The reserve's simulated is 215 , in late January 2027
How low could the Strategic Petroleum Reserve go on the deliveries already announced and the market's , and when could it refill?
In brief
Where the reserve goes on the barrels already promised, a new sale if the US strikes Iran, and the return of the borrowed barrels- The reserve's is 215 mb in late January 2027, from 279.5 in EIA's latest week.→ Exhibit 1
- 100% of the simulated paths pass the 252.4 mb drawdown floor by 31 Dec 2027.→ Exhibit 2
- With no sale after a strike, deliveries already announced still take the median low to 245 mb.→ Exhibit 3
- The shorter return schedule leaves the median reserve 69 mb higher by 30 Jun 2028 than the longer one.→ Exhibit 5
- A US strike on Iran landed by 3 Nov 2026, as Polymarket settled it, and the simulation fixes it there.→ Exhibit 6
| What would change this view | Trigger | Today | Decided by |
|---|---|---|---|
| The reserve leaves the simulated band in the coming weeks | SPR for the week to 27 Nov 2026 below 248.5 or above 260.9 mb | 279.5 mb, week to 6 Nov 2026pending | EIA weekly report, 2 Dec 2026 |
| The reserve leaves the simulated band by the end of the year | SPR for the week to 25 Dec 2026 below 217.5 or above 227.7 mb | 279.5 mb, week to 6 Nov 2026pending | EIA weekly report, 30 Dec 2026 |
| DOE's September award differs from the inference | an award from the 29 Sep offer other than 23 mb | not published; inferred as 40 − 17 mbpending | DOE's award notice |
| The borrowed barrels come back late | no weekly rise in EIA's SPR stock by the week to 25 Jun 2027 | last weekly rise in the week to 6 Mar 2026pending | EIA weekly report, 30 Jun 2027 |
The reasoning behind each point
- lowThe reserve's median low is 215 mb in late January 2027, from 279.5 mb in EIA's latest week.
Nine paths in ten bottom between 213 and 224 mb. Every path has the strike, so the spread comes from the uptake, a sale's weekly pace and when the returns start.
- line100% of the simulated paths pass the 252.4 mb drawdown floor by 31 Dec 2027.
The line is the floor of the President's limited-drawdown authority. Exchanges and emergency sales are not bound by it, so it is a line to report against, not a stop.
- floorWith no sale after a strike, deliveries already announced still take the median low to 245 mb.
The 23 mb inferred from the 29 Sep offer and the 16 Oct offer carry the reserve down on their own.
- refillThe shorter return schedule leaves the median reserve 69 mb higher by 30 Jun 2028 than the longer one.
What the return timing does to Brent is on the producers-and-refills page, which repays the reserve on this page's inputs and copies them with their date.
- strikeA US strike on Iran landed by 3 Nov 2026, as Polymarket settled it, and the simulation fixes it there.
Polymarket settled its “ceasefire continues through …” contracts from 3 Nov onward as broken. Every path has the strike in the week to 6 Nov 2026, and the weeks since it are EIA's record.
Next: Brent price simulation — the price under the same strike odds
Half the paths take the reserve below 215 mb before it refills
Where does the reserve go week by week from today's level, and how often does it pass the drawdown floor?Exhibit 1, at the top of the page: The reserve's weekly level and its simulated range.
The median low of 215 mb is a level the reserve last held in 1981.
The reserve stands at 279.5 mb in EIA's week to 6 Nov 2026. Half the simulated paths take it below 215 mb, the median low falling in late January 2027.
Exhibit 1 · The reserve's weekly level and its simulated range
About this exhibit
Exhibit 2 · Share of simulated paths past the President's drawdown floor, by week
100% of the paths reach the drawdown floor of 252.4 mb by 31 Dec 2027.
About this exhibit
The floor is a line to report against, not a stop: exchanges and emergency sales are not bound by it. The next section shows which flows set the low.
Deliveries already announced set the floor and a strike sets the depth
Which flows take the reserve down, and which bring it back?Deliveries already announced take 32 mb out of the reserve, and with no sale after a strike half the paths still bottom below 245 mb.
Exhibit 3 · Weekly flows of the reserve by kind, average of the simulated paths
Returns start on a day drawn from 1 Jan 2027 to 30 Jun 2027, so the average return rises over that half-year.
About this exhibit
Exhibit 4 · Distribution of the reserve's simulated low, with the year last seen
Paths at the 5th percentile bottom below 213 mb.
About this exhibit
A sale of 90 mb after a strike takes the median low to 161 mb instead. The next section asks when the borrowed barrels come back, and what that does to Brent.
The shorter return schedule leaves the reserve 69 mb higher
When do the borrowed barrels come back, and how full is the reserve by the horizon?By 30 Jun 2028 the median reserve stands at 344 mb with returns over 104 weeks, and at 413 mb over 52.
Exhibit 5 · The reserve's low and refill under each sale and return setting
| Sale after a strike | Returns | After a strike | Risk after 31 Dec | Low | Low, 1 in 20 | Week of the low | Past 252.4 by end-2027 | end of June 2027 | end of June 2028 | Strike drawn |
|---|---|---|---|---|---|---|---|---|---|---|
| None | 1 year | No delay | None priced | 244.9 | 243.1 | 22 Jan 2027 | 68.5% | 283.8 | 443.1 | 100.0% |
| None | 1 year | 26 weeks later | None priced | 244.5 | 240.3 | 22 Jan 2027 | 73.9% | 244.6 | 366.0 | 100.0% |
| None | 2 years | No delay | None priced | 244.9 | 243.1 | 22 Jan 2027 | 68.5% | 271.7 | 374.0 | 100.0% |
| None | 2 years | 26 weeks later | None priced | 244.5 | 240.3 | 22 Jan 2027 | 74.0% | 244.6 | 323.3 | 100.0% |
| 30 mb | 1 year | No delay | None priced | 214.9 | 213.1 | 22 Jan 2027 | 100.0% | 253.8 | 413.1 | 100.0% |
| 30 mb | 1 year | 26 weeks later | None priced | 214.5 | 210.3 | 22 Jan 2027 | 100.0% | 214.6 | 336.0 | 100.0% |
| 30 mbdefault | 2 years | No delay | None priced | 214.9 | 213.1 | 22 Jan 2027 | 100.0% | 241.7 | 344.0 | 100.0% |
| 30 mb | 2 years | 26 weeks later | None priced | 214.5 | 210.3 | 22 Jan 2027 | 100.0% | 214.6 | 293.3 | 100.0% |
| 60 mb | 1 year | No delay | None priced | 185.2 | 183.1 | 29 Jan 2027 | 100.0% | 223.8 | 383.1 | 100.0% |
| 60 mb | 1 year | 26 weeks later | None priced | 184.5 | 180.3 | 5 Feb 2027 | 100.0% | 184.6 | 306.0 | 100.0% |
| 60 mb | 2 years | No delay | None priced | 185.2 | 183.1 | 29 Jan 2027 | 100.0% | 211.7 | 314.0 | 100.0% |
| 60 mb | 2 years | 26 weeks later | None priced | 184.5 | 180.3 | 5 Feb 2027 | 100.0% | 184.6 | 263.3 | 100.0% |
| 90 mb | 1 year | No delay | None priced | 162.5 | 153.2 | 5 Mar 2027 | 100.0% | 193.8 | 353.1 | 100.0% |
| 90 mb | 1 year | 26 weeks later | None priced | 154.5 | 150.3 | 12 Mar 2027 | 100.0% | 154.6 | 276.0 | 100.0% |
| 90 mb | 2 years | No delay | None priced | 161.2 | 153.2 | 5 Mar 2027 | 100.0% | 181.7 | 284.0 | 100.0% |
| 90 mb | 2 years | 26 weeks later | None priced | 154.5 | 150.3 | 12 Mar 2027 | 100.0% | 154.6 | 233.3 | 100.0% |
Each row is one setting: its median low, its low at the 5th percentile and its median level at each half-year.
About this exhibit
Both schedules run no faster than the sites take the barrels, between 0.252 and 0.672 mb a day. What the return timing does to Brent is on the producers-and-refills page, which repays the reserve on these inputs.
Polymarket settled that the US struck Iran by 3 Nov 2026
How does the market time a US strike, and how far do its contracts reach?Polymarket settled its ceasefire contracts broken by 3 Nov 2026, so every path has the strike in the week to 6 Nov 2026.
Exhibit 6 · Market-implied chance of no US strike on Iran, day by day
| Through | Contract | Settled |
|---|---|---|
| 20 Sept 2026 | September 20 | the ceasefire held |
| 25 Sept 2026 | September 25 | the ceasefire held |
| 30 Sept 2026 | September 30 | the ceasefire held |
| 5 Oct 2026 | October 5 | the ceasefire held |
| 7 Oct 2026 | October 7 | the ceasefire held |
| 9 Oct 2026 | October 9 | the ceasefire held |
| 12 Oct 2026 | October 12 | the ceasefire held |
| 15 Oct 2026 | October 15 | the ceasefire held |
| 16 Oct 2026 | October 16 | the ceasefire held |
| 17 Oct 2026 | October 17 | the ceasefire held |
| 18 Oct 2026 | October 18 | the ceasefire held |
| 22 Oct 2026 | October 22 | the ceasefire held |
| 24 Oct 2026 | October 24 | the ceasefire held |
| 28 Oct 2026 | October 28 | the ceasefire held |
| 31 Oct 2026 | October 31 | the ceasefire held |
| 2 Nov 2026 | November 2 | the ceasefire held |
| 3 Nov 2026 | November 3 | the ceasefire broke |
| 4 Nov 2026 | November 4 | the ceasefire broke |
| 5 Nov 2026 | November 5 | the ceasefire broke |
| 6 Nov 2026 | November 6 | the ceasefire broke |
| 7 Nov 2026 | November 7 | the ceasefire broke |
| 8 Nov 2026 | November 8 | the ceasefire broke |
| 15 Nov 2026 | November 15 | the ceasefire broke |
| 22 Nov 2026 | November 22 | the ceasefire broke |
| 30 Nov 2026 | November 30 | the ceasefire broke |
| 31 Dec 2026 | December 31 | the ceasefire broke |
The contracts dated before 3 Nov 2026 settled held, and those from it on settled broken.
About this exhibit
No further strike is drawn. A new sale, where a setting asks for one, starts after the strike's week. Weeks before the first simulated one are EIA's record.
Every input and where it comes from
Each input with its value, its kind and its source
Exhibit 7 · Each input to the simulation with its value, kind and source
| Input | In the model | Kind | Source |
|---|---|---|---|
| Starting level | 279.535 mb, week to 6 Nov 2026 | EIA | EIA, weekly SPR stocks (WCSSTUS1); Weekly Petroleum Status Report, Table 1, 12 Nov 2026 |
| Weather exchange, October | 2–4 mb over the first 3 weeks | Published range | DOE notice, via BIC Magazine — emergency exchange authorised for ExxonMobil and BP, 13 Oct 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3 |
| The 29 Sep offer's award | 23 mb = 40 offered − 17 re-offered, over 13 Nov to 25 Dec 2026 | Inferred | DOE request for proposals, as reported by Oil & Gas Journal — US seeks to release another 40 million bbl from SPR, 30 Sept 2026read in research note: the Middle East and Russia–Ukraine baseline (14 Oct 2026) §3a timeline (29 Sep)Reuters, via BOE Report — 17 mb of the 29 Sep offer re-offered on 16 Oct, 16 Oct 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3 |
| The 16 Oct offer | 17 mb × an uptake drawn from 2026's three: 52.6%, 1.2%, 57.5%; over 27 Nov to 22 Jan 2027assumption: delivery beginning about six weeks after the offer, for nine weeks — the 29 Sep offer's lag and length (prototype note §3) | 2026 record | Atlantic Council — the March request for proposals: 45.2 mb awarded of 86 mb offered, 2 Oct 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3Oil & Gas Journal — US seeks to release another 40 million bbl from SPR despite low inventory levels, 30 Sept 2026read in research note: the Middle East and Russia–Ukraine baseline (14 Oct 2026) §2 (the reserve test's row); the prototype cites the same figure from DOE via Reuters |
| Weekly pace of a new sale | drawn from 2026's 23 delivery weeks, 0.38–9.92 mb a week, capped at 18.9 mb | EIA record | DOE, Strategic Petroleum Reserve inventory sheet (monthly movements) — deliveries from 20 March, peaking in May and nearly stopping in September, 9 Oct 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3; the prototype's Reserves exhibit 9GAO, December 2025, via the Atlantic Council — drawdown capability available, 2 Oct 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3 |
| Timing of a US strike | fixed: the first week ending on or after 3 Nov 2026, when the venue settled the November 3 contract broken | Market record | Polymarket, “US-Iran ceasefire continues through...?” — settled contracts (10 broken), 16 Nov 2026 |
| A new sale after a strike | 0, 30, 60 or 90 mb, starting 2 weeks after the strike | Setting | IEA collective actions of 23 June 2011 and 1 March 2022 — the US share: 30 mb in each, 1 Mar 2022read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3DOE, the March 2026 collective action — announced 11 Mar, first shipments 20 Mar, 20 Mar 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3 |
| Barrels owed back | 133 mb plus the new loans, times 1 + a premium of 18–28% | Published | DOE notice of the 29 Sep exchange — more than 133 mb awarded across four completed exchanges, 29 Sept 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3Atlantic Council — return premiums of 18–22% on the first exchanges and 28% on later tranches, 2 Oct 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3 |
| Start of the returns | a day drawn from 1 Jan 2027 to 30 Jun 2027 | Published | DOE — the March exchange's borrowed barrels returned “by next year”, 17 Apr 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3 |
| Pace of the returns | the slower of the contract schedule (one or two years) and a fill rate of 0.252–0.672 mb a day0.252 mb a day is EIA's best week since 2023, 1.764 mb in the week to 30 Aug 2024, over seven days. | Published and EIA record | Atlantic Council, citing GAO (Dec 2025) — the sum of the sites' maximum fill rates, 0.672 mb/d, 2 Oct 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3DOE notice of the 29 Sep exchange — about 200 mb to come back over the next year, 29 Sept 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3Energy Secretary Chris Wright, via Reuters — full return of the exchanged barrels beyond late 2028, 29 Sept 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3 |
| Weather exchanges in 2027 | 0.32 a season, 1.0–5.4 mb each, June to November | History | DOE, history of SPR emergency exchanges, with the 13 Oct 2026 exchange — weather and hurricane exchanges, 2002–2026, 13 Oct 2026read in research note: the Gulf, diesel and reserves prototype (17 Oct 2026) §3 |
| The limited-drawdown line | 252.4 mb, a reporting line, not a stopA reporting line, not a stop: exchanges and §161(d) sales are not bound by it (Atlantic Council, 2 Oct 2026; prototype note §4.1). | Statute | 42 U.S.C. §6241(h), as DOE's SPR FAQs state it — research note: Oman postponement spr (14 Sep 2026) §B2a authorities, 14 Sept 2026 |
About this exhibit
Appendix
How one path is made, what each draw is, and the checks the run passes83 checks, all pass
The Strategic Petroleum Reserve run forward one week at a time from EIA's latest week to the end of June 2028. Each path removes the barrels DOE has already announced — the October weather exchange, the September award and the October offer at one of 2026's uptakes — and, if a US strike on Iran falls in the horizon, a new sale of the size the settings choose, timed by the market's own term structure of strike odds. From the first half of 2027 the borrowed barrels come back, with their premium, at the slower of the contract's pace and what the sites can physically take; the producers-and-refills page prices what that timing does to Brent, on these same inputs. Every input is a published figure, EIA's own record or a market price, each with its source; the one inference, the September award, is printed with its arithmetic. Model-conditional: the ranges are what these inputs imply, not a forecast of policy.
One path, week by week
| Symbol | Meaning | From | Assumed or estimated |
|---|---|---|---|
| EIA's stock in the week to 6 Nov 2026, 279.535 mb | EIA, WCSSTUS1; Table 1 | – | |
| the September award, 23 mb (40 − 17), over n_Sep weeks | DOE; Reuters (inferred) | – | |
| an uptake drawn from 2026's three offers | Atlantic Council; Oil & Gas Journal; inferred | – | |
| the week of the US strike, fixed where the venue settled it | Polymarket | – | |
| a weekly draw resampled from 2026's delivery weeks, capped at 18.9 mb | EIA; GAO via the Atlantic Council | – | |
| the new sale's size (a setting) and what it has delivered so far | setting | – | |
| the owed pool, the contract weeks (52 or 104), the fill rate, the premium and the first return week | DOE; Atlantic Council; GAO; EIA | – |
Model-conditional: what the inputs imply, path by path; not a forecast. Every other symbol is a published figure, EIA's record or a market price.
Exhibit 8 · The engine reproduces the reference's 6 settings on its 17 Oct inputs
Model-conditional: every range on this page is what the model's inputs imply, drawn from 20,000 seeded paths; it is not a forecast of policy or of the reserve. The strike's timing is no longer a probability: Polymarket settled its ceasefire contracts broken by 3 Nov 2026, and the size of any new sale is a setting the reader chooses.
- One path
Each path starts at EIA's latest weekly stock and moves one week at a time to the horizon. It removes the October weather exchange over three weeks, the September award evenly over November and December, and the 16 October offer at one of 2026's three uptakes over the weeks to 22 January 2027. The strike is fixed in the week the venue settled it, on every path; a new sale of the chosen size starts two weeks after that week, and any of its weeks before the first simulated one are EIA's record, at weekly draws resampled from 2026's delivery weeks and capped at the effective drawdown rate. The next season's weather exchanges come out in June to November 2027. From a day drawn in the first half of 2027 the owed barrels come back at the slower of the contract's pace and a physical fill rate.
- The settings
16 settings: a new sale of 0, 30, 60 or 90 mb after a strike; returns over one or two years; returns deferred 0 or 26 weeks after a strike; no further strike is drawn after the settled one. Each is precomputed; the controls select one.
- One set of draws
One seed (20261017) draws every random input once — the strike's uniform, the exchange sizes, the uptake, the weekly pace of a sale, the weather exchanges, the premium, the start of the returns and the fill rate — and every setting reads the same draws (common random numbers). A larger sale therefore lowers every path, and the run checks that it does.
- The inference
The September award is not published. It is inferred as the 40 mb offered on 29 Sep less the 17 mb DOE re-offered on 16 Oct: 23 mb, delivered over November and December as DOE scheduled the offer.
- The line
A reporting line, not a stop: exchanges and §161(d) sales are not bound by it (Atlantic Council, 2 Oct 2026; prototype note §4.1).
- Precision
Levels are printed to 0.1 mb and shares of paths to 0.1 percentage point. A Monte Carlo comparison passes within four standard errors of the difference of two independent runs, from the run's own order statistics for a percentile and the binomial error for a share.
The reference run (research note: the Gulf, diesel and reserves prototype (17 Oct 2026)) against this run
Each cell: the reference's figure, then this run on the reference's inputs · this run on the latest inputs; 20,000 paths, compared within 4 standard errors (the checks).
| Setting | Median low, mb | Low, 1 in 20 | Week of the low | Past 252.4 by end-2027 | End of June 2027 | End of June 2028 | Strike drawn |
|---|---|---|---|---|---|---|---|
| Default | 226.3226.2 · 214.9 | 216.7216.7 · 213.1 | 22 Jan22 Jan · 22 Jan | 89.1%89.7% · 100.0% | 254.8255.0 · 241.7 | 357.1357.3 · 344.0 | 67.5%68.2% · 100.0% |
| No new sale | 248.4248.4 · 244.9 | 246.5246.5 · 243.1 | 22 Jan22 Jan · 22 Jan | 66.4%67.1% · 68.5% | 275.6275.7 · 271.7 | 377.7377.8 · 374.0 | 67.5%68.2% · 100.0% |
| 90 mb on strike | 169.6169.5 · 161.2 | 156.9157.0 · 153.2 | 5 Mar5 Mar · 5 Mar | 89.2%89.7% · 100.0% | 197.3197.3 · 181.7 | 299.5299.6 · 284.0 | 67.5%68.2% · 100.0% |
| Strike risk carried past 31 Dec | 221.4221.4 · – | 216.7216.6 · – | 22 Jan22 Jan · – | 95.3%95.3% · – | 249.0248.9 · – | 347.9348.0 · – | 99.3%99.2% · – |
| Returns wait 6 months after a strike | 225.9225.8 · 214.5 | 214.8214.8 · 210.3 | 22 Jan22 Jan · 22 Jan | 89.1%89.7% · 100.0% | 226.1226.1 · 214.6 | 308.8309.0 · 293.3 | 67.5%68.2% · 100.0% |
| Returns over 1 year (DOE schedule) | 226.3226.2 · 214.9 | 216.7216.7 · 213.1 | 22 Jan22 Jan · 22 Jan | 89.1%89.6% · 100.0% | 267.8267.8 · 253.8 | 419.1419.3 · 413.1 | 67.5%68.2% · 100.0% |
About this exhibit
How the numbers are made
spr_weekly + wpsr_table1 ──▶ the start: EIA's latest week, checked against Table 1; the weekly draws of 2026's delivery weeks; the best weekly rise since 2023 as the slowest fill ──▶ the reserve's record
spr_monthly + spr_weekly ──▶ the year the reserve last stood at or below each level: monthly since October 1977, then weekly ──▶ the last-seen years
strike_contracts ──▶ the survival curve: the mid of bid and ask of each dated contract with $100,000 or more traded; a constant daily hazard between dates; the midterms contract as a cross-check ──▶ the strike's timing
oct_exchange + sep_offer + oct_reoffer + uptake_march + uptake_june + prior_awarded + premium + return_start + fill_max + drawdown + release_lag + weather_history + line_161h ──▶ the announced flows, the owed barrels and the pace limits as published; the September award inferred as 40 − 17 mb ──▶ the published flows
the reserve's record + the strike's timing + the published flows ──▶ themes._spr_mc: one seed, 20,000 paths, common random numbers over the 32 settings; S_t = S_{t−1} − out_t + in_t ──▶ the paths
the paths + the last-seen years ──▶ quantiles by week, the share past 252.4 mb, the low's distribution and histogram, the half-year levels; seeded rerun, mass balance, loop, independent seed and reference checks ──▶ the settings
the paths ──▶ the average weekly flows by kind (announced deliveries, weather exchanges, a sale after a strike, returns), each setting ──▶ the flows
the settings + the flows ──▶ the two return schedules at the horizon; the reproduction checks (validate, check, publish) ──▶ the data the page readsSources
- publicU.S. Energy Information Administration: weekly Strategic Petroleum Reserve stocks, monthly Strategic Petroleum Reserve stocks since October 1977, and the Weekly Petroleum Status Report's Table 1
Source: U.S. Energy Information Administration · https://www.eia.gov/dnav/pet/hist_xls/WCSSTUS1w.xls
- publicPolymarket, Gamma API: the “US-Iran ceasefire continues through …?” contracts and the “… through the midterms?” contract, quoted on 16 Nov 2026 at 21:57 UTC (mid of bid and ask)
Source: Polymarket · https://gamma-api.polymarket.com/events?slug=us-iran-ceasefire-continues-throughptptpt
- computed hereFigures quoted from the U.S. Department of Energy's notices, the Atlantic Council (2 Oct 2026), GAO (Dec 2025), Reuters and Oil & Gas Journal, each with its document and date in the inputs table
Quoted as published; each input names its document and date
- computed hereThe simulation: 20,000 seeded paths with common random numbers over 32 settings, computed here
Model-conditional distributions computed by the site's engine; not forecasts.
Reproduction checks
| Check | Published | Computed | Tolerance | Result |
|---|---|---|---|---|
| Start: EIA's weekly SPR stock for the week to 6 Nov 2026 (WCSSTUS1, 279.535 mb) equals Table 1 of the Weekly Petroleum Status Report (279.535 mb; abs difference, mb) | 0 | 0 | 0.0005 | pass |
| Start: Table 1's week is the workbook's latest week | 2026-11-06 | 2026-11-06 | 0 | pass |
| Delivery weeks: every week from 27 Mar 2026 to 28 Aug 2026 is a fall in EIA's SPR stock (weeks without one) | 0 | 0 | 0 | pass |
| Delivery weeks: the first weekly fall after the 11 Mar 2026 announcement is the window's first week | 2026-03-27 | 2026-03-27 | 0 | pass |
| Strike odds: every contract of the event is dated from its title (26 contracts; unparsed) | 0 | 0 | 0 | pass |
| Strike: the venue's first contract settled broken (6588210, November 3, dated 3 Nov 2026) is dated on or before 17 Nov 2026, and every later contract has settled broken too (10 broken; later contracts held or not settled) | 0 | 0 | 0 | pass |
| Strike: the contracts dated before it settled held (16; settled otherwise) | 0 | 0 | 0 | pass |
| Weeks: the run starts the week after EIA's latest (13 Nov 2026) and ends on the horizon (30 Jun 2028); 86 Fridays seven days apart (gaps) | 0 | 0 | 0 | pass |
| Inference: the September award is the 29 Sep offer less the 16 Oct re-offer (40 − 17 mb) and equals the prototype's (mb) | 23 | 23 | 0 | pass |
| Inference: 2026's three uptakes (awarded ÷ offered: March, June and the inferred September) equal the prototype's (max abs difference) | 0 | 0 | 0 | pass |
| Arithmetic: the effective drawdown, 61% × 4.415 mb/d × 7 days, equals the prototype's cap (mb a week) | 18.85205 | 18.85205 | 0 | pass |
| Arithmetic: weather exchanges a season, 8 in 25 seasons, equal the prototype's rate | 0.32 | 0.32 | 0 | pass |
| Mass balance: on every path of every setting the outflows add to their parts and the level closes, S0 − S_T = Σ out − Σ in, with the returns at the slower of the two paces (16 settings × 20,000 paths; max abs residual, mb) | 0 | 0 | 0 | pass |
| Loop: 25 sampled paths of every setting, re-simulated one week at a time as the prototype writes a path, equal the vectorised paths (max abs difference, mb) | 0 | 0 | 0 | pass |
| Quantiles: the 5th, 25th, 50th, 75th and 95th percentiles are ordered at every week of every setting (violations) | 0 | 0 | 0 | pass |
| Line: the share of paths that have passed 252.4 mb never falls from one week to the next, in any setting (falls) | 0 | 0 | 0 | pass |
| Common draws: path by path, a larger sale after a strike never raises the low (paths where it does, over the 4 other settings) | 0 | 0 | 0 | pass |
| Seed: a second run of the default setting from seed 20261017 reproduces every path (max abs difference, mb) | 0 | 0 | 0 | pass |
| Strike: every path has the strike in the week ending 6 Nov 2026, the first on or after the venue's 3 Nov 2026 (share of paths) | 1 | 1 | 0 | pass |
| Independent seed: the default's low, 5th percentile, from seed 20261018 (213.08 mb) within 4 standard errors of the difference (mb) | 213.084927 | 213.077729 | 0.054411 | pass |
| Independent seed: the default's low, 50th percentile, from seed 20261018 (214.98 mb) within 4 standard errors of the difference (mb) | 214.933832 | 214.979125 | 0.120426 | pass |
| Independent seed: the default's low, 95th percentile, from seed 20261018 (224.04 mb) within 4 standard errors of the difference (mb) | 224.034882 | 224.037428 | 0.042477 | pass |
| Independent seed: the share past 252.4 mb by 31 Dec 2027 (100.00%) within 4 standard errors | 1 | 1 | 0 | pass |
| Independent seed: the median level at the end of June 2027 within 4 standard errors (mb) | 241.690061 | 242.503389 | 1.055931 | pass |
| Independent seed: the median level at the end of June 2028 within 4 standard errors (mb) | 343.950136 | 344.624199 | 1.021045 | pass |
| Independent seed: the median week of the low is the same week | 2027-01-22 | 2027-01-22 | 0 | pass |
| Expected flows: with no new sale and returns over one year, the exact average of the barrels delivered and returned to each week equals this run's mean of 20,000 paths within 4 standard errors (largest gap, standard errors) | 0 | 2.556212 | 4 | pass |
| Expected flows: with no new sale and returns over two years, the exact average of the barrels delivered and returned to each week equals this run's mean of 20,000 paths within 4 standard errors (largest gap, standard errors) | 0 | 1.899373 | 4 | pass |
| Flows: in every setting a sale's average weekly barrels are its outflow less a no-sale setting's, never negative, and sum to its paths' mean sale (max abs residual, mb) | 0 | 0 | 0 | pass |
| Flows: the weather exchanges' average is never negative and sums to the paths' mean weather exchanges (max abs residual, mb) | 0 | 0 | 0 | pass |
| Flows: announced deliveries, weather exchanges and a sale add up to each setting's average outflow, week by week (max abs residual, mb) | 0 | 0 | 0 | pass |
| Start: EIA's weekly SPR stock in the prototype's fetch of 17 Oct 2026 (week to 9 Oct 2026, data/raw/eia/2026-10-17) equals the prototype's 282.983 mb (abs difference, mb) | 282.983 | 282.983 | 0.0005 | pass |
| Delivery weeks: the prototype's fetch of 17 Oct 2026 has as many as the prototype (count) | 23 | 23 | 0 | pass |
| Delivery weeks: each week's draw in the prototype's fetch of 17 Oct 2026 equals the prototype's, printed to two decimals (23 weeks; max abs difference, mb) | 0 | 0.005 | 0.0051 | pass |
| Fill: the best weekly rise in EIA's SPR stock 2023–2026 in the prototype's fetch of 17 Oct 2026 (1.764 mb in the week to 30 Aug 2024) over seven days equals the prototype's slowest fill, printed to two decimals (mb a day) | 0.25 | 0.252 | 0.0051 | pass |
| Strike odds: the anchor dates of the prototype's fetch of 17 Oct 2026 are the prototype's (the same contracts pass the $100,000 floor) | 2026-10-172026-10-182026-10-222026-10-312026-11-152026-11-302026-12-31 | 2026-10-172026-10-182026-10-222026-10-312026-11-152026-11-302026-12-31 | 0 | pass |
| Strike odds: each anchor of the prototype's fetch of 17 Oct 2026 is within 1 point of the prototype's, quoted 81 minutes earlier (largest difference 0.60 points, 18 Oct 2026; max abs difference) | 0 | 0.006 | 0.01 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Default, low p50 against the prototype's 226.3 within 4 standard errors of two independent runs plus half its last digit | 226.3 | 226.164024 | 0.262172 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Default, low p5 against the prototype's 216.7 within 4 standard errors of two independent runs plus half its last digit | 216.7 | 216.716407 | 0.168256 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Default, passed 2027 against the prototype's 0.8908 within 4 standard errors of two independent runs plus half its last digit | 0.8908 | 0.8966 | 0.012229 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Default, jun27 p50 against the prototype's 254.8 within 4 standard errors of two independent runs plus half its last digit | 254.8 | 255.024862 | 1.138051 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Default, jun28 p50 against the prototype's 357.1 within 4 standard errors of two independent runs plus half its last digit | 357.1 | 357.250361 | 0.992384 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Default, p strike against the prototype's 0.6748 within 4 standard errors of two independent runs plus half its last digit | 0.6748 | 0.68195 | 0.018679 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Default, the median week of the low is the prototype's | 2027-01-22 | 2027-01-22 | 0 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): No new sale, low p50 against the prototype's 248.4 within 4 standard errors of two independent runs plus half its last digit | 248.4 | 248.379493 | 0.171884 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): No new sale, low p5 against the prototype's 246.5 within 4 standard errors of two independent runs plus half its last digit | 246.5 | 246.53811 | 0.110325 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): No new sale, passed 2027 against the prototype's 0.6639 within 4 standard errors of two independent runs plus half its last digit | 0.6639 | 0.67065 | 0.018849 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): No new sale, jun27 p50 against the prototype's 275.6 within 4 standard errors of two independent runs plus half its last digit | 275.6 | 275.714168 | 1.00223 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): No new sale, jun28 p50 against the prototype's 377.7 within 4 standard errors of two independent runs plus half its last digit | 377.7 | 377.776907 | 0.915129 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): No new sale, p strike against the prototype's 0.6748 within 4 standard errors of two independent runs plus half its last digit | 0.6748 | 0.68195 | 0.018679 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): No new sale, the median week of the low is the prototype's | 2027-01-22 | 2027-01-22 | 0 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): 90 mb on strike, low p50 against the prototype's 169.6 within 4 standard errors of two independent runs plus half its last digit | 169.6 | 169.506013 | 1.265592 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): 90 mb on strike, low p5 against the prototype's 156.9 within 4 standard errors of two independent runs plus half its last digit | 156.9 | 156.988758 | 0.219664 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): 90 mb on strike, passed 2027 against the prototype's 0.8918 within 4 standard errors of two independent runs plus half its last digit | 0.8918 | 0.8966 | 0.012229 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): 90 mb on strike, jun27 p50 against the prototype's 197.3 within 4 standard errors of two independent runs plus half its last digit | 197.3 | 197.272855 | 1.319498 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): 90 mb on strike, jun28 p50 against the prototype's 299.5 within 4 standard errors of two independent runs plus half its last digit | 299.5 | 299.620407 | 1.71493 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): 90 mb on strike, p strike against the prototype's 0.6748 within 4 standard errors of two independent runs plus half its last digit | 0.6748 | 0.68195 | 0.018679 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): 90 mb on strike, the median week of the low is the prototype's | 2027-03-05 | 2027-03-05 | 0 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Strike risk carried past 31 Dec, low p50 against the prototype's 221.4 within 4 standard errors of two independent runs plus half its last digit | 221.4 | 221.392613 | 1.72176 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Strike risk carried past 31 Dec, low p5 against the prototype's 216.7 within 4 standard errors of two independent runs plus half its last digit | 216.7 | 216.645662 | 0.119 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Strike risk carried past 31 Dec, passed 2027 against the prototype's 0.9532 within 4 standard errors of two independent runs plus half its last digit | 0.9532 | 0.9527 | 0.008541 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Strike risk carried past 31 Dec, jun27 p50 against the prototype's 249.0 within 4 standard errors of two independent runs plus half its last digit | 249 | 248.904898 | 1.362399 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Strike risk carried past 31 Dec, jun28 p50 against the prototype's 347.9 within 4 standard errors of two independent runs plus half its last digit | 347.9 | 348.025653 | 1.148747 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Strike risk carried past 31 Dec, p strike against the prototype's 0.9928 within 4 standard errors of two independent runs plus half its last digit | 0.9928 | 0.99185 | 0.003646 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Strike risk carried past 31 Dec, the median week of the low is the prototype's | 2027-01-22 | 2027-01-22 | 0 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns wait 6 months after a strike, low p50 against the prototype's 225.9 within 4 standard errors of two independent runs plus half its last digit | 225.9 | 225.803019 | 0.905711 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns wait 6 months after a strike, low p5 against the prototype's 214.8 within 4 standard errors of two independent runs plus half its last digit | 214.8 | 214.787306 | 0.588634 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns wait 6 months after a strike, passed 2027 against the prototype's 0.8908 within 4 standard errors of two independent runs plus half its last digit | 0.8908 | 0.8966 | 0.012229 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns wait 6 months after a strike, jun27 p50 against the prototype's 226.1 within 4 standard errors of two independent runs plus half its last digit | 226.1 | 226.052551 | 0.219374 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns wait 6 months after a strike, jun28 p50 against the prototype's 308.8 within 4 standard errors of two independent runs plus half its last digit | 308.8 | 309.000375 | 1.513575 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns wait 6 months after a strike, p strike against the prototype's 0.6748 within 4 standard errors of two independent runs plus half its last digit | 0.6748 | 0.68195 | 0.018679 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns wait 6 months after a strike, the median week of the low is the prototype's | 2027-01-22 | 2027-01-22 | 0 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns over 1 year (DOE schedule), low p50 against the prototype's 226.3 within 4 standard errors of two independent runs plus half its last digit | 226.3 | 226.203374 | 0.244972 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns over 1 year (DOE schedule), low p5 against the prototype's 216.7 within 4 standard errors of two independent runs plus half its last digit | 216.7 | 216.717014 | 0.16689 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns over 1 year (DOE schedule), passed 2027 against the prototype's 0.8908 within 4 standard errors of two independent runs plus half its last digit | 0.8908 | 0.89645 | 0.012237 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns over 1 year (DOE schedule), jun27 p50 against the prototype's 267.8 within 4 standard errors of two independent runs plus half its last digit | 267.8 | 267.818165 | 1.567313 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns over 1 year (DOE schedule), jun28 p50 against the prototype's 419.1 within 4 standard errors of two independent runs plus half its last digit | 419.1 | 419.275973 | 0.897291 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns over 1 year (DOE schedule), p strike against the prototype's 0.6748 within 4 standard errors of two independent runs plus half its last digit | 0.6748 | 0.68195 | 0.018679 | pass |
| Reference (the prototype’s inputs of 17 Oct 2026): Returns over 1 year (DOE schedule), the median week of the low is the prototype's | 2027-01-22 | 2027-01-22 | 0 | pass |
| Histogram: every path's low falls in a band (23 bands of 5 mb, 16 settings; paths outside) | 320000 | 320000 | 0 | pass |
| Range kinds: the fan and the low's spread are simulated ranges with their quantiles (problems) | 0 | 0 | 0 | pass |
| Grammar: every tile carries a comparison and its date, and the headline keeps the house grammar; the takeaways are claims (faults) | 0 | 0 | 0 | pass |
| Decisions: every desk decision this page's specification names is in the decisions file (D-2026-10-24-TH1, D-2026-11-05-OB7; ids missing) | 0 | 0 | 0 | pass |
- U.S. Energy Information Administration: weekly Strategic Petroleum Reserve stocks, monthly Strategic Petroleum Reserve stocks since October 1977, and the Weekly Petroleum Status Report's Table 1 — Source: U.S. Energy Information Administration
- Polymarket, Gamma API: the “US-Iran ceasefire continues through …?” contracts and the “… through the midterms?” contract, quoted on 16 Nov 2026 at 21:57 UTC (mid of bid and ask) — Source: Polymarket
- Figures quoted from the U.S. Department of Energy's notices, the Atlantic Council (2 Oct 2026), GAO (Dec 2025), Reuters and Oil & Gas Journal, each with its document and date in the inputs table — Quoted as published; each input names its document and date
- The simulation: 20,000 seeded paths with common random numbers over 32 settings, computed here — Model-conditional distributions computed by the site's engine; not forecasts.