Scene summary
- Objective: diagnose a “delta hedge failed” incident without blaming either trader or model prematurely.
- Products: storage + swing hedged with an exchange winter strip; exposures are shape- and policy-dependent.
- Key tension: a single bucket delta is not a hedge plan; mapping, basis, and non-linearities dominate.
Transcript
- Time: later that week, 11:00.
- Location: power desk. Anna’s screens show Winter-26, a stack of intraday fills, and a PnL widget with one red line that will not stop being red.
- Ambient: the desk is loud. Anna is not. She is controlled-annoyed, which is worse.
Anna calls Lukas over with one finger, eyes still on her blotter.
Anna: “I hedged according to your deltas on Winter-26 gas. Next day, market moved, and PnL didn’t match. We were supposed to be flat. What went wrong?”
Lukas pulls a chair in. He does not defend the report. He asks for the hedge ticket.
Lukas: “Let’s reconstruct it. What hedge did you put on? Exact instrument, size, timestamp.”
Anna swivels her monitor toward him.
Anna: “I sold 50 MW Winter-26 baseload futures to neutralise the delta from a storage plus swing combination. Your report said ‘delta = +50’ on that bucket. Market went down 1 €/MWh. I expected near-zero PnL. We lost more.”
Lukas nods slowly. He opens the trade-level risk screen and the hedge blotter side by side.
Lukas: “Okay. We need three checks before we talk about wrong:
- bucket definition,
- instrument mapping and basis,
- non-linearities and policy effects.”
Anna folds her arms.
Anna: “Fine. But I’m telling you: I did what your number told me.”
Lukas: “Understood. Let’s see if the number matched the instrument you traded.”
1) Bucket definition: what does Winter-26 delta actually mean?
Lukas clicks into the risk factor mapping.
Lukas: “Your report’s Winter-26 delta is the sensitivity to our quoted Winter-26 forward bucket. That bucket is defined as a weighted blend of delivery months—typically Dec/Jan/Feb, sometimes with March depending on convention.”
Anna interrupts.
Anna: “I hedge the exchange winter strip. That’s Dec–Mar on that venue, not your internal bucket.”
Lukas does not argue. He highlights the definition field.
Lukas: “Exactly. Here: our Winter bucket is Dec–Feb for this index. Your future is Dec–Mar. You hedged a slightly different object. That alone creates slippage.”
Anna’s eyes narrow.
Anna: “How big could that be on a one-euro move?”
Lukas keeps it concrete.
Lukas: “If the move was a clean parallel shift, small. If the move was shape—March moved less than January—then it matters immediately.”
He pulls up the curve change chart: Dec, Jan, Feb, Mar.
Lukas: “Look. The day wasn’t parallel. Jan and Feb dropped more than Mar. So your hedge instrument didn’t cancel the move in the months your storage and swing cares about most.”
Anna looks at the monthly deltas section.
Anna: “And your report didn’t show monthly deltas. It gave me one winter number.”
Lukas nods once.
Lukas: “Yes. That is on us.”
2) Basis and mapping: same name is not same exposure
Anna points at the hedge ticket.
Anna: “Still, the winter strip is the liquid hedge. That’s why we use it.”
Lukas agrees, then adds the uncomfortable detail.
Lukas: “And it’s a good hedge for level. But your storage and swing are priced off an OTC index and our internal forward curve build. Your hedge is an exchange future. Even when they track, there is basis and timing noise.”
He opens a small panel: “Index mapping”.
Lukas: “Here’s the mapping chain: Trade indexed to: OTC monthly index → curated forward curve build → internal Winter bucket. Your hedge: exchange winter future with its own settlement rules and liquidity dynamics.”
Anna: “So I’m carrying basis risk.”
Lukas: “Yes. Not huge most days. But it shows up exactly when liquidity and shape move.”
Anna exhales through her nose.
Anna: “And this is why people hate quants. You give me a clean delta, but the hedge instrument is a slightly different universe.”
Lukas stays neutral.
Lukas: “Then the report should say that. A delta without a hedge mapping is incomplete.”
3) Non-linearities: delta is not a contract
Lukas scrolls down to the structured trades.
Lukas: “Now the bigger issue: storage and swing. Their delta is not like a vanilla delta. It is policy-dependent.”
Anna looks at him.
Anna: “Meaning what in practice?”
Lukas draws three quick arrows on a piece of paper.
Lukas: “Meaning:
- if the curve moves, the optimal exercise policy can change,
- if vol moves, option value shifts even at unchanged forwards,
- if you’re near constraints, small changes can flip decisions.”
Anna points at the one-euro move.
Anna: “But the market move was just one euro.”
Lukas nods.
Lukas: “And vol moved as well. And the intra-winter shape moved. And your storage policy boundary for hold versus withdraw sits right around that region.”
Anna’s face hardens.
Anna: “So my delta hedge neutralised only the flat-forward move, not shape changes or convexity.”
Lukas: “Exactly.”
He taps the report.
Lukas: “We gave you a first-order sensitivity to a single bucket. But your PnL came from:
- month-to-month differentials inside winter,
- gamma and vega effects,
- basis between hedge and pricing curve,
- and potentially a policy regime shift.”
4) The escalation: what Anna needs from the tools
Anna sits forward.
Anna: “So we fix the tooling. I want the report to stop pretending a winter bucket is sufficient.”
Lukas nods.
Anna: “We should show:
- delta decomposed by month inside winter,
- a non-linear risk indicator—gamma/vega, or at least boundary proximity,
- explicit basis exposure between the pricing index and the hedge instruments.”
Lukas writes as she speaks.
Lukas: “Agreed. And from my side I’ll add two more:
- scenario shape steepening and shape twist, not just parallel shift,
- a hedge suggestion map: if you hedge with winter strip, expect residual if month shape moves; consider overlay hedges in Jan/Feb or spreads.”
Anna nods.
Anna: “Good. Because the hedge strategy is: delta hedge for level moves, overlay for shape, and optionality hedges when vol shifts.”
Lukas adds one operational point.
Lukas: “And we’ll label it clearly: delta near zero means neutral to small parallel shifts under today’s policy. It does not mean zero risk.”
Anna looks at him, satisfied but still annoyed.
Anna: “Just don’t sell me the fiction that delta near zero means flat. Flat is a strategy. Delta is a local derivative.”
Lukas nods.
Lukas: “Exactly. That’s the line we should both use with risk.”
Anna stands up and picks up her phone.
Anna: “One more thing. I want this change before month-end. I don’t want to have this argument again with the same screenshot.”
Lukas closes his laptop.
Lukas: “Understood. I’ll deliver: month-level deltas, basis flags, and shape scenarios. And I’ll add a hedgeable mapping section so the report speaks in the instruments you actually trade.”
Anna pauses, then gives him the smallest possible concession.
Anna: “Good. Then I can argue with the market. Not with the tooling.”
They both return to their screens. The disagreement is not resolved by persuasion. It is resolved by making the analytics match the reality of how the book is hedged.
Key takeaways
- A bucket delta is only meaningful relative to a specific hedge instrument definition.
- Hedging “Winter” with a winter strip embeds shape mismatch and basis risk by construction.
- For flex products, policy-dependent non-linearities can dominate even on small moves.