# What protecting $10,000 of ETH for a year actually costs

> Hedging is asked for more than almost anything. Priced at every strike, spend and tenor, it lost in every single configuration tested.

- Source: https://defiloops.com/blog/what-a-year-of-downside-protection-costs
- Published: 2026-07-07
- Category: Strategy
- Tags: hedging, options, backtesting, risk
- Author: DeFiLoops

---
"Put 10,000 USDC into WETH and protect it against a fall of more than 15% over the next
year."

That is a real request, typed by a real person, and it is one of the most common things
anyone asks for. It is also the request where the honest answer is least welcome.

## The short version

<StatRow>
  <Stat value="30.5%" label="a year, BTC" note="At-the-money protection, 90-day puts rolled four times." />
  <Stat value="40.1%" label="a year, ETH" />
  <Stat value="0" label="configurations that won" accent note="Every strike, every spend, every tenor tested." />
</StatRow>

Insuring a long position at the money costs about a third of the position's value per year
on BTC and about two fifths on ETH. The premium is larger than the asset's own drift, which
means no choice of strike and no size of budget rescues it. There is no clever corner.

<Quote>
  Every strike, every spend, every tenor — buying protection has not worked. Not "worked
  badly". Not worked.
</Quote>

## What was actually tested

This is not one backtest with a discouraging result. It is a sweep, and the sweep is the
point: a single configuration that loses is an anecdote, and a single configuration that
wins is usually a bug.

<Spec title="The sweep" rows={[
  ['Strikes', '0.70, 0.80, 0.90, 0.95 and 1.00 — as moneyness, so the same shape prices on BTC at $90,000 and ETH at $3,000'],
  ['Premium budget', '0.5% to 10% of the sleeve per roll'],
  ['Tenors', '30, 90, 180 and 365 days'],
  ['Position weight', '50% to 70% spot'],
  ['Windows', 'Four, across both assets'],
]} />

Every combination lost. Repricing at the skew a live option chain actually showed did not
change it.

## It barely even protects

This is the part that surprises people who have already accepted the cost. The premium buys
much less protection than it looks like it should.

<Spec rows={[
  ['$8,363 a year of BTC puts', 'Reduced the worst drawdown by 1.2 points — from 56.2% to 55.0% — and cost 4.7 points of annual return'],
  ['Heavy protection on ETH', 'Made the drawdown WORSE. Paying premium every roll is itself a drawdown, and it arrives whether or not the crash does'],
  ['Through the 2022 crash', 'Saved 1.8 points on BTC and 0.8 on ETH — after paying for four years to be holding it on the day'],
]} />

<Callout type="danger" title="A hedge that pays out can still be a bad trade">
  The 2022 line is the one worth sitting with. Protection did its job in the worst
  drawdown of the period and still lost, because you pay for it in all the years that are
  not 2022. "It would have paid off in the crash" is true and is not the same as "it was
  worth owning".
</Callout>

## Two ways this measurement goes wrong

Both of these produced results we had to throw away, and both are easy to reproduce if you
go looking for this yourself.

<Steps>
  <Step title="Spending 100% of the sleeve on premium is not a hedge">
    Set the premium budget to the whole account and every roll spends everything on
    options. That reads as −96.6% and it is not a finding — it is a leveraged buyer of
    volatility, not somebody insuring a position. A buying structure is sized *by* its
    premium, so the budget is the whole trade.
  </Step>
  <Step title="Never judge an option strategy at one placement">
    A 365-day put looked like +3.4 points on BTC at a single placement. Its drawdown saving
    was −0.1, which was the tell: that is a directional bet wearing a hedge's clothes. Laddered
    across four placements the advantage vanished and it lost in six of eight windows.
  </Step>
</Steps>

<Callout type="warn" title="The second one is general">
  A strategy placed once is measured at one moment, and the moment is doing the work. A
  180-day straddle read +2.0% at one placement and −2.5% at three. If a result moves that
  much when you ladder it, the result was about the start date.
</Callout>

## What the prices are built on, exactly

We would rather you knew where the soft part is.

There is **one measured input**: DVOL, Deribit's 30-day at-the-money volatility index.
Strike is handled by a skew we set, tenor by a term structure we set. Historical option
chains are not available from anyone, so this is model-priced and always will be.

<Callout type="note" title="The skew is worth more than any other assumption">
  Sweeping it on a headline result moves annual return from +9.3% to +22.2%. Nothing else
  in the model has that leverage, which is why every answer says which skew priced it.
  Checked against a live chain, DVOL itself is sound — within two or three volatility
  points of the market at every tenor from eight days to nine months.
</Callout>

One known bias, stated plainly: real volatility is a smile and this model is not, so it
prices selling calls too cheaply. That means covered-call results are pessimistic by an
amount we have not measured.

## So what do you do instead

Three things, in order.

<Steps>
  <Step title="Price it before you assume it is prudent">
    "I'll hedge the downside" sounds responsible and costs a third of the position a year.
    Nine structures can be priced at once — protective put, collar, put spread, straddle,
    strangle, covered call and more — today and at every roll since 2021. That is one
    question and it spends nothing.
  </Step>
  <Step title="Size the position instead of insuring it">
    If a 55% drawdown is unacceptable, holding less is free. Protection that reduces it to
    54% is not a smaller position, it is the same position with a fee.
  </Step>
  <Step title="Check what the other side pays">
    The side that worked in this data is the opposite one — selling puts with a regime
    filter rather than buying them. That is a different risk with a different failure mode,
    and it is not advice. It is the direction the measurement points.
  </Step>
</Steps>

## The honest summary

If you came here to find out what a hedge costs, the number is 30–40% a year against the
asset's own drift, and it does not improve at any strike or budget we tested.

That is a real answer to a real question, and it is worth more than a tool that cheerfully
builds you the position. The measurement takes one question and no money. The position
takes a year and a third of your capital to reach the same conclusion.