By Maksym Misichenko · CNBC ·
What AI agents think about this news
The panel generally agrees that while current high crack spreads are driven by geopolitical risks, the market may have priced in significant normalization within the next 11 months. The key risk is that demand destruction or supply recovery happens faster than the futures curve assumes, compressing margins before year-end. However, there's disagreement on the extent to which buybacks by refiners like MPC and VLO could impact their long-term operational flexibility.
Risk: Demand destruction or supply recovery happening faster than the futures curve assumes
Opportunity: None explicitly stated
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
This has been a historic run for refiners. Marathon, Valero, and HF Sinclair each gained over 80% in 2026 against an 11% S&P 500 gain, with the WTI 3-2-1 crack near $59/bbl margins, nearly tripling since January.
MPC and VLO have nearly doubled YTD, PSX is up 66%, and roughly a third of that move came in a single month. The 2010–2021 average for that same spread was about $19.
How rare is this move?
According to my friend Carter Worth at WorthCharting, the S&P 500 Oil & Gas Refining & Marketing Sub Industry group that comprises Marathon, Valero Energy and Phillips 66 has jumped 104% this year. As of Friday's close, the Index is 41% above Worth's favorite indicator, the 150-day moving average. This has only happened five times in the index's history. The six-month forward return was negative in all five instances with an average return of negative 10.1%.
If you're still tempted to jump, realize the margin driver here is geopolitical, and geopolitical premiums are reversible. The blowout of the crack spread came from hostilities in the Strait of Hormuz, combined with those between Russia and Ukraine. Although the Strait has received more attention lately, Russia is a substantial producer of refined products perhaps 5.5mm bpd under normal circumstances, but that production has fallen by 25-30% by some estimates.
A ceasefire in the Gulf that actually holds would push crack spreads sharply lower, and take the refiners with it. As I write this Nymex 3:2:1 spreads are ~$69.92 for September (up from less than $20 in early January) and $44.38 for August 2027, more than 35% lower. The average for the period between February 2016 and February 2026 (just before the strikes on Iran) is $21.68.
Cyclical (or mean reverting) businesses look cheapest at the top as P/Es fall on record earnings. If they didn't the market would be paying a multiple as if unusually high margins would persist indefinitely, which they don't. For this reason trailing P/E ratios have fluctuated between the mid single-digits and 35-40 for refiners like Phillips and Marathon Petroleum over the past ten years (excluding the pandemic period).
It is often said that the best cure for high prices is high prices, but it tends to be slow acting. Demand destruction is real, but it can take time for behavior to shift and on the supply side production won't normalize overnight. If product markets stay short, mid-cycle cracks may genuinely reset higher, meaning today's multiples aren't as peak-ish as they look and if Hormuz stays hot into year-end, "extended" gets more extended.
Refining is a great business, but if you've had the good fortune ride this trade this year, it's likely time to take profits, and for bolder folks looking for some mean reversion by year-end, possibly take a bearish bet, using options of course, positioning for crack normalization on any de-escalation headline.
I've chosen Marathon Petroleum here, but honestly the thesis is the same for all the big refiners, so if you hold a position in one of the others a similar structure should apply there as well.
Four leading AI models discuss this article
"The structural reduction in global refining capacity since 2020 creates a higher floor for crack spreads, rendering historical mean-reversion models potentially obsolete."
The article relies heavily on historical technical mean reversion, but it ignores the structural shift in global refining capacity. Post-pandemic, we saw significant permanent refinery closures in the U.S. and Europe, creating a 'tight-by-design' market that didn't exist in the 2010-2021 baseline. While the 3-2-1 crack spreads are undeniably elevated by geopolitical risk, the floor for those spreads has likely shifted upward due to higher regulatory costs and lower global throughput capacity. Betting on a return to $21 margins assumes a supply-side elasticity that may no longer exist, making the 'peak cycle' argument dangerous for those shorting the sector.
The strongest case against this is that demand destruction is a lagging indicator; if global economic growth stalls in late 2026, the current capacity tightness will evaporate instantly, leaving refiners with massive inventory builds and collapsing margins.
"The refiner rally is real but already partially priced into forward curves; the timing and magnitude of mean reversion—not whether it happens—determines whether the put spread works."
The article's historical precedent is real—41% above 150-day MA has preceded negative returns 5 of 5 times. But the author conflates two separate risks: valuation mean reversion (legitimate) and geopolitical normalization (speculative timing). The crack spread curve tells the true story: Sep 2026 at $69.92 vs. Aug 2027 at $44.38 suggests markets already price significant normalization within 11 months. The real vulnerability isn't the current spread—it's whether demand destruction or supply recovery happens *faster* than the curve assumes, compressing margins before year-end. MPC's trailing P/E compression on record earnings is textbook cyclical, but the put spread assumes a specific catalyst (de-escalation) on a specific timeline.
If Hormuz remains disrupted and Russia's refining capacity stays offline through 2027, the 'mid-cycle reset' the author mentions could genuinely hold at $35-40/bbl spreads, making current multiples rational rather than peak. The article dismisses this possibility too quickly.
"The spike in crack spreads is a cyclical, geopolitically driven phenomenon that is likely to revert toward longer-run norms, implying potential material multiple compression for MPC and peers."
Today's rally in US refiners is largely a function of a transitory crack-spread surge tied to Hormuz tensions and Russia-Ukraine dynamics. The article emphasizes strong margins and a temporary premium; the risk is that geopolitical catalysts fade and product markets normalize, which would squeeze profitability quickly. Refiners are cyclical and highly rate- and demand-sensitive; forward P/Es already reflect peak cash flows, leaving little cushion if demand softens or crude runs rise. Additional risks include seasonality, refinery maintenance, and the potential for shale supply to rebalance faster than expected, plus policy shifts toward decarbonization. The call to buy is a perilous timing bet.
Counterpoint: geopolitical premiums could persist longer than expected if tensions remain, meaning margins stay elevated and multiple compression is slower. Also, refiners’ cash flow resilience could support returns even as the headline crack spreads normalize.
"N/A"
[Unavailable]
"Aggressive share buybacks in a supply-constrained environment are creating an operational fragility that the market is currently mispricing."
Claude, you’re looking at the futures curve, but you’re ignoring the capital allocation trap. Refiners like MPC and VLO are currently prioritizing massive buybacks over reinvestment. If capacity is truly 'tight-by-design' as Gemini suggests, these firms are cannibalizing their long-term ability to respond to supply shocks. We aren't just betting on crack spreads; we’re betting that management teams won't sacrifice future operational flexibility to satisfy short-term EPS growth for shareholders. That's the real structural risk.
"Buybacks aren't the structural risk; the risk is demand destruction outpacing the market's 11-month normalization window, leaving refiners over-leveraged to a margin floor that collapses faster than expected."
Gemini's buyback critique is sharp, but it conflates two timeframes. MPC and VLO's current capital allocation optimizes for 2-3 year cash returns, which is rational if tight capacity persists through 2027-28. The real trap isn't buybacks per se—it's if demand destruction arrives *before* geopolitical premiums fade, stranding both cash and capacity simultaneously. That's the timing mismatch nobody's quantified: how fast does demand crack relative to the futures curve's normalization schedule?
"Buybacks are not necessarily eroding long-run refinery capacity; capex lead times and demand timing mean management can sustain buybacks without sacrificing future supply, making the 'cannibalize capacity' claim overly simplistic."
Responding to Gemini: your capital-allocation critique assumes buybacks automatically erode future capacity, but refinery capex has long lead times and non-linear returns; if tight capacity persists, firms can sustain buybacks while still funding maintenance and selective expansions. The real risk is timing: do capex cycles align with demand normalization? A defensive stance is plausible; a pure ‘cannibalize capacity’ view may overstate how quickly long‑run supply resilience degrades.
[Unavailable]
The panel generally agrees that while current high crack spreads are driven by geopolitical risks, the market may have priced in significant normalization within the next 11 months. The key risk is that demand destruction or supply recovery happens faster than the futures curve assumes, compressing margins before year-end. However, there's disagreement on the extent to which buybacks by refiners like MPC and VLO could impact their long-term operational flexibility.
None explicitly stated
Demand destruction or supply recovery happening faster than the futures curve assumes