Delek US Holdings is a mid-continent refiner that owns 63% of DKL ($2.04B at market), operates Big Spring / Krotz Springs / El Dorado, runs ≈250 retail stations, and has an idled biodiesel business in wind-down. The Q1 2026 10-Q (filed April 29) reports a -$190.9M GAAP net loss while operating cash flow swung +$523M YoY and refining EBITDA improved +$95M. Two unrelated facts make the filing worth a second look: a D.C. Circuit ruling the tape didn't price, and a buyback pattern peers don't share.

What the filing says

The headline loss is three non-cash mark-to-market items: -$180.8M RINs fair-value change, -$144M Inventory Intermediation fair-value change (Citigroup facility, crude price spike), -$65.8M commodity derivative loss. Refining EBITDA was +$79.2M vs -$15.8M Q1 2025. Operating cash flow was +$461.1M vs -$62.4M YoY. Crack spreads: 5-3-2 +57%, 3-2-1 +55%, 2-1-1 +85% YoY.

Big Spring throughput averaged 30.5K bpd (vs 59.4K Q1 2025) on a planned turnaround completed "safely, on budget and on-time." Q2 should normalize toward 60K bpd.

DK repurchased zero shares in Q1. The repurchase authorization remains $464.2M; cash on hand is $624M. Q1 2025 buybacks were $31.5M.

DKL closed a new $1.3B revolving credit facility maturing March 2031 (+$400M capacity). DK's ABL was extended to April 2031, capacity raised to $1.25B.

The MD&A says DK intends to "monetize our investment in Delek Logistics... deconsolidating Delek Logistics by reducing Delek's ownership." The language is unchanged from the Q4 2025 10-K — no counterparty named, no timeline, no "board authorized."

Separately, on April 7, 2026, the D.C. Circuit (Cases 25-1187 / 25-1197 consolidated) vacated EPA's denial of Krotz Springs' 2024 Small Refinery Exemption petition. Only DK and DINO were directly named. Pending: Big Spring 2020 SRE petition, refund of expired RINs 2019-2023.

What the market thinks

DK is +273% 1Y, RSI 61.8, short interest 17.3% (5-7x peer). SOTP at the current DKL price is ≈$45/share (DKL stake $33 + cash $10 + refining $16 + retail $5 − net debt $19), against $46.59 — fair AT CURRENT DKL. The trade is the DKL repricing gap closing or a take-out at the CEO's stated $70+ anchor.

The 14-day, 48-day, and 77-day options chains are all priced at ≈56.5% IV. For a known earnings catalyst inside the 14-day window, that's flat term structure where it shouldn't be. P/C 12.00 with put-skew inversion looks like longs hedging, not informed bears (the dominant put OI is at $32, far OTM).

On April 7, DK closed -1.4% and DINO -0.6%. The Sidley press release announcing the win wasn't published until April 13. The cohort sold off April 8-9 on EPA's separate March 27 final RFS rule, which reallocated ≈70% of SRE-exempted volumes back to compliance.

Cross-refiner buybacks Q1 2026: VLO doubled to $564M, PSX stable at $269M, PBF zero (consistent with structural distress), CVI in debt refi, DK is the only refiner with a YoY change from active deployment to zero.

Why the gap exists

The court ruling is buried — court docket on the day, press release six days later, sector cohort selling off the same week on a separate (and real) RFS negative. Most analysts read MD&A and earnings releases, not D.C. Circuit dockets.

The GAAP-vs-cash split requires decomposing three non-cash items to see the underlying business. A "$190M loss" headline doesn't.

The corporate signals — Spiegel's June 30 employment cliff, the July 1 Omnibus fee step-up ($13M → $21M/year), advisors retained, the Israel/Russell EVP transition that dropped "Renewables" from the title, the buyback pause — show up across separate 8-Ks and footnotes. None individually meaningful. Together they describe a capital-conservation posture peers aren't running.

SRE relief isn't in any factor model. It materializes when EPA reprocesses the vacated petition, typically 6-12 months out, and shows up in a 10-Q footnote.

Risks

1. Crude collapses on Hormuz resolution (≈30%, ~-12%). Refining cycle peaks; crack spreads compress; DK gives back recent gains.
2. Soreq stays flat on May 6-8 call (≈35%, ~-5%). Deconsolidation timing pushes to H2; July 1 Omnibus fee becomes the next forcing function rather than a transaction trigger.
3. RINs prices keep climbing sector-wide (≈30%, ~-5%). The $461M deficit grows; DK already exposed.
4. EPA seeks en banc review (≈25%, ~-3%). Krotz Springs relief delays into 2027.
5. Omnibus fee absorbed without transaction (≈40%, ~-3%). $12M/year permanent drag begins July 1.

Catalysts

• May 6-8 — Q1 earnings call. Tests whether deconsolidation language escalates to named counterparty or timeline (45%).
• June 30 — Spiegel employment cliff; structural-transaction deadline (22%).
• July 1 — Omnibus fee step-up to $21M/year.
• August ≈15 — Q2 10-Q. Big Spring throughput recovery (85%); continued zero buybacks (55%).
• Q3-Q4 — EPA reprocesses Krotz Springs SRE petition.
• December 31 — Big Spring 2020 SRE deadline (55%); DKL ownership ≥10pp reduction deadline (50%).

What would change our mind

Bull confirmation: Soreq names a counterparty or retains a named advisor on May 6-8. Form 4 P-code insider buying pre-call. EPA voluntary remand of Krotz Springs without appeal. DKL announces a large third-party customer contract (validates the standalone-vehicle thesis).

Bear confirmation: Soreq stays flat for a fourth consecutive quarter. DK announces buyback resumption (invalidates the capital-hoarding read). Crude collapse on Hormuz resolution. EPA seeks en banc review with merits-stage briefing on Krotz Springs.

Evidence

Memo LR signal: 1.4 — bullish; market underweighting the convergent positioning into the May 6-8 catalyst, with the unpriced regulatory leg as the most distinctive item. Resists higher because the dominant catalyst (deconsolidation) has 35% probability of staying flat on the call, and SRE relief magnitude flowing to actual cash payment depends on EPA reprocessing pace.