What Is an Inverse ETF? The Daily Reset, Real Sell-Offs and 20 Years of SH and SDS

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-06 · Updated 2026-11-02 · Reviewed 2026-11-01

An inverse ETF is a fund built to return the opposite of its index for one day: up 1% when the index falls 1%, or up 2% for a -2x fund. Held longer, it drifts from the simple inverse; in 2011 the S&P 500 ended flat, yet ProShares’ -1x SH lost 7.82% and the -2x SDS lost 18.81%.

The drift is not only a cost. In the seven real S&P 500 sell-offs since 2006 the funds usually rose by more than the index fell, and then lost it again on the way back up. Both halves are measured below.

How it works

An inverse fund holds swaps, futures or short positions so that it rises when its index falls. It gives short exposure without a margin account and without the borrowing mechanics of short selling. The most you can lose is what you put in.

The objective is stated per day. ProShares’ own page for SH, read on 1 Nov 2026, says the fund seeks daily results that correspond to -1x the S&P 500, and that over any period other than a day the return “may be higher or lower” than that target.

Exposure resets at each close, so tomorrow’s inverse applies to tomorrow’s starting value rather than to the price you paid. FINRA’s Regulatory Notice 09-31 says these funds “typically are unsuitable for retail investors who plan to hold them for longer than one trading session, particularly in volatile markets.”

Each reset compounds. After a fall the fund’s base is larger and the next rise costs more; after a rise the base is smaller and the recovery earns less. In a choppy market that works against the holder. In a market that falls steadily, the same compounding adds to the gain, because each day’s rise is taken on a bigger base.

The three S&P 500 versions

Targets, launch dates, expense ratios and returns are from the ProShares fund pages, read 1 Nov 2026. Over a decade in which the index rose, every version lost money each year on average, and the loss grew with the multiple. The same issuer and others run inverse funds on the Nasdaq-100, the Dow, small caps and single sectors; the mechanics are identical.

A worked example

First the arithmetic, on a made-up two-day path. The index rises 10% from 100 to 110, then falls 9.09% back to 100. A daily-reset -1x fund starting at 100 falls 10% to 90, then rises 9.09% to 98.18. The index is flat and the fund is down 1.82%, with no fee involved. The -2x version falls to 80 and recovers only to 94.55, a loss of 5.45%.

Now a real one: SH through the 2025 sell-off. The S&P 500 closed at a record 6,144.15 on 19 February 2025. SH closed that day at $40.87, so $10,000 bought 244.68 shares.

At the low. On 8 April 2025, 34 sessions later, the index closed at 4,982.77, down 18.90%. SH closed at $49.94, so the position was worth $12,219.23, a gain of 22.19% on price.

The next day. On 9 April the index jumped 9.52% in one session. SH fell 9.35% to $45.27 and SDS fell 18.82%. A single day took back more than half of SH’s gain.

The round trip. On 27 June 2025 the index closed at 6,173.07, its first close above the February high. SH closed at $39.87, so the $10,000 was worth $9,755.32. Counting the two distributions SH paid along the way ($0.311 and $0.418 a share), the total return from the high was -0.73%. The index was back where it started and the hedge was slightly behind.

What it means in practice. Day by day the fund did roughly what it promised: on 9 April it fell 9.35% as the index rose 9.52%. What decided the result was when the position was opened and closed.

In a real sell-off

Every S&P 500 decline of 10% or more since SH and SDS both began trading, measured the same way. A decline runs from a closing high to the lowest close before the index closes at or above that high again. There were seven, from the 56.78% fall that ended on 9 March 2009 to the 18.90% fall that ended on 8 April 2025.

For each, the fund’s return with distributions (adjusted close) was set against the stated multiple of the index’s price fall, then followed until the index got back to its old high.

On the way down the funds paid more than their multiple in six of the seven (SPXU: five of its six). SH’s gain from high to low was a median 1.20 times the index’s fall, SDS’s 1.29 times twice the fall, and SPXU’s 1.26 times three times the fall in its six. The exception each time was 2015-16, a slow, choppy 14.16% decline over 183 sessions, where SH rose 11.63%. In the 2020 crash SH rose 42.33% while the index fell 33.92%.

Held to the recovery, they gave it back. From the high to the day the index recovered, SH ended below its starting value in six of the seven, with a median of -0.96%; its one gain was +0.87% across 2022-24. SDS lost money in all seven, a median 11.61%, and SPXU in all six, a median 17.87%. The 2007-09 round trip, which took until 28 March 2013, left SH down 36.53% and SDS down 71.77%.

The highs and lows are picked with hindsight. Nobody knows the top while it is happening. A point you could know at the time is the first close 10% below the high. Bought there and held until the index got back, SH lost money in all seven, a median 10.89%; SDS lost a median 27.30% and SPXU 39.77%. Most of that is simply the index rising again, which is what a recovery means.

Inverse ETF vs short selling vs puts

An inverse fund’s edge is access. It sits in a cash account or an IRA that cannot short, and the loss is capped at the amount invested. A short sale has no daily reset but needs margin, a borrow and unlimited upside risk. A put option has a known maximum cost and an expiry you choose. A short future has no daily reset at all.

For a genuine hedge, the alternatives are more direct. Selling part of the position removes the risk and costs one round trip. If an inverse fund is the tool because the account allows nothing else, hold it for days rather than months, and size it knowing the tracking will drift the whole time.

The original data

Twenty years of the two oldest S&P 500 inverse funds, by calendar year. Yahoo Finance adjusted closes for SH and SDS were set against the index’s own close over 5,104 shared sessions from 13 July 2006 to 26 October 2026. The gap is the fund’s return minus the stated multiple of the index’s return over the same days, so each gap is the daily reset plus fees and the interest earned on the fund’s cash.

In 2011 the index ended 0.0032% lower with annualized daily volatility of 23.35%: SH lost 7.82% and SDS lost 18.81%, 18.82 points behind -2x, the widest gap of the 19 years. In calm 2024 (12.67%) the index rose 23.31% and SDS lost 29.45%, 17.16 points less than twice the gain. A -2x position reset daily with no fees or interest would have lost 14.91% in 2011 and 37.34% in 2024, so the reset itself made 14.92 of 2011’s 18.82 points and 9.28 of 2024’s 17.16.

Across the 19 full years SH finished behind -1x in 10 and SDS behind -2x in 11. The correlation between the index’s volatility and the year’s gap was -0.44 for SH and -0.68 for SDS: the choppier the year, the further behind the funds ended. Every year is in the calendar-year file.

For one day the funds did their job: behind or ahead about equally often, by a median of about two thousandths of a point. From a month on they were behind more often than not. Every window length is in the holding-window table, and every sell-off figure above is in the sell-off file.

Over the whole span the index rose 510.17% on price to 30 October 2026, while SH fell 91.02% and SDS 99.69%. That is the cost of holding a short view through a rising market.

Checked against Nasdaq. Over the 2,513 to 2,514 sessions each fund shares with Nasdaq from 31 October 2016, Nasdaq’s closes differ from Yahoo’s by more than a cent on 1 day for SH, 2 for SDS and 3 for SPXU, and Nasdaq shows the same three SH closes used in the 2025 example: $40.87, $49.94 and $39.87.

When it fails

A range produces losses on both legs. 2011 is the measured version: an index back where it started, a -2x fund down 18.81%, and a holder whose view was never tested.

Holding it as insurance until the storm passes. The sell-off data shows the pattern: the gain is real at the low, and the recovery takes it back, faster for -2x and -3x. A hedge needs an exit plan before it is bought.

Buying after the fall has started. From the first close 10% below the high, SDS lost a median 27.30% by the time the index recovered.

One violent up day. On 9 April 2025 SDS lost 18.82% in a session. A gap up at the open hits a -2x fund before any order can be placed.

Questions people ask

What is an inverse ETF?

A fund built to return the opposite of its index for one day, at -1x, -2x or -3x. It holds swaps, futures or short positions sized for that each day, and the exposure resets at every close.

Can you hold an inverse ETF long term?

You can, but the result drifts from the simple inverse. In 2011 the S&P 500 ended flat while SH lost 7.82%; FINRA calls these funds typically unsuitable for retail investors holding longer than one session.

Do inverse ETFs go up in a crash?

Yes, often by more than the fall: from 19 February to 23 March 2020 the index fell 33.92% and SH rose 42.33%. The question is what happens after; by 18 August 2020, when the index was back, SH was down 11.07% from the February high.

What are some 2x inverse ETFs?

On the S&P 500, ProShares’ SDS; on the Nasdaq-100, its QID. Both launched on 11 Jul 2006 and both reset daily. In the seven sell-off round trips measured here SDS lost money every time, a median 11.61%.

Are inverse ETFs a good idea?

It depends on the holding period. They offset falls day by day. Held to the recovery in the seven declines since 2006, SH ended lower than it started in six, so the measured pattern fits a short, planned hedge better than a standing one.

Why did SDS not double the S&P 500’s 2008 fall?

The path. The index fell 38.49% with daily volatility of 41.02%, and SDS rose 60.50%, not the 76.97% that twice the fall suggests.

Related

Leveraged ETF is the same mechanism pointed upward. Hedging covers what a hedge is meant to do. And short selling is the direct alternative and its own mechanics.

This page is educational, not financial advice. Test every idea on your own charts before risking money.