# The hedge is the one step you place yourself

> Three operations do nothing on purpose. They are instructions for a person, carried in the plan so a hedge sits in the right order beside what it protects.

- Source: https://defiloops.com/blog/the-hedge-is-the-one-step-you-place-yourself
- Published: 2026-09-16
- Category: Architecture
- Tags: hedging, perpetuals, options, delta-neutral
- Author: DeFiLoops

---
Three of the operations here execute nothing. They reach no venue, sign no transaction and
move no money. When a run arrives at one it reports it, steps over it, and carries on.

They are instructions for a human being, written into the machine's plan on purpose.

## Why they exist at all

Options and perpetual futures trade at venues this system does not reach. The honest responses
to that are to say nothing about hedging, or to say it in the wrong place — a note at the end,
a paragraph in a summary, a suggestion after the fact.

Both are worse than they sound, because **a hedge is a thing with a position in a sequence.**
"Buy a put" is not useful. "Buy a put *before* the leveraged position opens" is a different
instruction, and the difference is the entire value of writing it down.

So the step exists where it belongs in the order, and is honest about who performs it.

<Callout type="note" title="What that costs, stated plainly">
  The run does not wait. Nothing later can be sized from it. The system never sees the position
  and cannot tell whether you opened it — so a plan cannot check, cannot retry, and cannot
  react to the result.
</Callout>

## The instruction has to be specific enough to act on

A step that says "hedge this" is a note to self. The parameters force it into something a
person can carry out without thinking: for an option, the asset, put or call, the strike as a
multiple of spot, and how long.

That is the difference between "buy a put" and an instruction you can take to a venue and fill.

The survey half of the system prices it first. What that protection has historically cost is a
question with a real answer, measured across every strike, spend and tenor since 2021 — and
the answer is usually uncomfortable. Insurance at the money runs 30–40% a year against the
asset's own drift. Knowing that before you write the step is the point of being able to ask.

## The perp case, and the thing people get wrong about it

A short perpetual against spot you already hold is the delta-neutral carry. The two legs
cancel, the price stops mattering, and what is left is the funding rate.

It is the most misunderstood trade in this category, because the risk is not where the name
suggests.

<Spec rows={[
  ['Not price', 'That is the part the structure removes. Both legs move together and cancel'],
  ['Liquidation', 'A hedge that gets liquidated stops hedging at the exact moment you needed it. The structure does not protect the leg that failed'],
  ['Funding turning', 'Funding is the entire return. If it flips negative you are paying it rather than earning it, and the trade has quietly inverted'],
]} />

Both of those get worse with leverage, which is the argument for keeping it low in a trade
whose whole appeal is that it removed the price risk.

Whether the rate is worth having, and which venue is paying it, is another survey question —
four venues, back to 2019.

## A hedge needs an ending, and plans forget

The third of the three is closing a perpetual, and it exists because of a specific failure:
**a plan that opens a hedge and never says how it ends.**

That leaves somebody paying funding on a position that is no longer hedging anything — the
thing it protected was unwound three steps later, and the hedge simply stayed open. Nobody
notices, because it costs a little at a time.

So the closing step is there for the cases where the plan already knows the end: funding has
turned, the position being protected is unwound by a later step, or the carry was always meant
to run for a while and stop.

<Callout type="warn" title="It names the position by asset and venue, never by an id">
  An id belongs to the venue's account, and the plan has never seen one. This is the same rule
  that runs through the whole system — a plan cannot name a thing it was never shown — and here
  it means the instruction reads as "close the short ETH perp on this venue", which is what a
  person can act on anyway.
</Callout>

## The design argument

It would be easy to leave these out and let the product look more automated than it is. The
reason not to is that the plan is the artefact people read, approve and keep — and a plan with
a hole where the hedge should be is a plan that describes a strategy nobody is running.

A step that says "this part is yours, and it goes here" is less impressive and more useful
than silence.