In this article
← All articles
Investment allocation

Nasdaq correction in 2026: what past falls mean for your investing plan


The Nasdaq Composite closed 13.5% below its record on 19 November 2026, its lowest close of this fall; history supports planning for both a short correction and a prolonged loss, without telling us which comes next. Of 31 completed falls of at least 10% since 1971, 13 reached 20% and four reached 35%. Most regained their record within a year, while the March 2000 peak took fifteen years to recover.

For someone building wealth through monthly ETF or stock purchases, the useful question is whether the plan can continue under either path. Check the same technology holdings across your funds, the money you need to withdraw and how secure your contributions would be during a downturn. A recovery forecast cannot do that work for you.

Long-term compounded growth is the objective. For an investment without external cash flows, its compound annual growth rate (CAGR) is the constant annual rate connecting its starting and ending value. Diversification or protection helps that objective only when the benefit to the whole portfolio justifies its costs. The historical counts below describe the Nasdaq's price record; they do not establish that a hedge, a sale or another ETF would improve your future return.

Key takeaways

  1. A 13% fall leaves several paths open. Of 19 completed episodes reaching the 2026 low of 13.5%, 13 reached 20% and 6 stopped short; those counts are not a forecast.
  2. The recovery can matter for years. Median peak-to-recovery time was 3.8 months for 10-20% falls and 47.6 months for falls of at least 35%, with wide variation within each group.
  3. Another fund may repeat your largest exposures. Six AI-linked companies were 18.3% of a widely held MSCI World ETF on 5 November; add their weights across all your holdings before judging diversification.
  4. Contributions change the account's recovery clock. In the March 2000 scenario, regular purchases brought contribution break-even forward by about three years, without improving the index's investment return.
  5. Protection needs a whole-portfolio growth case. Compare its cost in ordinary markets with its benefit through losses; a smaller drawdown alone does not establish a higher long-term CAGR.

Where does the Nasdaq stand after the 19 November fall?

The Nasdaq Composite closed at 23,879.20 on 19 November, 13.48% below its 6 October record of 27,599.89, its sixth close of the episode at least 10% below and the lowest so far. The previous low was 5 November, when the index closed at 24,315.00, 11.90% below the record. Regaining the record from the 19 November close needs a 15.58% gain.

On 16 November, the last close in this study's history, the index was 11.20% below the record. It then closed at 24,328.38 on 17 November and 24,429.89 on 18 November, before falling 2.3% on 19 November. AP wrote that day that "AI stocks kept falling on worries their boom may be over."

AP's market report for 16 November said: "Nvidia led another sell-off in technology stocks and helped drag the U.S. market lower." It put the Nasdaq down 626.52 points, or 2.5%, and the S&P 500 down 1.9% to 7,416.59. Nvidia published its quarterly results after the US close on 17 November; nothing below uses them.

With dividends reinvested, Nasdaq's total-return index stood 11.16% below its own 6 October record on 16 November, practically the same loss. This is the second Nasdaq correction in about a year: the index fell 13.2% from October 2025 to March 2026 and regained its record in April. Our November AI sell-off analysis covers the week of 6 November and the world-index weights; this article looks at what happened next in the index's own history.

How often did a 10% Nasdaq fall become a bear market?

In 13 of 31 completed episodes since 1971 the Nasdaq went on from a 10% fall to a close at least 20% below its record, and in 4 of them to 35% or more. We count an episode from a record close to the first later close at or above it, and keep it if any close fell 10% or more below the record. The current episode, open since 6 October, is the 32nd.

The base rate changes with the depth already reached. Of the 19 completed falls that closed at least as far below their record as the Nasdaq did on 19 November, 13 reached 20%, or 68%. We compare unrounded losses: the Nasdaq was 13.481% below its record, while the 1997 fall stopped at 13.476%. Including that near miss would give 13 of 20, or 65%. Of the 17 that reached 15%, 13 reached 20%. Once the 20% line was crossed, 4 of 13 went on to 35% or more.

Completed Nasdaq Composite episodes since 1971. Had closed 10% below the record, 31 episodes: 41.9% went on to 20% or more below and 12.9% to 35% or more. 13.5% below, the 2026 low on 19 November, 19 episodes: 68.4% and 21.1%. 15% below, 17 episodes: 76.5% and 23.5%. 20% below, 13 episodes: 100% and 30.8%.

Reached 20% after a 10% fall: 13 of 31 (41.9%); after a fall as deep as 19 November: 13 of 19 (68.4%); after a 15% fall: 13 of 17 (76.5%); 35% after a 20% fall: 4 of 13 (30.8%). Source: FRED NASDAQCOM and Nasdaq Global Indexes, COMP history; base rates (CSV), all 32 episodes (CSV).

The groups shrink as the depth threshold rises, because shallower falls drop out. In the 19-episode group, one more or one fewer bear market changes the reported share by about five percentage points. The sample also spans very different markets and index constituents. Treat these figures as historical scenario evidence, rather than the probability that today's fall crosses a particular threshold.

Does an 11% fall after six weeks look like the crashes or the corrections?

Similar early losses preceded both recoveries and much deeper falls. After 29 trading days, the time elapsed from the 6 October record to 16 November, six of the 31 completed episodes were already at least 11.2% below their record; five of those went on to fall 20% or more, and the 1996 fall stopped at 16.6%. Some of the deepest falls, however, started slowly.

Nasdaq Composite, percentage below the record close 29 trading days after the record, and the episode's lowest close. 2026, from the 6 October record, open: minus 11.2% and minus 13.5% at its lowest close so far, 19 November. 2000 dot-com: minus 27.8% and minus 77.9%. 2020 pandemic: minus 21.6% and minus 30.1%. 1998: minus 18.6% and minus 29.6%. 1980: minus 16.0% and minus 24.9%. 1978: minus 13.0% and minus 20.4%. 1996: minus 13.0% and minus 16.6%. 1987 Black Monday: minus 2.3% and minus 35.9%. 2021-22 rate rises: minus 1.4% and minus 36.4%. 2024-25 tariffs: minus 2.4% and minus 24.3%.

29 trading days after the record: 2026 −11.2%; 2000 −27.8% (lowest −77.9%); 2020 −21.6% (−30.1%); 1987 −2.3% (−35.9%); 2021-22 −1.4% (−36.4%). Source: FRED NASDAQCOM and Nasdaq Global Indexes; same-point comparison for all 31 episodes (CSV).

The 2021-22 fall was only 1.4% below its November 2021 record after six weeks and ended 36.4% below it. The 1987 fall was 2.3% below after six weeks and reached 35.9% below three weeks later, in October. In 1973 the index was 10.6% below at the same point, close to the 2026 figure, and fell 59.9% before its October 1974 trough.

Speed varied as much as depth. Of the 13 bear markets, the median gap between the first close 10% below the record and the first close 20% below was 29 days; it took 3 days in 1987 and 246 days in 1989-90. In six of the 31 completed episodes the first close 10% below the record was also the lowest close.

How long did the Nasdaq take to recover from each kind of fall?

The median wait from record to new record was 3.8 months for falls that stopped between 10% and 20%, 7.8 months for falls of 20% to 35% and 47.6 months for falls of 35% or more. The deep group splits into two clocks of very different length: the fall itself, then a rebound that took longer still.

Depth of the fall (completed episodes)Record to lowest closeLowest close to new recordRecord to new record
10% to 20% (18)median 1.7 months, at most 6.8median 1.8 months, at most 8.6median 3.8 months, 1.8 to 12.5
20% to 35% (9)median 3.7 months, at most 14.5median 3.6 months, at most 17.4median 7.8 months, 3.6 to 30.5
35% or more (4)median 17.0 months, 2.1 to 31.0median 34.2 months, 14.1 to 150.4median 47.6 months, 23.3 to 181.4

The four deep falls were 1973-74 (59.9%, record regained after 5.7 years), 1987 (35.9%, 1.9 years), 2000-02 (77.9%, 15.1 years) and 2021-22 (36.4%, 2.3 years). Across all 31 episodes the median wait was 5.5 months, a figure that hides the 181 months after March 2000. The episode list (CSV) gives each record, trough and recovery date.

Counting from records also hides a fall. The Nasdaq dropped 55.6% from 2,859.12 on 31 October 2007 to 1,268.64 on 9 March 2009, but both levels were below the March 2000 record, so the 2008 crash sits inside the 2000 episode rather than counting as a 33rd. A household that started investing in 2007 lived through a bear market our count does not list separately.

The general arithmetic of the two clocks, and why a 50% loss needs a 100% gain, is in our guide to recovery time and loss maths. Falls of 75% or more in whole national markets, and what bonds and leverage did in them, are covered in our 80% crash history.

Do dividends and the index's changing make-up alter these counts?

Dividends barely moved recent Nasdaq recoveries, but the index measured in 1973 was a different market from today's. Nasdaq's total-return index (XCMP) compounded at 12.93% a year from 25 September 2003, the start of our retained series, to 16 November 2026, against 11.90% for the price index. That 1.03-point gap matters over decades but little within a short fall.

In the ten completed episodes since 2003, the total-return index regained its record on the same day as the price index or up to three weeks earlier. After the 2021-22 fall it took 26.6 months instead of 27.3. The retained total-return series does not cover the 2000 episode. Its 15.1-year recovery is therefore a nominal price-index observation, without a matched estimate including reinvested dividends.

Nasdaq's index methodology includes "all domestic and international common type stocks listed on the Nasdaq Stock Market", reconstituted daily, with no sector, size or liquidity criterion. Nasdaq's index page listed 3,395 components on 16 November 2026. Companies leave when they delist and are not replaced, so failed companies stay in the record until they go.

What changed is the mix. The 1970s and 1980s index held mostly smaller over-the-counter companies; today its value is dominated by large technology companies that also sit at the top of the S&P 500 and MSCI World. Each episode reflects a different mix of businesses and market conditions. More recent observations share more of today's technology exposure, but recency does not make their loss path a forecast either.

What does this mean for a household that holds a world index?

A world ETF can repeat much of your technology exposure, even though it holds many other companies. In our AI sell-off analysis, Nvidia, Microsoft, Amazon, Alphabet, Meta and Broadcom were 18.3% of iShares' Core MSCI World ETF on 5 November, before counting Apple. The S&P 500 and MSCI World comparison examines another common overlap. The Nasdaq's percentage fall is not a forecast for either fund or for a household account.

To measure your exposure, multiply each fund's portfolio weight by its holding weight in the company, then add direct shares. Use dated holdings files and a consistent currency. That gives you a company exposure you can stress-test; the number of ETF names in your app does not tell you how many different risks you own.

Diversification can work even when two assets both fall. What matters is how their returns, losses and costs combine across the full holding period. A lower-return asset may improve portfolio compounding through that interaction, while a low-correlation holding may still fail to justify its cost. Our independent-bets analysis explains how to examine shared return drivers.

For money invested at a peak, regaining the same price leaves no nominal price gain over that interval. A household's result can differ because of dividends, contributions, withdrawals, fees and currency. That is why a useful plan tests the loss path and cash needs alongside the eventual return, rather than treating the index's recovery date as the account's recovery date.

Protection needs the same whole-portfolio test. Our tail-hedge study found that a small TAIL allocation reduced the sampled 2026 fall while lowering annualised growth over the full observation window. That result does not rule out protection; it shows why its cost and payoff must be assessed together before calling it growth-enhancing.

What do continued contributions do to the recovery clock?

Continued purchases brought contribution break-even forward in these scenarios, most noticeably after the longest fall. In a scenario starting at the March 2000 record with 100 invested and 1 added on the first trading day of each month, the balance last regained the total paid in during June 2012, after 12.2 years. The index needed 15.1 years.

Scenario on the price index, start at the recordIndex back to recordBalance back above money paid in, adding 1 a monthAdding 2 a month
2000-02 (−77.9%)15.1 years12.2 years11.7 years
1973-74 (−59.9%)5.7 years5.3 years5.2 years
2021-22 (−36.4%)2.3 years2.2 years2.1 years

The break-even in this table is the last date from which the balance stayed at or above the money paid in until the index regained its record. In the 2000 scenario with 2 a month, the balance first passed its contributions in May 2007 and fell back below in the 2008 crash. Data: contribution scenarios for all 13 bear markets (CSV).

This is a scenario on a price index without dividends, costs or taxes, not a suggested contribution. It separates two clocks a saver can confuse: the index's return to its record, and the account's return to a profit on money paid in. Contributions do not change the investment return; the lump-sum study compares the timing of new money directly.

The reverse also holds. A household that sells units during the fall to meet spending has fewer units in the rebound, which the withdrawal model quantifies.

What can you check now?

Check what would interrupt your plan before guessing the recovery date. Use your broker statements, dated fund holdings and a calendar of known spending. The scenarios below are checks on your ability to keep investing, not estimates of the next market move or suggested allocations.

  1. Combine repeated exposures. Add direct shares and fund look-through holdings in the same large companies, using each position's weight in your total portfolio.
  2. Test different loss durations. Examine a stated 12% fall lasting four months, a 36% fall lasting two years and a longer period without a new peak. Include the possibility that the rest of your portfolio falls too.
  3. Date withdrawals and debt payments. Compare those commitments with accessible money outside the stressed investments. Forced sales can leave fewer units for a recovery; company failure or lasting impairment can cause permanent loss even without a sale.
  4. Stress-test contributions. If your salary or bonus depends on the same industry as your investments, model a pause in contributions during the loss.
  5. Compare protection after its cost. Test the same portfolio and cash flows with and without the proposed defence, including premiums, fees and returns forgone. Ask whether it improves long-term compounding or serves a specific funding need.

Sources and method

Checked on 17 November 2026, with closes to 16 November 2026; the 17 November close was added on 18 November from Nasdaq's export and AP's report, and the 18 November close on 19 November from Nasdaq's COMP quote, without changing any count.

The 19 November close of 23,879.20 is in Nasdaq's COMP history and moves the base rate from 13 of 24 to 13 of 19, with the same 31 completed episodes. The charts round the threshold to 13.5%; the inclusion rule uses the full loss at that close.

Index levels: FRED NASDAQCOM, Nasdaq Composite price index, daily closes from 5 February 1971 to 13 November 2026. The 16 November 2026 close (24,509.76) is from Nasdaq's Global Indexes history export for COMP and matches AP's figure. Total return: Nasdaq's XCMP history export, from 25 September 2003.

Market report: AP via WTOP, 16 November 2026; its attribution of the fall is a press account. Market reports for 17 November and 19 November: AP via WTOP, 17 November and 19 November. Results: Nvidia press release, 17 November 2026, announced on 28 October. Index rules: Nasdaq Composite methodology and index overview.

FRED and Nasdaq's export agree on every common date since 1 January 2015. They differ by more than 0.01 points on 75 of 14,057 earlier dates, the last in October 2002; running the episode count on Nasdaq's series alone gives the same 31 episodes and the same 18, 9 and 4 split. The MSCI World weights are from iShares' holdings file of 5 November 2026, as cited in our AI sell-off article.

All episodes, base rates, durations, total-return comparisons and contribution scenarios are copio calculations; the CSV files linked above contain every derived figure. A record is the highest close to date; recovery is the first close at or above it; months are calendar days divided by 30.44 and years by 365.25. Depth bands use the lowest close: 10% to under 20%, 20% to under 35%, 35% or more.

Limits: the price index excludes dividends, and index returns precede fund costs, taxes and currency effects for a euro investor. The counts are small, the index's composition has changed since 1971, and the current episode is open. In the contribution scenario, contributions buy at the first close of each month after the record and are measured against the starting balance of 100.

This article is educational, not personalised investment advice.

FAQ

Is the Nasdaq in a correction in November 2026?

Yes, on the usual 10% definition. The Nasdaq Composite closed at 23,879.20 on 19 November 2026, 13.5% below its record close of 27,599.89 on 6 October and its lowest close so far. The previous low was 24,315.00 on 5 November, 11.9% below the record. The index needs a 15.6% gain from the 19 November close to regain the record.

How often has a Nasdaq correction turned into a bear market?

Of 31 completed Nasdaq Composite falls of 10% or more since 1971, 13 went on to a close at least 20% below the record and 4 reached 35% or more. Among the 19 falls that had already reached 13.5%, the 2026 low on 19 November, 13 reached 20%. These are counts from one price index, not a probability for the current episode.

How long did the Nasdaq take to recover from corrections and bear markets?

From record close to a new record close, the 18 falls that stopped between 10% and 20% took a median 3.8 months, at most 12.5. The 9 falls of 20% to 35% took a median 7.8 months, from 3.6 to 30.5. The 4 falls of 35% or more took 23.3 to 181.4 months; the March 2000 record was regained only in April 2015.

Do dividends change the Nasdaq's recovery time?

Only a little in recent episodes. The retained Nasdaq total-return series starts in September 2003 and compounded 1.03 percentage points a year faster than the price index to 16 November 2026. It regained its 2021 record on 8 February 2024, three weeks before the price index. That series does not cover the 2000 episode.

Do regular contributions shorten the recovery after a Nasdaq crash?

They brought contribution break-even forward in our historical scenarios, without changing the index's return. In a scenario starting at the March 2000 record with 100 invested and 1 added each month, the balance last regained the money paid in during June 2012, after 12.2 years, against 15.1 years for the index. In falls that recovered within two years, contributions changed the date by weeks.

Optional cookies help us measure visits and advertising. Choose each purpose, or decline both. You can change your choice using Cookie settings in the footer. Cookie policy.