Initiated 2026-10-05 · Price $80.78 (as of 2026-10-02, stockanalysis) · Mkt cap $9.64bn · Energy / petroleum refining · Model: verified

Rating: Sell — Conviction: High

PBF is earning a geopolitical windfall: Hormuz and Red Sea disruption pushed the NYH 3-2-1 crack to ~$65 in July–August and ~$73 on 2026-09-22, against a 2025 average of ~$23. The stock is up ~165% in a year and now discounts margins staying well above mid-cycle into 2027 and beyond. We think the windfall is real but temporary, and that a high-cost, reliability-challenged coastal refiner should not be valued as if it were permanent.

Conviction tests (3a-v-c): T1 pass (Sell at both ±1pp) · T2 pass · T3 pass (one est input, TRA liability, moves Base 6.5%) · T4 pass

The rule gives High; the call is still a timing call against a live supply shock, and the price can keep rising while the disruption lasts.

Model value range vs price
Bear $0.00Base $37.37Bull $98.55Price $80.78

Business overview

PBF runs six refineries (Delaware City, Paulsboro, Toledo, Chalmette, Torrance, Martinez), with ~1.0mm bpd of capacity across the East Coast, Mid-continent, Gulf Coast and West Coast. It sells gasoline, distillates and other products into wholesale and export markets, and it has a small logistics segment. GAAP revenue was $29.3bn in FY2025, but what matters for earnings is the gross refining margin: $8.77/bbl in FY2025 and $16.67/bbl in H1 2026, against refinery opex of ~$8.0–8.4/bbl. Three variables drive earnings: regional crack spreads and how much of them PBF captures, throughput (304mm bbl in FY2025 after the February 2025 Martinez fire, vs 330.9mm in FY2024), and opex per barrel, where the RBI program targets $350mm of run-rate savings by year-end 2026.

Competition

PBF competes with Valero, Marathon, Phillips 66 and HF Sinclair, and on the East Coast with European and Asian product imports. The thesis turns on margins, and the cleanest structural datapoint is on the West Coast. Phillips 66's 139k bpd Los Angeles refinery stopped processing crude at year-end 2025 and Valero's 145k bpd Benicia plant was set to close by April 2026. Together that cuts California capacity ~17%, to at most ~1.3mm bpd (Industrial Info, 2025), which supports Martinez and Torrance (Q3 West Coast guide 270–290k bpd). Pressure would show up first in East Coast cracks as import arbitrage reopens. The closest listed comparable, HF Sinclair, trades at 6.2x forward P/E against PBF's 3.35x. The discount reflects PBF's leverage history, outage record and the market's view that its forward earnings are peak earnings.

Bull case

  1. The windfall lasts — the September strikes on Saudi Red Sea infrastructure keep global product markets short, and cracks stay near Q3 levels through 2027. Plays out if Gulf export flows stay impaired past the Q1–Q2 2027 normalization ADNOC has guided to. Model: realized_price
  2. A structurally higher West Coast mid-cycle — California capacity loss makes Martinez and Torrance the marginal suppliers, which lifts through-cycle GRM to ~$14/bbl and justifies a ~6x multiple. Plays out if more California capacity exits and imports stay costly. Model: realized_price, exit_ev_ebitda
  3. Self-help on volume and cost — Martinez restarted in May, the Q3 guide is 900–960k bpd, and RBI pulls opex toward $7.8/bbl. Plays out if PBF runs a clean year without unplanned outages. Model: volume_growth, unit_cash_cost

Bear case

  1. Fast mean reversion — Ras Tanura restarted on 2026-06-26, ADNOC expects full normalization by Q1–Q2 2027, and Western refiners running at near-max utilization add product. GRM falls back toward FY2024–25's $8–9/bbl, plus ~$1 for the West Coast. Plays out if Gulf flows recover on schedule. Model: realized_price
  2. Reliability and cost — PBF has a history of unplanned outages, including the 2025 Martinez fire that cut FY2025 throughput 8%. Opex per barrel rises when volume falls. Plays out if outages recur. Model: volume_growth, unit_cash_cost
  3. No multiple for peak earnings — investors pay ~4x mid-cycle for a high-cost coastal refiner facing secular gasoline decline. Plays out if peers de-rate as cracks fade. Model: exit_ev_ebitda, terminal_growth

Valuation & balance sheet

Metric (definition) Current Own history (range or 5y avg) Peers Source, as-of
Forward P/E (stockanalysis consensus) 3.35x n/a (unverified) VLO 7.68x · MPC 6.40x · DINO 6.22x stockanalysis, 2026-10-02
EV/EBITDA, trailing (lease-inclusive, stockanalysis) 11.63x n/a (unverified) VLO 9.00x · MPC 9.43x · DINO 6.62x stockanalysis, 2026-10-02
EV/EBITDA, model basis (mkt cap − cash + notes + TRA; EBITDA ex special items) 70.6x FY2025 · 2.6x FY2026E base — — model; FY2025 release
GRM ex special items, $/bbl $16.67 (H1 2026) $7.89–$22.00 (FY2022–25, avg $13.7) — FY2023/FY2025/Q2 2026 releases

Model-implied value range (from model-summary.json; energy module, DCF (Gordon) and DCF (exit multiple), midpoints): Bear $0.00 · Base $37.37 · Bull $98.55 per share, i.e. implied returns of −100% / −54% / +22% vs $80.78. These ranges show how the bull and bear drivers translate into value; they are not price targets. The price sits much closer to the bull case: the market is paying for 2027 GRM near $18–20/bbl and a mid-cycle near $14. On re-rating, today's model-basis multiple is meaningless on trough FY2025 EBITDA (70.6x) and only 2.6x on FY2026E. Base's 5.0x exit on Y5 mid-cycle EBITDA is a re-rating from the peak-year multiple, while the Gordon value ($28.51) embeds only ~2.8x terminal EBITDA. The exit method ($46.23) is therefore the higher one, and the gap between the two is a terminal-multiple difference, not noise. Both are well below the price.

Floored bear: both bear methods print $0.00. That is not a forecast of bankruptcy. At FY2024–25-like margins plus $1, EBITDA (~$0.36bn) does not cover ~$0.77bn of sustaining capex and turnarounds, so FCF stays negative after the 2026 windfall is banked. That says refining at trough margins destroys value, and in practice capacity would close. Base is not floored: it keeps ~$1.2bn of mid-cycle EBITDA.

Balance sheet: at the FY0 (2025-12-31) model basis, net leverage was 10.05x and coverage 0.96x on trough EBITDA. Both are stale: by 2026-06-30 PBF had repaid the $801.6mm 2028 notes and the revolver, leaving $1.8bn of notes (9.875% and 7.875% due 2030, 7.25% due 2034) against $894.1mm of cash, so net debt was ~$0.9bn. The nearest material maturity is $1.3bn in 2030. Base reaches net cash in FY2027. Ratings: Moody's Ba3, negative outlook (May 2025, per a secondary source; current rating n/a (unverified)).

Model note: verified (LibreOffice matched all 3,511 formula cells). Unverified input: market.other_claims, the tax receivable agreement liability of $293.6mm, taken at 2024-12-31 because the FY2025 figure wasn't retrieved; removing it lifts Base to $39.79. Assumptions without basis: none. No scenario consistency CHECKs. In this model "revenue" is gross refining margin (company $/bbl × throughput), not GAAP revenue. Fixed costs are a calibrated residual against EBITDA ex special items (FY2022–25 avg ~$90mm). FY0 cash and debt are deliberately taken at 2025-12-31 so that H1 2026 cash generation isn't counted twice, which leaves out H1 Martinez insurance receipts of $356.5mm (~$2.9/sh). There is no company EPS guidance to reconcile Base Y1 EPS ($22.79) against.

Scenario stress test

Reasoned from the bull/bear drivers above. The model column comes from the scenario overlays (Base case + shock).

Scenario Effect Mechanism Magnitude Model Δ value vs Base ($/sh)
S1 Fast equity crash − High-beta cyclical de-rates; cracks unaffected over weeks Low −$2.56
S2 Slow bear / recession − Product demand falls; cracks −30% in 2027–28. Milder than Base − Bear because it's transitory and 2026 is largely banked High −$11.80
S3 Rapid rate shock − Higher discount rate and HY coupons Low −$1.94
S4 Slow rate grind − Sticky inflation lifts opex 3% and the discount rate Med −$5.56
S5 Soft-landing cuts + Steady demand supports cracks; lower discount rate Low +$2.15
S6 Recession-driven cuts − Demand recession compresses cracks 25%; cuts only partly offset High −$8.06
S7a Credit liquidity shock − HY issuer de-rates; no maturity before 2030 Low −$2.13
S7b Slow default cycle − Weaker diesel demand plus wider spreads Med −$3.60
S8 Stagflation + Supply-driven product inflation widens cracks faster than opex (2022 analogue) Med +$6.09
S9a Dollar spike − US product exports less competitive Low −$2.79
S9b Dollar slide + Better export margins Low +$2.79
S10 Melt-up + Beta to a broad rally Low +$1.07
S11 Energy supply shock + Tighter global product markets; coastal US refiners capture the crack (the current regime) High +$13.96
S12 Mega-cap/AI derating 0 no material effect, not modeled Low $0.00

Currently active/on watch per the playbook: S3 partially active; S8, S10, S11 on watch (S11 escalated 2026-09-19/20).

Model value change vs Base, by scenario
S2 Slow bear / recession−$11.80S6 Recession-driven cuts−$8.06S4 Slow rate grind−$5.56S7b Slow default cycle−$3.60S9a Dollar spike−$2.79S1 Fast equity crash−$2.56S7a Credit liquidity shock−$2.13S3 Rapid rate shock−$1.94S12 Mega-cap/AI derating$0.00S10 Melt-up+$1.07S5 Soft-landing cuts+$2.15S9b Dollar slide+$2.79S8 Stagflation+$6.09S11 Energy supply shock+$13.96

What would change the call

Upgrades if: NYH 3-2-1 stays above ~$45 through Q1 2027 with Gulf export flows still impaired, or PBF shows mid-cycle-quality capture (GRM above $14/bbl at cracks near $30), or the stock falls toward the $40s while cracks hold. Downgrades if: n/a; already Sell. The Sell would strengthen on a Q3/Q4 unplanned outage or a fast crack collapse after a Red Sea ceasefire.

Watch items

Sources