LNG vs XLE forward 20-day relative returns after OCF expansion and weak Brent
The tidy version of this trade is almost too appealing: LNG’s operating cash flow just ticked up quarter-over-quarter, Brent is in a soft patch, and a contracted exporter with visible cash-flow durability should therefore attract bids against the rest of the energy complex. That defensive story doesn’t survive contact with the data.
Across 594 evaluable trading days, the 138 days that actually met the dual condition produced a mean next-20-day LNG-minus-XLE return of roughly +0.24% — slightly worse than the +0.86% average on the other 456 days. The median signal day was a net loser against XLE, and LNG only won 46% of the time. On top of that, those daily signals are built on just five distinct quarterly cash-flow episodes.
The full statistical breakdown, including the HAC test and the event-by-event detail, is in the analysis below.
Over the past ~3 years, when LNG's quarterly operating cash flow rises quarter-over-quarter while Brent crude's trailing 20-day return is negative, does LNG outperform XLE over the next 20 trading days? I expect cash-flow expansion at a contracted LNG exporter during a soft crude tape to signal cash-flow durability that is not crude beta, so LNG gets bid defensively against the energy sector.
How this was measured
Daily closes for LNG and XLE were aligned on a common trading calendar. For each trading day, the most recently released LNG quarterly operating-cash-flow figure was selected using actual earnings release dates where available and a 60-day-lag fallback. The condition required that OCF to be higher than the preceding reported quarter and Brent crude's trailing 20-trading-day return to be negative. The outcome is the next-20-trading-day return of LNG minus XLE, close-to-close. The event sample was compared with the full evaluable non-event sample, and the mean gap was tested with HAC standard errors to account for overlapping 20-day windows.
The key numbers
Reading the numbers
Across 594 evaluable days, the signal fired 138 times: the average 20-day LNG-XLE edge was +0.24%, versus +0.86% on other days, and the difference is not statistically clear (p=0.7113). Plainly, the expected durability edge doesn't show up.
The charts
The two groups of forward LNG-minus-XLE returns overlap heavily. Signal days average +0.24% and span from -12.53% to +17.50%, while non-signal days average +0.86% with a range from -13.45% to +20.75%. A signal day can still produce a large loss or a large gain relative to XLE, so this chart shows why the 138 signal days don't separate themselves from the other 456.
This breaks the spread into its two pieces. On signal days LNG's own average forward return is +1.72%, below the +2.35% average on non-signal days, while XLE is basically unchanged at +1.48% versus +1.49%. So the signal's lower relative return comes from LNG lagging its baseline, not from LNG acting unusually defensive when crude is weak.
The scatter covers all 594 evaluable days with Brent's trailing 20-day return on the horizontal axis and LNG's forward 20-day return relative to XLE on the vertical axis. Brent ranges from -32.61% to +70.32%, while relative returns run from -13.45% to +20.75%, and the vertical spread is wide throughout. The broad cloud means a negative crude tape does not by itself line up with a reliable LNG-XLE edge.
Forward 20d relative return summary
| State | N | Mean | Median | Std | Win rate |
|---|---|---|---|---|---|
| Signal | 138 | 0.0024 | -0.012 | 0.072 | 0.4638 |
| Non-signal | 456 | 0.0086 | 0.0034 | 0.067 | 0.5197 |
Quarterly signal episodes
| Fiscal quarter end | First signal date | Brent 20d return | OCF current ($M) | OCF prior ($M) | LNG-XLE fwd 20d |
|---|---|---|---|---|---|
| 2023-12-31 | 2024-02-27 | -0.0023 | 1,720 | 1,698 | -0.0389 |
| 2024-09-30 | 2024-10-31 | -0.0557 | 1,391 | 1,116 | 0.0945 |
| 2024-12-31 | 2025-02-20 | -0.0381 | 1,641 | 1,391 | 0.0158 |
| 2025-09-30 | 2025-10-30 | -0.005 | 1,425 | 831 | -0.0485 |
| 2025-12-31 | 2026-04-16 | -0.0124 | 2,055 | 1,425 | -0.1061 |
The takeaway
No — the data don't support the defensive cash-flow story. Across 594 tradable days, the 138 signal days (OCF up quarter-over-quarter and Brent's trailing 20-day return negative) produced a next-20-day average LNG-minus-XLE return of roughly +0.24%, and that was actually below the +0.86% average on the other 456 days, a -0.63% gap in the wrong direction. The median was worse, about -1.2%, and LNG only beat XLE 46.4% of the time, so a couple of strong quarters are carrying the mean. The test says this is noise: HAC t-stat -0.37, p-value 0.71, meaning about a 71% chance this gap is just random variation. With only five distinct quarterly cash-flow episodes behind 138 daily signals, the independent evidence is even thinner than it first appears. Takeaway: this isn't a tradable edge we can hang a defensive-tilt thesis on — at best it's a coin flip, and the point estimate leans slightly against the hypothesis.
The fine print
- Only 5 unique quarterly OCF-up episodes behind the 138 signal days, and forward 20-day windows overlap; effective independent sample is tiny.
- OCF dates use reported earnings dates where available and a 60-day lag fallback, so the public release date can be off by a day or two.
- The spread is raw long LNG / short XLE, not beta- or dollar-neutral, so residual energy exposure may still drive part of the result.
- Same-day close execution is assumed, and Brent's 20-day return is measured through the signal-day close.