What a bear market is, and what one bad week is not

Where the 20 percent line comes from

Twenty percent is a convention. No regulator set it, and no exchange rulebook contains it. The definition shows up the same way across financial education material. A bear market is a fall of at least 20 percent from the most recent closing high. A correction is 10 to 19.9 percent. A new bull market begins once price has recovered 20 percent from its low (NISM, National Institute of Securities Markets, updated 5 August 2026). Nobody enforces the number. It stayed in use because it is round, repeatable, and close to where sentiment has historically turned.

Three ways one bad week actually gets measured

The same fall reads as three different numbers depending on where the count starts. None of them is the wrong number. They measure different things. A headline that swaps one for another without saying so leaves the reader doing arithmetic they never agreed to do.

A daily move is close to close

A daily move compares today's closing price with yesterday's. On 24 September 2026, the Sensex closed at 73,580.54. That was down 1,247.71 points, or 1.67 percent, from the previous session. The Nifty 50 closed at 23,063.10 the same day, down 383.70 points, or 1.64 percent (Free Press Journal, 24 September 2026). That single number describes one session. It carries no information about the four other sessions around it.

A weekly move is five closes, not the roughest day inside it

A weekly move compares Friday's close with the Friday before, five sessions apart. The week to 25 September 2026 closed with the Nifty 50 down 0.88 percent. That was its seventh straight weekly decline, the longest losing run since 2020 (Mint, 25 September 2026). Seven-week losing streaks are uncommon. Over the last 25 years the Nifty has recorded one only four times: in 2008, in 2020, and twice in 2021 (Mint, 25 September 2026). One of those 2021 runs reached nine straight weeks.

A weekly number like that can sit on top of a single day that looked far worse. Thursday's 1.64 percent fall was the sharpest single session in ten weeks. The week it sat inside still closed down by less than a point. Consider an illustrative index that opens a week at 23,300. It falls to 23,100 on Monday, recovers to 23,250 on Tuesday, then drops hard to 22,850 on Wednesday, a 380-point single-day fall. It climbs back to 23,150 by Friday's close. The week's number is Friday's close against the prior week's Friday close: 23,150 against 23,300, a decline of 0.64 percent. Wednesday's 380-point session and an average of the five daily changes are both different numbers from that one.

A correction and a bear market are peak to trough

A correction and a bear market are not measured week to week at all. Both start counting from the most recent peak. Both keep counting until the lowest close reached before a new high is set, however many weeks that takes. The Nifty 50's most recent peak was 26,373.20 on 5 January 2026 (NSE India, 5 January 2026). Against the 24 September close of 23,063.10, that is a decline of 3,310.10 points, or 12.55 percent. It took over eight and a half months to build, folded here into a single figure a one-week headline cannot show.

What the label does not tell you

A bear market describes price against its own recent history. It says nothing about a company's earnings, an economy's growth rate, or what price does next. A correction can deepen into a bear market within a month. It can also reverse within days. Commentary on the week to 25 September 2026 was split on both outcomes. Some pointed to rising crude prices and higher global bond yields as reasons for further weakness. Others pointed to improving valuations after the pullback and resilient domestic liquidity (Mint content via TradingView, 25 September 2026). Neither view is the label's job to settle. The number only measures how far price has moved from a point it once reached.

How long India's past bear markets have lasted

NISM's own count of the last 30 years lists eight episodes in the Nifty 50 (NISM, updated 5 August 2026). The shortest was the 2006 Liquidity Crunch. That was a 31.22 percent fall from May 2006's peak of 3,773 to June 2006's trough of 2,595, reached in 34 days. Recovery to the earlier peak took a further 138 days. The slowest to reach its low was the Asian Crisis. That fall ran 39.68 percent, from February 1996's 1,260 to October 1998's 760, and took 996 days to bottom, with a further 260 days to recover the earlier peak. The average decline across all eight episodes was 38.97 percent. Even the COVID-19 crash, a 39.57 percent fall from January 2020's peak of 12,430 to March 2020's trough of 7,511, took 63 days to reach its low. The fastest of the eight still took 34 days, well past a single week. Every one of the eight was confirmed only after the 20 percent line had actually been crossed and held.

What a holder does with the word

The word is useful only as an input to a decision made before the week that tests it. A rule written in advance does not need real-time judgement calls. Rebalance on a fixed calendar rather than on a headline. Size a position as a percentage of the whole portfolio rather than as a rupee amount chosen in the moment. Three percent of a ₹1 crore portfolio held inside a monthly-rebalanced index behaves the same way on a down week as on a flat one. The rebalance date was fixed on the calendar before this week's headline existed.

What crypto has instead of a closing bell

Equity indices have a closing bell. Their peak and trough are always a printed number from a specific session. Crypto has no closing bell. A drawdown, the drop from a peak to whatever low follows it, can be set at any hour. That includes 3 a.m. on a Sunday, when most Indian holders are asleep and no exchange floor is open to react. That weekend print is still a real number. Almost nobody was watching when it was set. Qatobit's four QSI indices are rebalanced monthly on a published methodology, regardless of which day or hour a peak or trough happened to print. A fixed schedule decides when a position changes. The clock the market happened to be running on has no say in it.

Two other gauges moved through the same week and are worth reading alongside this one. What India VIX measures, and how to read a low print covers one. What a bond yield is, and why it rises when the price falls covers the other. How much of my portfolio should be in crypto walks through how a position that already includes crypto gets sized against moves like these. What a crypto index is covers the mechanics behind that position. The glossary carries every term this piece used in one place.

What one bad week actually was

The week to 25 September 2026 was real: a seventh straight weekly decline, the sharpest single session in ten weeks, and a 12.55 percent gap to the January peak. None of that crosses the 20 percent line, and none of it was decided inside five closing sessions. The next question a holder needs answered is simple. Does the position already run on a rule for weeks like this one, or does it wait to find out what it will do only after the headline appears?

Frequently asked questions

What is a bear market?

A bear market is a fall of 20 percent or more from a recent peak in a broad index, sustained over an extended period. A single session or one bad week is never enough on its own (NISM, updated 5 August 2026). It is a convention used across financial education material rather than a rule any exchange or regulator enforces.

How is a bear market different from a correction?

A correction is a fall of 10 to 19.9 percent from a recent peak. A bear market begins once that fall reaches 20 percent, confirmed on the closing price rather than an intraday print (NISM, updated 5 August 2026). Both are measured peak to trough, never week to week.

Is a 20 percent fall in one stock a bear market?

No. The convention applies to a broad market index, such as the Nifty 50 or the Sensex, rather than to a single stock. One stock can fall 20 percent or more inside a single earnings reaction without the broader index reaching the same threshold.

How long do bear markets last?

India's last eight, tracked by NISM across 30 years of Nifty 50 data, took between 34 and 996 days to reach their low. Recovery to the earlier peak ranged from 138 days to 773 days. Duration varies by episode. No single past bear market predicts the length of the next one.

Does a bear market apply to crypto?

The same 20 percent convention can be applied to any priced asset, including crypto. Crypto has no closing bell to mark which session a peak or trough belongs to. A crypto drawdown is measured from whichever hour the peak or trough actually printed, weekday or weekend, rather than from a fixed market close.

Crypto investments are subject to market risk. Not financial advice.