BRENT $100.71 ▲3.6%  GAS $4.2245   DIESEL $5.94 ▲ RECORD  SPR 286.6M  DISTILLATES 104.2M   CRISIS DAY 193

The Depletion Ledger

Strait of Hormuz, day . Brent is back above $100, US diesel has set an all-time record, and the Strategic Petroleum Reserve has fallen from 415.4 million barrels to 286.6 million in six months. The depletion model's question: at the current draw rate, how much buffer is left before the corridor breaks — and what does each branch cost?

Crisis day · model day · data as of 2026-09-09 · scenario model, not a forecast

40030020010030025018070415.4 pre-war286.6 now
SPR, million bbl (EIA w/e Aug 28)−128.8M since Feb 27 · lowest since Dec 1982red lines: 300 breached · 250 GEF min · 180 · 70

Brent

$100.71

Sep 9 settle · first >$100 since Jul 24 · +32% vs pre-crisis ~$76

US diesel (AAA)

$5.94

Sep 9 · all-time record (Jun 2022 record broken Sep 4) · +60% vs pre-war $3.72

US gasoline (AAA)

$4.22

Sep 9 · highest Labor Day ever · +50% vs Jan $2.81

SPR

286.6M

w/e Aug 28 · −128.8M since pre-war 415.4M · lowest since Dec 1982

US distillates

104.2M

w/e Aug 28 · −14% vs 5-yr avg · PADD1 −27% YoY — the binding constraint

Likely outcomes

10 / 50 / 40

corridor holds / standoff drift / corridor lapses (Sep 9 reweight)

Brent: $126 peak in March, back above $100 on Sep 9

monthly avgs — Mar $103.0 (EIA) · Apr $117.29 (STEO) · Jul $83.76 (STEO)

Observed points only — hollow markers are a source's rounded estimate, not a measured close. The path: $76 before the closure → $126 intraday peak (March, +65% — the largest monthly rise in oil-market history) → $95 by mid-April on ceasefire hopes → $74 on Jun 30 → back through $100 on Sep 9 as the tanker war took hold.

Gasoline up 50% since January — May peak $4.50

EIA weekly, national

From $2.81 in January to $4.22 now — +50%, with the May peak at $4.50. Every month of 2026 since March has averaged above pre-crisis levels.

Diesel at an all-time record: $5.94, up 68% since January

AAA national, sourced points

From $3.52 in January to an all-time record $5.94 on Sep 9 — +68%. The old record (June 2022, ≈$5.85) fell on Sep 4. Diesel leads gasoline: product, not crude, is the scarce thing.

SPR down 129M barrels since the war began

EIA weekly ending stocks, million bbl

The red lines are floors, not forecasts: ~300M is the cavern-damage threshold — below it the caverns cannot safely be refilled after a draw (breached Aug 14). 250M is the GEF operational minimum for a sustained draw, 180M the hard operable limit, and 70M the DOE's stated safe minimum — the draw stops there, it does not continue. The dashed paths project three draw rates from the Aug 28 level out to ~March 2027: 0.45M b/d (corridor holds, ~205M by then), 0.70M b/d (standoff, reaches 250M in late October and ~160M by March), 1.35M b/d (corridor lapses, through 250M within a month and pinned at the 70M floor from ~Feb 2027).

Supply snapshot

SPR

286.6M bbl

−128.8M (−31%) since pre-war 415.4M

lowest since Dec 1982

US distillates

104.2M bbl

−14% vs 5-yr avg (121.2M)

PADD1 −27% YoY — the binding constraint

US crude

424.5M bbl

+1% vs 5-yr avg (420.3M)

products, not crude, are the scarce thing

Global inventories

−400M bbl YTD

STEO Sep 9 estimate

falling through end of 2026

The release began in earnest Apr 3 (413.3M) and ran fastest in May (≈1.2M b/d). The draw has slowed to ≈0.45M b/d — the corridor-holds pace — while distillates sit 14% below their 5-year average.

Likely outcomes

probability of each track — reweighted only on regime events

Sep 9
10%
50%
40%

regime event: 10 tankers/week + $100 + Jazan

Sep 7
15%
55%
30%

regime event: exclusion zone + first third-country base hit

Sep 2
35%
60%
5%

post-8.6M-claim bust

Jun 30
65%
35%

original model (de-escalation 65%) — approx

corridor holdsstandoff driftcorridor lapses

Corridor holds

10%

Transit resumes under some regime — an Iran–Oman safe corridor or US-escorted lanes — and Hormuz trends back toward normal over one to two quarters.

Brent mean-reverts toward $70–80; the SPR draw slows to ~0.45M b/d and the depletion timeline stretches out again.

Standoff drift

50%

The war continues at current intensity — periodic tanker strikes, the exclusion zone holding, Iranian infrastructure partially offline — but no sustained closure.

Brent ranges $95–125; the SPR draws ~0.70M b/d; global stocks keep falling; the breaking-points cascade is delayed, not cancelled.

Corridor lapses

40%

Sustained closure or escalation — tanker losses spike, the exclusion zone hardens into permanence, Abqaiq and Jazan stay offline for months.

Brent $130+; the SPR draws 1.35M b/d and the inventory cascade begins (PADD1 → Russia → ARA → China → jet fuel).

How these odds are set. They are judgment priors, not a forecast — revised only on regime events: verified structural changes (tanker-loss spikes, exclusion-zone enforcement, Iranian infrastructure offline, diplomatic breaks), never on daily price noise. On Sep 9 the weights moved from 15/55/30 to 10/50/40 on the week's tanker losses (10 ships), the exclusion zone being enforced, the first strike on a third-country base, and the Jazan hit. A data check runs alongside the judgment: branch_filter.py in the model directory is a three-state Bayesian filter — weekly verified transits and tanker losses (Poisson) plus Brent (Normal) feed a transition matrix that outputs horizon weights. On Sep 9 it said 14.9 / 49.8 / 35.3 against judgment's 10 / 50 / 40; v1 has no trend memory, so that gap is logged, not averaged in.

What would move them. A verified corridor re-opening (IMO filing, sustained transits ≥40% of the 85/day baseline for two consecutive weeks) or a durable ceasefire shifts weight toward holds. Tanker losses above ~15/week, a Jazan or Abqaiq restart slipping, the Russia product-ban expiry on Sep 30, a widening EIA–market spread (STEO has 2H26 at ~$90 while spot is over $100 — a gap beyond ~$10–15 means the market is pricing lapse above 40%), or an adverse Polymarket settlement on Sep 14/30 shifts weight toward lapses. Between regime events, the weights hold — a quiet week is not evidence.

Crisis timeline

key events, as verified in the research logs

2026-09-09STEO: 2H26 ~$90 / 2027 $74. Brent breaks $100 ($100.71, high $101.25). Novorossiysk drone strike. 32 Saudi airstrikes on Yemen
2026-09-08Jazan hit (400 kb/d — offline since at least late July); diesel $5.90 record
2026-09-07Regime event: exclusion zone enforced; first strike on a third-country base (Jordan)
2026-09-0410 tankers lost in a week; US retail diesel $5.85 — all-time record
2026-09-02US strikes Iranian tankers for the first time; 80 AIS-dark tankers in 24h
2026-08-31Kpler claims 8.6M bpd through Hormuz; two supertankers struck — the claim does not hold up
2026-08-29Russia: >30% of actual refining capacity knocked out; product export ban extended to Sep 30
2026-08-28WPSR: SPR 286.6M — lowest since Dec 1982; drawing ~0.45M b/d
2026-08-23PortWatch: 3 vessels through Hormuz = 4% of the 85/day baseline
2026-08-21US distillates 103.4M — record seasonal low, ~14% below 5-yr avg
2026-08-19US claims 'controlling the southern lane' (Axios) — the lane is contested, not controlled
2026-08-14SPR breaks below the ~300M cavern-damage floor (293.4M)
2026-07-27Abqaiq halted (~7M bpd of processing)
2026-06-30Day 1 of the depletion model — Brent $74 called a 'false signal'; Hormuz ~5% of normal traffic
2026-05Gulf export shut-in peaks at 11.2M bpd; US diesel peaks $5.60 (May)
2026-04April monthly avg $117.29 (STEO); mid-April ceasefire hopes pull the settle to $95.48 (Apr 20)
2026-03-30Consumer prices turn: gasoline $4.11, diesel $5.62 (AAA) — +38% / +49% in a month
2026-03-18Brent intraday peak $126 — March closes +65%, the largest monthly rise in oil-market history
2026-03-11IEA 172M-bbl coordinated release authorized (~120 days; ends ~early July)
2026-03-08Brent above $100 for the first time in four years
2026-03-04Iran declares the Strait of Hormuz closed
2026-02-28War begins: strikes on Gulf energy infrastructure start — ACLED damage inventory runs from Feb 28 (172+ non-military strikes, 48% energy/power/desal); IEA counts ~1.3B bbl of supply loss from February

Upcoming watch list

dated events the model is watching

Sep 10

WPSR (w/e Sep 5) — distillate sub-100M breach? SPR draw pace after the escalation

the leading indicator

Sep 10

August PPI — consensus 5.4% vs 4.7% prior

macro tripwire

Sep 14/15

Polymarket 'Hormuz normal by Sep 30' settles (3.8%)

expected ~0

Sep 30

Russia diesel export ban expiry (T-21) · Polymarket Sep 30 settlements · Swedish petrol excise cut expires

three dated events in one week

Oct 7

Next STEO — the first one that actually prices the tanker war + Jazan + exclusion zone

the real test of EIA's 'constraints, not closure'

Nov 30

Russian jet-fuel export ban

jet fuel cascade

Breaking-points cascade

§11, compressed twice

PADD1 (US East Coast)

Sep 14–21

distillate days-of-cover collapse; −27% YoY already

Russia

Sep 30

diesel export ban expiry — consequence of >30% capacity loss, not policy choice

ARA (Europe)

Early Oct

gasoil trader floor 8.5–9M bbl; mid-Oct in the lapse branch

China

~Oct 10–17

buffer drawdown accelerates

Europe (visible)

~Oct 25–30

shortages expected mid/late October — price-triggered political breaks first

Jet fuel

Nov 10–30

Russia jet export ban (Nov 30) + 45.7M bbl / ~26 days cover

Research log

local-only files (research/), not published

2026-09-09Full pass: STEO, Russia/Red Sea deep-dive, regime reweight 10/50/40, literature survey, model scripts (draw_rate / elasticity / branch_filter)
2026-09-08Jazan + diesel record (backfilled)
2026-09-07Regime event #1 — exclusion zone, Jordan strike (backfilled)
2026-09-06OPEC+ ministerial, OMR (backfilled)
2026-09-02First US strike on Iranian tankers (backfilled)
2026-08-30Traffic claims vs verified transits (backfilled)
2026-08-29WPSR week + Russia deep-dive (backfilled)
2026-08-1914-day deep-dive sweep, Aug 19 – Sep 2 (backfilled)
2026-06-30Day 1 — model established (backfilled)