The Buffer Is a Promise About One Day
A buffered ETF makes one promise, about one day, measured from one level, to one holder who was there at the open. On every other day you own a portfolio of options, and the price says so.
The complaint arrives in the same shape every time. The market fell, the fund had a 10% buffer, the account is down, and somebody wants to know why the protection did not work.
It did work. It just has not been delivered yet.
A defined-outcome ETF is a contract about a single date. The buffer and the cap are properties of the outcome period's final day, calculated from the index level on its first day. Between those two dates the fund holds options, and its price is whatever those options are worth — which is not the same number, and is not supposed to be.
Nobody writes that down. The fact sheet says "10% downside buffer" in large type and "over the outcome period" in small type, and the second half is the half that governs.
What the fund actually owns
Strip the wrapper and a 10% buffered fund on a one-year outcome period holds four options on the index, all expiring on the reset date:
- a call struck at zero, which is synthetic index exposure
- a short call at the cap, which pays for everything else
- a long put at the starting level
- a short put 10% below the starting level
The long put and the short put together are the buffer. They absorb the first 10% of decline and stop. Below that, the fund falls with the market, ten points better off.
That structure reproduces the advertised outcome exactly. Priced at expiry across the full range — down 50%, down 30%, at the buffer edge, at the cap, up 50% — the four options settle to the published payoff at every point, to twelve decimal places. The contract is real and it is precise.
The precision is the problem, because it is precision about one day.
The gap between the mark and the promise
Here is the same fund, a 10% buffer with a 23.46% cap, priced through the outcome period at 20% volatility. The middle column is what the fund is worth. The right column is what it would pay if today were the reset date.
One month in, eleven months left:
| Index | Fund is marked | Would pay at reset | Gap |
|---|---|---|---|
| −15% | −9.71% | −5.00% | −4.71% |
| −10% | −6.13% | 0.00% | −6.13% |
| −5% | −2.76% | 0.00% | −2.76% |
| 0% | +0.38% | 0.00% | +0.38% |
| +10% | +5.95% | +10.00% | −4.05% |
| +25% | +12.15% | +23.46% | −11.31% |
The index is down 10%, the buffer covers exactly that, and the fund is marked at −6.13%. Every dollar of protection is still there. None of it is available yet.
The same mechanism runs the other way. The index is up 10% and the fund shows +5.95%, because the short call has moved against the position and eleven months of time value still separates the mark from the outcome.
Now the same fund with two weeks left:
| Index | Fund is marked | Would pay at reset | Gap |
|---|---|---|---|
| −15% | −5.18% | −5.00% | −0.18% |
| −10% | −1.52% | 0.00% | −1.52% |
| −5% | −0.12% | 0.00% | −0.12% |
| +10% | +9.94% | +10.00% | −0.06% |
| +25% | +22.03% | +23.46% | −1.43% |
The gap closes as the options run out of time. It always does. The product is not broken at eleven months and fixed at two weeks — it is converging, on schedule, toward a number that was fixed on day one.
Anyone who judges the fund continuously against a promise that resolves discretely will conclude it is failing, right up until the day it does exactly what it said.
Whose buffer is it
This is the part that costs people money, and it is not a subtlety — it is the whole economics of the position.
The buffer runs from the starting level down 10%. It does not follow the buyer. Purchase mid-period and you inherit whatever is left of it, priced into the NAV you pay.
Buying six months into the same outcome period:
| Index when you buy | NAV you pay | Index decline you absorb before losing money | Advertised buffer |
|---|---|---|---|
| +10% | 108.34 | 1.51% | 10% |
| +5% | 105.27 | none — index must rise 0.26% to break even | 10% |
| 0% | 102.06 | none — index must rise 2.06% to break even | 10% |
| −5% | 98.79 | 6.54% | 10% |
| −10% | 95.42 | 5.09% | 10% |
| Held from day one | 100.00 | 10.00% | 10% |
The last row assumes a cap struck at exactly fair value. At the published 23.46% the day-one NAV is 100.0004 rather than 100.00, so the fund is fractionally rich and the break-even sits at the start level instead of ten points below it. That is a rounding artifact of the cap, not a feature of the product — but it is why the workbook's break-even column reads 0% for a day-one holder, and it is worth knowing before you conclude the model is broken.
Buy after a 10% rally and you have roughly a point and a half of protection while the marketing material on your screen still says ten. Buy when the index is flat but six months of decay have accrued and the fund needs the index to rise before you are whole.
None of this is hidden. All of it is in the pricing. But it is not in the name of the product, and the name is what gets remembered.
This is also the honest explanation for the underperformance complaints that surface after every drawdown. The fund trails plain index exposure. The account is down. The buffer is fully intact the entire time, exactly as written. Nothing failed — continuous judgement was applied to a discrete instrument.
The cap is not set by volatility
There is a standard explanation for why caps move, and it is repeated everywhere, including by people selling the funds: high volatility makes protection expensive, so investors get less upside.
The arithmetic does not support it.
Hold the buffer at 10% and the tenor at one year, and solve for the cap the structure can fund:
| Volatility | Cap |
|---|---|
| 12% | 15.62% |
| 15% | 17.88% |
| 20% | 23.46% |
| 25% | 30.64% |
| 30% | 39.21% |
Caps go up with volatility, and substantially. The reason is that both legs reprice, and they do not reprice equally. The buffer is a put spread — its cost is bounded, because it stops at ten points no matter how bad things get. The call being sold to finance it is unbounded. When volatility rises, the call's premium grows faster than the put spread's cost, so the same funding requirement is met by selling a call further out. More upside, not less.
Skew narrows this and does not reverse it. Price the call at a lower volatility than the puts — the shape every equity index surface actually has — and steepen that spread as volatility rises, which is also what surfaces actually do: say 4 points of spread in calm conditions and 6 in stressed ones. A 10% buffer then funds 16.43% at 20% volatility and 26.62% at 30%. On a flat surface those two regimes were 15.75 points of cap apart; with skew they are 10.19. Narrower, same direction.
What actually cuts caps is rates, through the forward:
| Risk-free rate | Cap |
|---|---|
| 0% | 12.54% |
| 2% | 17.14% |
| 4.5% | 23.46% |
| 6% | 27.62% |
| 8% | 33.69% |
Take the stressed case above — 30% volatility, 6 points of skew, a 26.62% cap — and drop rates 300 basis points. The cap falls to 19.89%, below the 23.46% that a calm, flat surface funds at 4.5%. That is the real mechanism, and it explains the pattern better than the volatility story does. Caps collapsed in 2020 when rates went to zero. Caps were generous in 2022 and 2023 when volatility and rates were high, which the standard explanation cannot account for at all.
If you are choosing an entry point, you are making a bet on the forward far more than on implied volatility.
What the price of the protection actually is
Stated without adjectives, for the structure above, held start to finish:
| Index return | Buffered outcome | Difference |
|---|---|---|
| −40% | −30.00% | +10.00% |
| −20% | −10.00% | +10.00% |
| −10% | 0.00% | +10.00% |
| 0% | 0.00% | 0.00% |
| +10% | +10.00% | 0.00% |
| +30% | +23.46% | −6.54% |
| +40% | +23.46% | −16.54% |
Ten points of protection, always. Everything above 23.46%, forfeited. That is the trade, and it is a defensible one. It is simply not the trade most buyers think they are making, because they are shown the left half of the table.
What to write down
Four lines, before the position is on, in language somebody can check later.
- The date. Which reset the outcome refers to, and that no promise applies before it.
- The level. The index level the buffer is measured from — not today's, and not the buyer's.
- The entry. What was paid relative to the outcome, and therefore how much of the advertised buffer this holder actually owns.
- The ceiling. The cap, and the fact that it was set by the forward and the surface on one particular morning and cannot be renegotiated.
A position with those four lines written down produces no surprises, because every disappointing month has already been described in advance. A position without them produces a meeting after the drawdown, in which somebody explains option mechanics to a board that would rather have read it a year earlier.
The buffer was never the interesting part. The date was.
The calculator
Every figure in this piece comes from a Black-Scholes-Merton workbook you can download and check. It prices the four-option structure, solves for the cap a given buffer can fund, and shows the interim mark beside the outcome for any index level and any point in the period — the gap in the tables above, reproduced on your own inputs.
It is on the Tools page.
Scarborough Road works with insurers and asset managers on derivatives governance, hedging operations, and the investment data infrastructure that demonstrates the governance was followed.
General commentary on product mechanics. Not legal, actuarial or investment advice, and not a recommendation of any fund. Figures are illustrative outputs of a Black-Scholes-Merton model under stated assumptions — constant volatility, constant rates, continuous trading, lognormal returns, a single flat volatility except where a two-volatility skew is stated — none of which hold exactly. Real quotes reflect a full surface, bid-offer and financing terms, and will differ. Verify against fund prospectuses and your own pricing systems. No specific fund's outcome is described or predicted.
Illustrative parameters: one-year outcome period, 10% buffer, cap solved at 23.46%, S = 100 at inception, r = 4.5%, q = 1.5%, sigma = 20%.
