One freeze made six percent of the year: underwrite storage revenue as event risk, not seasonality

Storage revenue looks lumpy because it is lumpy. On a real ERCOT year, the shape of the whole year was set by a handful of days — and mistaking that concentration for seasonality is one of the most common ways a storage underwriting model goes wrong.

The year hides in a few days

At the best hub, January was the largest month — but not because winter is structurally rich. Its median day was unremarkable. What made January was a single freeze event: one day produced 37% of that month and 6% of the entire year. Across the year, something like a fifth to a quarter of P&L came from the ten best days. That is event risk, not a season you can schedule around, and one year of data cannot tell you when the next freeze lands.

The widest spread was not the best asset

It is tempting to site and underwrite on spread width. On the same year, the hub with the largest basis of any point had the worst experience of it — the lowest hit rate and a maximum drawdown many times larger than at a calmer hub. Wind-driven volatility is hard to forecast, so it arrives as drawdown rather than revenue. Spread width alone is not a siting thesis.

Duration and degradation are constraints, not footnotes

More duration helps, but sub-linearly. Hold power fixed and step a unit from two to eight hours and revenue rises — yet the fifth and sixth hours of an evening peak are worth materially less than the first, which is the trade-off behind every sizing decision. And the cycles that earn those event days are finite: across a real year the schedule ran on the order of one full cycle a day, comfortably inside a typical four-hour warranty envelope. The schedule has to respect the warranty rather than chase every marginal day, or the revenue in year one is borrowed from the asset’s life.

Underwrite a storage asset on its median day and you’ll be wrong in both directions — too low on the freeze, too high on the quiet summer. The revenue is concentrated, the widest spread can be the worst asset, and duration pays less at the margin than the brochure implies. Put the event risk, the drawdown, and the cycle budget into the underwriting up front — not into a surprised memo in year two.