VOLTSTACKTRADE THESIS
CLIENT DISTRIBUTION
DELIVERABILITY INDEX92.2%LIVEMAR-26 TROUGH64.4%RECORD52.9% MAR-18TROUGH WINDOW19 MAR-1 APRTTF Q1-27€65.3428 AUG, ELEXYS
TRADE THESIS — 31 AUG 2026

The Deliverability Cliff: Fill Percent Counts Molecules. Winter Runs on Sendout.

Every storage headline this year quotes one number: percent full. But a storage site cannot deliver its nameplate withdrawal rate at low inventory. Capability decays with pressure, fastest below 40-50% fill, and fills bottom in late March. We built the index that measures this, backfilled it across 11.6 years, and the record is unambiguous: every measured winter trough lands between 19 March and 1 April, the one season that touched the realized-peak zone was 2018, and on current reads this winter opens nearer that zone than any since the crisis. The market prices Q1-27 as one flat block of risk. The trade is the shape.

POSITIONING — CONSTRUCTIVE LATE WINTERISSUED: 31 Aug 2026AUTHOR: Voltstack IntelligenceHORIZON: Now → Apr 2027
KEY TAKEAWAYS
  • Nobody publishes the number winter runs on: available withdrawal capability at the current fill. Voltstack computes it live (ENTSOG/GSE Annex A deliverability curves × GIE AGSI+ fills and declared technical maxima, for DE+FR+IT+NL+AT, ~80% of EU technical — not EU-total) and backfilled it to 2015. No comparable public index exists anywhere our research could find.
  • The cliff is a date, not just a level: all eleven measured winter troughs land 19 March-1 April, after the price-weighted cold has usually passed. Comfortable winters trough at 87-89%; the crisis winters at 63-66%; the record is 52.9% on 30 March 2018, when the Beast from the East met draining storage.
  • This winter opens short of every post-crisis position on the late-August reads: the index stood at 91.9% on 30 August, versus 95-98% at every 1 November since 2021 (the live figure runs in the metrics below) — and running last winter's own 55-point fill draw from the 70-78% forecast start lands at 15-23% spring fill, mapping the index to roughly 54-63%: the 2018 zone at the low end.
  • The expression is the intra-winter steepener — long Feb-27 + Mar-27 against short Dec-26 + Jan-27 — plus Q1-27 call spreads with a daily-measured engine under them, and FR/DE late-winter basis for desks with hub marks. A parallel move in the whole Q1 strip nets to roughly zero across the steepener: this trades the calendar of the cliff, not the winter level our sibling theses already own.
  • It invalidates on the instrument: the index holding at or above 80% through February 2027, or the ENTSOG winter-edition curves (due ~October) revising the knee away.
Index · LIVE (gas day 2026-09-04)
92.2%
14.82 of 16.08 TWh/d available — vs 95-98% at every 1 Nov start since 2021
Winter 25/26 Trough
64.4%
10.31 TWh/d on 31 Mar 2026, from a 96.0% start
Record Trough
52.9%
30 Mar 2018 — the Beast-from-the-East season
Trough Timing, 11 Winters
19 Mar-1 Apr
every measured winter trough lands in this window
This Winter, Mapped
54-63%
last winter’s 55-point draw from a 70-78% start (uniform-fill approx.)
DE Curve at 30% Fill
×0.67
a fifth of capability shed between 40% and 30%; a third by 20%
derived · covered DE+FR+IT+NL+AT, not EU-total · per-country curves are unweighted node means · one curve vintage across the historical window
01 — The Record

11.6 Years of a Line Nobody Had Drawn

50%60%70%80%90%100%20152017201920212023202552.9% · 30 Mar 201864.4% · 31 Mar 202630 Aug · 91.9%
Weekly samples of 4,130 daily values, 2015-01-01 to 2026-08-30. Available withdrawal capability as % of declared technical maximum, DE+FR+IT+NL+AT covered total (not EU-total); per-country curves are unweighted node means. Each day scored with that day's own declared capacity; one curve vintage (the current annex edition) applied across the whole window. Two multi-week AGSI backfill gaps (Jan-Mar 2023, Jan-Feb 2024) are drawn broken, not interpolated. Derived · Voltstack computation on sourced inputs.

Three facts jump out. First, the cliff is seasonal and late: every measured winter trough across eleven seasons lands between 19 March and 1 April, when attention has moved on. Second, comfortable winters stay far from it: 2022/23 and 2023/24 troughed at 87.7% and 89.4% on the days on record (both winters carry a multi-week backfill hole across Jan-Feb, so their true minima could sit somewhat lower). Third, the one season that touched the realized-peak zone was 2018: a late-February cold wave met draining storage and availability fell to 52.9% of technical by 30 March. That is the season this winter's starting arithmetic most resembles.

02 — The Counterweight, Taken Seriously

Why Smart Desks Dismiss This, and Where the Dismissal Fails

The standard objection is aggregate and historical: OIES notes that since 2017, EU-wide daily withdrawal has never exceeded roughly 60% of daily withdrawal capacity (Insight 159), and last winter's realized peak was 10.38 TWh/d against a ~20.2 TWh/d EU technical maximum. On that arithmetic, sendout capacity has never been the binding constraint. The objection is correct at the winter's start, and the record shows exactly where it stops being correct.

  • The margin is a level meeting a decaying line. Peak-day need is roughly stable per unit of cold; available capability falls all winter. Last season the covered index fell from 15.66 TWh/d available on 1 November to 10.31 TWh/d by 31 March: into the same band as the whole winter's EU-wide realized peak (10.38 TWh/d, which landed earlier in the season, at higher fill; different basis, stated deliberately). A 2018-class cold wave arriving in March meets the year's minimum capability, not its average.
  • Aggregate arithmetic hides the geography. The knee is per-country: Germany sheds a fifth of its capability between 40% and 30% fill and a third by 20%; aquifer-heavy France decays earliest. A cold snap does not draw on the EU average; it draws on the systems behind the demand, through finite interconnection.
  • The starting point moved. Every winter since 2021 began at 95-98% availability; on 30 August the index stood at 91.9%, two months before the season. Whether 1 November lands above or below the post-crisis starts depends on the autumn injections still to come: a 78% fill start would roughly match 2021's 95%, a 70% start would sit clearly below it.
03 — This Winter's Arithmetic

From a 70-78% Start, the Knee Arrives on Schedule

Last winter is the template: a 96.0%-availability start drew down to a 64.4% trough by 31 March (that percentage is of technical withdrawal capability; by coincidence it matches the sibling note's 30 Aug fill print, a different quantity entirely). The EU fill went from 82.8% on 1 November to 27.9% on 1 April, a draw of 55 fill-points, both endpoints from our own backfilled AGSI record. This winter's central forecasts put the 1 November fill at 70-78%, five to thirteen points behind, and the deferral mechanics our sibling theses document (no forced autumn buyer, thinner commercial buffer, state-held inventory) do not soften the draw. Run last winter's own 55-point draw from this winter's start range: from 78% it ends near 23% fill, from 70% near 15%, the exact fill zone of the March-2018 record. On the current curves those endpoints map the covered index to roughly 63% and 54% of technical respectively (treating the covered set at one uniform fill, an approximation: the live index weights each country at its own fill, and the countries diverge). That is arithmetic on stated assumptions, not a forecast; prices can ration demand harder than last year and lift both numbers. What the arithmetic does not reach, from this starting point, is the comfortable 87-89% of 2023 and 2024.

SCENARIO A — NORMAL WINTER, THE KNEE BITES IN FEB-MAR (Probability: ~45%) ← BASE CASE

Fill crosses 40% in late January, a month earlier than last season; DE and FR cross their knees in February. The index troughs 56-63% in late March. Late-winter months reprice against early-winter months as the physical story becomes legible; prompt volatility rises into February-March.

SCENARIO B — A 2018 RE-RUN: LATE COLD ON A DRAINED SYSTEM (Probability: ~20%)

A blocking pattern lands after mid-February on sub-30% fills. Available capability sets a new record low below 52.9%; peak-day need and available sendout converge for the first time on record; late-winter prompt and within-Q1 spreads gap. This is the scenario the flat Q1 strip gives away for free.

SCENARIO C — MILD WINTER, SHALLOW DRAW (Probability: ~35%) ← INVALIDATION PRESSURE

A warm, windy winter from a 78% start keeps fills above 45% through February; the index holds 80%+ and the steepener bleeds carry to nothing. This branch is fatter here than in the sibling theses on purpose: a mild winter merely compresses their spreads, but it kills this thesis outright. The loss is bounded by construction; size to this branch.

04 — Trade Expression

The Core Trades the Shape; the Convexity Leg Buys the Tail With an Instrument Under It

▲ CORE — THE INTRA-WINTER STEEPENER: LONG FEB-27 + MAR-27 / SHORT DEC-26 + JAN-27

The market prices Q1-27 as one block; the record says the risk lives at the block's end. Long the months where every measured trough landed, short the months the panic premium already inhabits. Level, Hormuz and the winter-total risk largely cancel across the legs; a parallel repricing of the whole Q1 strip nets to roughly zero here, which is what makes this genuinely a shape trade rather than a second copy of the winter level our sibling theses own. Month marks need your own curve (public prints quote quarters); the platform's spread monitors take BYO marks.

△ CONVEXITY — Q1-27 CALL SPREADS, ONE BRANCH UPGRADED

Lineage, stated plainly: the calendar core of our June note, Short the Calm, was restated by The Empty Auction in August on the auction record; its convex overlay (Q1-27 call spreads) is the leg that survived, and this thesis is that leg's instrument. The cold-snap state where those calls pay now has a daily-measured engine; the Hormuz branch the June note also named remains un-instrumented and is not this note's claim. The June structure's funding (the soft summer) is gone, so entering now means paying post-repricing premium at higher strikes: re-underwriting the overlay, not re-running it. A sizing warning: for a desk already long Q1-27 via the Empty Auction or still carrying the June overlay, this leg is a replacement and a monitor, not an addition. Stacking all three is a triple-long on the same March tail.

△ BASIS — FRANCE AND GERMANY, LATE WINTER, BYO MARKS

Aquifers decay before caverns: France's blended curve is the earliest of the five to fall, Germany's knees hardest below 40%. If February-March tightness localizes, PEG and THE late-winter basis against TTF widens before the aggregate story does. No free feed carries those hub curves; this leg exists for desks with their own marks.

RELATION TO THE DESK — THE TRILOGY, AND ONE NAMED NETTING

This closes a trilogy: The Empty Auction prices this winter's level, The Second Bill the 2027 refill the deferral rolls into, and this note the shape and volatility inside the winter both siblings look across. The legs are designed to net rather than to avoid each other, and one netting is deliberate and named: the steepener's short Jan-27 sits inside the Empty Auction's long Q1-27 (and inside Short the Calm's, where carried). Sized 1:1 per leg, the combined book goes net short January and double-long February-March. That is the thesis, stated: Q1's repricing concentrates late. One honest edge: if the Empty Auction's base case plays early, with January inventory scares doing the repricing, the steepener's short leg pays against it for a stretch. A desk that wants the block AND the tilt sizes the steepener below 1:1.

05 — Risk & Invalidation

Written Rules, With a Live Instrument Under Them

INVALIDATION — CUT THE TRADE

The index holding at or above 80% through February 2027: the draw never reached the knee, and the steepener’s premise is gone.

INVALIDATION — RE-DERIVE FIRST

The ENTSOG Winter Supply Outlook 2026/27 (due ~October) revising the deliverability curves materially flatter: the cliff was re-surveyed away. The index is re-scored on the new edition either way.

CONFIRMING — UNPRECEDENTED IF SEEN

The index crossing below 75% before end-January. That has never happened in eleven measured winters (closest: 76.1% in January 2026, 78.0% in January 2022).

CONFIRMING — MONITOR WEEKLY

Fills crossing 40% before February; FR/DE basis firming while the TTF front is quiet; the draw pace outrunning last winter’s from a lower start.

The desk treats a rule as armed only when something watches it. The instrument runs live on the Voltstack platform (the index and its 11.6-year history), its alert system can watch these thresholds, and the desk record at app.voltstack.energy/thesis states at any moment whether the rule is armed. The headline on this page reads from the live index.

06 — Voltstack Platform Relevance

The Thesis Is a Feature of the Instrument

This note exists because we built the metric first. The storage deliverability index runs live in the Voltstack workspace (per-country curves charted against today's position, trend view over the full backfilled record, provenance and grades printed on the widget face), and the number at the top of this page is served by the same API.

CapabilityRole in This ThesisGeneric Alternative
Storage deliverability index, live + 11.6y historyThe instrument itself: fill-adjusted available withdrawal, per country and covered total.Does not exist elsewhere
AGSI+ storage corridor, by countryThe fill trajectory that drives the curves' input.Weekly CSV, aggregate only
Curve & spread monitors (BYO marks)The steepener and basis legs, tracked against your own month marks.Excel and screenshots
AlertsThe 80%/75% rule thresholds, watchable on the live index.Reading the chart and hoping

Voltstack — Built for European Energy Trading

Storage deliverability index · AGSI+/ALSI+ live · ENTSOG flows · Curve & spread monitors · REMIT II native

voltstack.energy
RELATED VOLTSTACK INTELLIGENCE
DISCLAIMER: This analysis is produced by Voltstack Intelligence for informational purposes only and does not constitute investment advice, a recommendation to trade, or an offer to buy or sell any financial instrument. All scenario probabilities, index mappings, and trade structures represent the analytical judgement of the author and are subject to material uncertainty. The deliverability index is a derived Voltstack computation on sourced inputs: ENTSOG/GSE Annex A per-node curves (averaged unweighted per country; the annex publishes no node weights; values held flat above the 90% inventory step) applied to GIE AGSI+ fills and declared technical maxima for DE, FR, IT, NL and AT (~80% of EU technical capacity, not EU-total), with one curve vintage applied across the historical window and two multi-week backfill gaps (Jan-Mar 2023, Jan-Feb 2024) disclosed on the chart. The section-03 index mappings treat the covered set at one uniform fill, an approximation. Futures, options and spread trading involves substantial risk of loss. Recipients should conduct their own due diligence and consult qualified advisors before making trading decisions. © 2026 Voltstack Ltd. All rights reserved.

SOURCES: ENTSOG Summer Supply Outlook 2026, Annex A withdrawal deliverability curves (provided by GSE, per-node rows exact to 5 decimal places); GIE AGSI+ (fills and technical maxima, API manual v006 field semantics; 63,502 backfilled daily points 2015-2026); OIES (realized peak withdrawal 10.38 TWh/d winter 2025/26; ~20.2 TWh/d EU technical maximum; the never-above-60%-of-capacity-since-2017 observation, Insight 159, Nov 2024; European Storage Refill in Summer 2026, Jul 2026); ICE Endex settlements via the Elexys public forward table (28 Aug 2026, Q1-27 reference); Voltstack storage deliverability index and history (app.voltstack.energy, the live source of this page's headline).