Are Buffer ETFs Worth the Hype?

From zero to $78 billion in seven years

The first buffered ETFs launched in 2018. Here is where the category stood at the end of 2025:

Source: Morningstar, Defined Outcome ETF Landscape, data as of December 31, 2025. Includes funds of funds.

• 2010: the idea starts in annuities. Equitable launched the first buffered annuity, now called a registered index-linked annuity, in October 2010, when low interest rates had made traditional guaranteed products less generous (Society of Actuaries).
• 2018: the first buffered ETFs. Innovator Capital Management brought the same idea to a fund that trades daily, has no surrender charges, and costs less than most annuities (etf.com).
• 2022: the breakout year. When stocks and bonds fell together, money poured in. The category held about $17 billion across roughly 150 funds by September 2022 (etf.com).
• 2025 and beyond. Assets more than quadrupled from 2022 to the end of 2025. In its optimistic scenario, Cerulli Associates projects the category could quadruple again by 2030 (Cerulli, November 2025).

Who’s buying

According to Cerulli, much of the demand comes from advisors serving clients near or in retirement, as baby boomers retire and near-retirees look for predictability. A large loss right before or early in retirement is hard to recover from, because withdrawals continue while the portfolio is down. Many of these investors want to stay in stocks for growth but can’t afford a 2008-style drawdown.

So is the protection worth what it costs? That’s what the rest of this page tests.

How a buffered ETF works

A buffered ETF, also called a defined-outcome ETF, uses options on an index, usually the S&P 500, to reshape its return over a fixed outcome period, typically 12 months. At the start of each period the fund sets two terms:

• The cap is the most you can earn for the period. If the index rises 25% and the cap is 10%, you get 10%.
• The buffer is how much of a decline the fund absorbs. With a 15% buffer, an index drop of 12% costs you nothing; a drop of 20% costs you 5%.

Some funds add a dual-direction feature: an index loss that stays inside the buffer is paid out as a gain of the same size, so a 7% decline pays +7%. Others offer full protection, where any decline returns 0%, in exchange for a much lower cap. When the period ends, the fund resets with new terms based on interest rates and option prices at that time.

Two costs are easy to miss. The payoff is based on the index’s price return, so you give up its dividends. And the fund charges an expense ratio, which we assume is 0.85% a year.

Horizontal axis: S&P 500 price return over one outcome period. Vertical axis: what the strategy pays, before fees. Strategy S&P 500

The three designs we tested

These are hypothetical designs, not specific funds. Results are before fees in the example column. We hold each design’s cap and buffer fixed in every year; the next section explains why that matters.

The biggest assumption: a fixed cap

Every result on this page assumes the same cap in every year: 6% for full protection, 15% for the standard buffer, and 10% for the dual-direction buffer. Real funds don’t work that way. At each annual reset the fund buys and sells a new set of options, and the cap is whatever those options can pay for at that moment. The main factors:

• Interest rates. Higher rates generally allow higher caps.
• Market volatility. Higher expected volatility has generally allowed higher caps on buffer funds, since the options the fund sells are worth more (Advisor Perspectives).
• Buffer design. A deeper buffer, full protection, or a dual-direction feature each cost more, so they come with lower caps.
• The index’s dividend yield and the fund’s fee also reduce what’s left for the cap.

The swings can be large. For a standard 15%-buffer S&P 500 fund, design 2 above, Innovator estimated a cap of 8.54% to 10.34% for 2022 (December 2021). After a year of rising rates and high volatility, the 2023 cap was set at 18.84% (January 2023). Both figures are before the fund’s 0.79% fee. Our 15% assumption sits between those two years.

So the projected returns here are only as good as the cap assumption. This table shows how the last 50 years would have looked at other caps, holding everything else the same:

Last 50 years, 600 overlapping 12-month periods, net of a 0.85% expense ratio. Highlighted rows are the caps used everywhere else on this page.

As a rough rule, each percentage point of cap was worth roughly 0.4 to 0.7 points of annual return. A fund bought in a year with a low cap can trail these figures meaningfully; one bought when caps are high can beat them. Check the actual cap before you buy.

One year at a time: how often each one made or lost money

Each row compounds every 12-month outcome period in the window. Choose a time period:

What the last 50 years show:

• The standard buffer earned the most of the three, 7.5% a year, because its 15% cap kept more of the strong years.
• The dual-direction buffer was the most consistent. It was positive in 90% of periods, against 76% for the standard buffer, because it turns small declines into gains instead of 0%.
• Both 15% buffers had the same crash risk. Each lost more than 1% in 7% of periods and 31.3% from March 2008, against 43.5% for the S&P 500.
• Full protection never lost more than its fee, but it compounded at 3.5%, below T-bills at 4.3%.

Where each design wins and loses

Grouping the periods by what the S&P 500 did shows the trade directly. Over the last 50 years the index rose more than 10% in 56% of periods, averaging +23.7%. The standard buffer kept +13.4%, the dual-direction buffer +9.2%, and full protection +5.1%. That gap is the price of the protection.

• Small declines are where dual-direction pays. When the S&P fell 0–15%, about 16% of periods, the dual-direction buffer gained (+3.7% and +11.4% on average), while the standard buffer and full protection returned about 0% less the fee.
• Strong years are where the standard buffer pays. In years the index rose more than 10%, it averaged 4.2% more than the dual-direction buffer, thanks to its higher cap.
• Deep declines are where full protection pays. When the index fell more than 15%, both buffers averaged −9.5%; full protection lost only its fee.

Six years at a time: rolling results

Most investors hold for longer than one outcome period. Here each design is rolled into a new 12-month period every year for six years.

• Typical six-year results: S&P 500 +111%, standard buffer +60%, dual-direction buffer +49%, full protection +23%, T-bills +20%.
• Six years did much of the buffer’s job. The S&P 500 lost money in only 3% of six-year periods (worst −4%, starting May 1999), and beat each buffered design in 84% to 88% of them.
• The buffers could still lose over six years, though not by much: the worst was −6% for the standard buffer and −6% for the dual-direction buffer. Full protection never lost money over six years.
• Cash was a real competitor. Full protection trailed T-bills in 51% of six-year periods, the dual-direction buffer in 29% and the standard buffer in 18%, mostly in the high-rate 1970s and 1980s or in periods that included the 2000–02 or 2008 bear markets.

Rolling six-year returns over time

Annualized return for every six-year holding period starting from 1934, when the Treasury bill data begin. Choose a design to compare with the S&P 500 and T-bills, and hover or tap the chart to read any start month.

The buffered designs stay in a narrow band: mostly 1% to 4% a year for full protection, 1% to 9% for the dual-direction buffer, and 1% to 11% for the standard buffer. The S&P 500 ranges from small losses for periods starting in 1936–37 and 1969 to gains above 20% a year. For six-year periods starting from 1970 through 1984, T-bills beat full protection every time and beat both buffers in roughly three of every four.

Pros and cons

Practical points before buying

• Compare with what cash pays today. A fully protected fund only makes sense when its cap, less the fee, clearly beats the Treasury bill or CD yield for the same period.
• Buy at or near the reset date. Buying mid-period gives you the remaining cap and buffer, which can be very different from the headline terms. Issuers publish them daily.
• Don’t pick a month; spread them. No start month was reliably better. Splitting money across several reset months smooths the effect of any single crash.
• Mind the taxes. In a taxable account, ETF gains held over a year are generally taxed as long-term capital gains, while T-bill interest is ordinary income. Inside an IRA, switching funds has no tax cost.
• Match the horizon. Annual buffers fit money you may need in one to five years. For longer horizons, the data favored owning the index.

Methodology and data

How each period is calculated

1. Each 12-month outcome period starts at a month-end index level and ends 12 months later. A new period starts every month, from February 1871 to October 2025: 1,857 periods in all.
2. The S&P 500 price return for the period equals its total return minus the trailing dividend yield at the start.
3. Each design’s payoff is applied to that price return, then the 0.85% expense ratio is subtracted.
4. Treasury bill returns compound the monthly 3-month T-bill yield (yield ÷ 12 each month) over the same 12 months.
5. Six-year results compound six consecutive 12-month periods that share the same start month.

R = S&P 500 price return over the period

Full protection (cap C):
  any R             payoff = min(max(R, 0), C)

Standard buffer (buffer B, cap C):
  if R ≥ 0          payoff = min(R, C)
  if −B ≤ R < 0     payoff = 0
  if R < −B         payoff = R + B

Dual-direction buffer (buffer B, cap C):
  if R ≥ 0          payoff = min(R, C)
  if −B ≤ R < 0     payoff = −R
  if R < −B         payoff = R + B

Net return = payoff − 0.85%

Sources

• S&P 500, January 1871 to June 2003: Robert Shiller’s monthly S&P Composite prices and dividends, as packaged in the datasets/s-and-p-500 repository. Total return each month = (price + dividend ÷ 12) ÷ prior price. Before 1957 this is the predecessor index to the S&P 500.
• S&P 500, July 2003 to September 2026: month-end total returns of SPY (dividends reinvested), chained to the series above.
• Treasury bills: 3-month Treasury bill secondary market rate, FRED series TB3MS, from January 1934.

Download the data

• Every 12-month period (CSV): S&P price and total return, dividend yield, T-bill return, and each design’s net return, by start month.
• Every six-year period (CSV): cumulative six-year returns by start month.
• Monthly index levels (CSV): the total-return index, dividend yield and T-bill yield used for everything above.

Limitations

• Caps and buffers are held constant. Real caps reset each year with interest rates, volatility and option prices, and have nearly doubled from one year to the next. See the cap section for how results change with the cap.
• Before July 2003 the index uses monthly average prices rather than month-end closes, which slightly smooths 12-month returns and understates how often caps and buffers were breached.
• After June 2023 the dividend yield is held at its last reported level.
• “Last 20 years” and “last 50 years” use the most recent 20 or 50 complete periods for each start month. “Since 1934” uses every period starting in January 1934 or later.
• No taxes, advisory fees, or trading costs are included.

Frequently asked questions