KMI vs XLE after Iran/Hormuz/tanker/sanctions headline-intensity spikes
When Iran, Hormuz, tanker and sanctions headlines cluster, the intuition is that midstream's contracted cash flows start looking dull next to oil-beta energy — and KMI should trail XLE for a few sessions.
Three years of data don't back that up. Across 17 spike sessions, KMI actually outran XLE over the following five days by an average of roughly 0.51%, against 0.23% on ordinary days — the opposite sign. The median, -0.06%, is essentially dead flat, which says the positive average comes from a couple of favorable draws rather than anything repeatable. A 53% event split where KMI lagged is what noise looks like, and with p = 0.75 from just 17 observations, none of the one-to-five-day horizons separate from chance.
The full evidence, horizon by horizon, is below.
Over the past ~3 years, when news headline intensity for Iran/Hormuz/tanker/sanctions keywords spikes into the top quintile of its prior-20-session distribution, does KMI underperform XLE over the next 5 trading sessions? I expect a Hormuz risk-premium spike to trigger rotation from midstream's contracted cash flows into oil-beta energy equities, so KMI lags XLE until the supply-shock premium fades.
How this was measured
Daily KMI and XLE closes were resampled from minute bars. A headline-intensity series was built from the union of KMI_news and XLE_news by flagging titles and summaries containing Iran, Hormuz, tanker, or sanctions, then assigning each matching article to the first trading session on or after its publication date. A spike was defined as a session whose keyword count exceeded the 80th percentile of its prior 20 sessions, with the current session excluded. Forward returns were measured close-to-close at horizons of one through five sessions; the key outcome is the cumulative log differential KMI minus XLE, so negative values mean KMI lagged XLE.
The key numbers
Reading the numbers
Across 752 sessions, only 17 were keyword spikes. After those spikes KMI actually beat XLE by about 0.51% on average over five days, versus +0.23% in non-event periods, but the gap is not statistically reliable (p=0.75); median event return was -0.06%.
The charts
This chart tracks the average gap between KMI and XLE after a keyword spike. The spike-event line starts near zero and climbs to about +0.51% by session five, while the non-event baseline ends around +0.23%. The right edge is the key: instead of KMI falling behind XLE, the spike line sits above the baseline and the gap widens over the five days, the opposite of the expected rotation into oil-beta XLE.
This splits the same post-spike window into KMI's and XLE's own cumulative returns. KMI starts slightly negative at -0.18% but recovers to +0.41% by session five, while XLE stays weak and ends at -0.09%, with a dip as low as -0.48%. The main detail is the widening gap between the two lines: KMI is the one outperforming after these headline spikes, not lagging XLE as the Hormuz risk-premium rotation story would predict.
This histogram shows the spread of outcomes across the 17 spike events, from -4.37% to +7.01% for the five-session KMI-minus-XLE return. The mean is +0.51%, but the median is only -0.06%, and 52.9% of spike events had KMI underperforming XLE. That tells you the positive average is driven by a right-tail outlier or a few big KMI wins, not by a consistent KMI lag or edge; most events cluster near flat or slightly negative.
Cumulative KMI−XLE differential by horizon
| Horizon (sessions) | Event N | Event mean diff | Baseline mean diff | t-stat | p-value |
|---|---|---|---|---|---|
| 1 | 17 | 0.0002 | 0.0004 | -0.055 | 0.9569 |
| 2 | 17 | 0.0022 | 0.0008 | 0.335 | 0.7418 |
| 3 | 17 | 0.0053 | 0.0012 | 0.602 | 0.5552 |
| 4 | 17 | 0.0004 | 0.0019 | -0.194 | 0.8485 |
| 5 | 17 | 0.0051 | 0.0023 | 0.329 | 0.7462 |
The takeaway
No — the rotation you were looking for didn't happen. Across 17 headline-spike sessions over the past three years, KMI actually did slightly better than XLE over the following five sessions, averaging +0.51% relative to +0.23% on ordinary days — the opposite sign of the hypothesis. The median flips that read: -0.06%, essentially dead flat, so the positive average is coming from a couple of favorable draws rather than a repeatable pattern. The event split was 53% of spikes where KMI lagged versus 45% on non-spikes, which is the kind of gap noise produces. With p = 0.75 and only 17 events, this is a coin flip, not a signal — and every horizon from one to five days tells the same story (p-values between 0.55 and 0.96). Practical read: a Hormuz/Iran headline spike has not been a tradeable trigger to rotate out of midstream into oil-beta energy, so don't position around it.
The fine print
- Only 17 spike sessions — far too few to separate a real edge from luck, and one or two big moves can swing the average.
- Some 'spikes' aren't crises: with sparse keyword counts, the top-quintile cutoff can be a single matching article.
- Headlines come only from KMI and XLE news feeds, so company-specific KMI stories can register as macro Hormuz risk.
- Publication timing within a session isn't modeled, so after-hours articles may be treated as same-day signals.