026. The unit, slug, dataPointId, and 2026-10-22 outer-bound resolution date are taken from the canonical ledger target. I attempted the required specs.json check; the hosted fetch returned no visible body, so I kept the ledger-provided slug rather than inventing a replacement.

▸ Reported tool use: usaspending.docsmodel report
Fetched USAspending API endpoint index for /api/v2/agency/<TOPTIER_AGENCY_CODE>/obligations_by_award_category/ and target code 097 with fiscal_year=2026.
↳ USAspending API docs list GET /api/v2/agency/<TOPTIER_AGENCY_CODE>/obligations_by_award_category/ as returning obligations by award category within a single fiscal year; target uses TOPTIER_AGENCY_CODE=097 and fiscal_year=2026. The FY2026 live endpoint was used for target identity and endpoint shape, not as an in-year amount input.
▸ Reported tool use: gao.fpds.dashboardmodel report
Fetched GAO Governmentwide Contracting FY2022 dashboard.
↳ GAO FPDS FY2022 defense agencies: services $205B, products $210B, total proxy $415B; source states data are GAO analysis of Federal Procurement Data System data as of May 2023.
▸ Reported tool use: gao.fpds.dashboardmodel report
Fetched GAO Governmentwide Contracting FY2023 and FY2024 dashboards.
↳ GAO FPDS FY2023 defense agencies: services $230B, products $226B, total proxy $456B. GAO FPDS FY2024 defense agencies: services $240B, products $205B, total proxy $445B.
▸ Reported tool use: gao.fpds.dashboardmodel report
Fetched GAO Governmentwide Contracting FY2025 dashboard.
↳ GAO FPDS FY2025 defense agencies: services $238B, products $253B, total proxy $491B; GAO also states governmentwide contract obligations were $793B and other transaction agreements were $18.2B, not included in contract obligations.
▸ Reported tool use: budget.current_signalmodel report
Fetched CBO FY2026 defense budget analysis and CRS FY2026 NDAA funding summary.
↳ CBO reports DoD FY2026 acquisition funding in the budget request was $384B, 18% more than appropriated for 2025; CRS reports enacted FY2026 NDAA DOD-Military authorization total $855.7B versus enacted FY2025 $849.9B, with procurement $161.7B and RDT&E $145.7B.

Base rate / reference class: the recent public FPDS defense-contract proxy totals are FY2022 415, FY2023 456, FY2024 445, FY2025 491 billions. Last-print persistence is 491.0. The simple one-year-change history is +41, -11, +46, so recent realized changes have mean +25.3 and a wide range from -11 to +46. The draft evidence did not include successful FY2022-FY2025 same-endpoint USAspending history fetches, so this proxy is not treated as the resolver series; I carry extra uncertainty for proxy mismatch and first-snapshot timing.

Model candidates under thesis_model_candidate_v1, computed from fetched history: persistence candidate point=491.0, p10=450.6, p50=491.0, p90=531.4, 80% interval 450.6-531.4, 90% interval 439.0-543.0, interval_method=successive-change residual normal, calibration_n=3, train_cutoff=FY2025; mean-change candidate point=516.3, p10=475.9, p50=516.3, p90=556.7, 80% interval 475.9-556.7, calibration_n=3. I select persistence-plus-current-budget-update because the sample is short and FY2026 budget evidence supports only a modest upward move, not the full mean-change rule.

Prior/update/interval: prior=last-print persistence 491.0 from FY2025 GAO FPDS defense services+products proxy; historical sample=FY2022-FY2025 values 415,456,445,491; adjustment components=+14.0 for FY2026 acquisition funding support and mandatory reconciliation availability, shrunk because enacted base procurement/RDT&E is not uniformly above FY2025 and funds are available through 2029; point=491.0+14.0=505.0. For uncertainty, successive changes are +41,-11,+46, mean change 25.3, sigma = 31.6; 80% half-width roughly 1.28*sigma = 40.4, widened by 20% to 48.5 for GAO-proxy-to-USAspending-target mismatch and October registered-snapshot reporting lag, so interval is 505.0-48.5=456.5 to 505.0+48.5=553.5.

Sanity check: an 80% band with half-width about $48.5B would have covered FY2023, FY2024, and FY2025 moves around a persistence prior, while adding room for the target-definition and snapshot-timing uncertainty that is not measured by the three GAO proxy changes alone.

Upside risk: the result would land above the interval if DoD obligates a large share of reconciliation-funded shipbuilding, missiles, and RDT&E-related contract actions before the registered snapshot or if late September closeout awards are reported unusually quickly. Downside risk: the result would land below the interval if FY2026 contract reporting lags past the October snapshot, if procurement execution is delayed, or if large prior-year deobligations hit the contracts row.

Review disposition: accepted the critique that the draft prior used a GAO FPDS proxy rather than same-endpoint USAspending history, so the final trace labels it as a proxy instead of the resolver series, clarifies the FY2026 API fetch was not used as an amount input, and widens the interval 20% for proxy-definition and first-snapshot timing risk. I did not replace the history with unfetched USAspending values.

calibrated forecast · 80% CI
505[456.5 · 553.5]
Target metadata

Data point: usaspending.dod.prime_contract_obligations.fy2026.registered_query_snapshot

More government data forecasts