Markets

Bull and Bear Markets, Measured

The twenty percent rule has no theory behind it, and two of the last decade's four big declines missed it. What the data shows about depth, duration and recovery.

By the Euphoria team · 2026-07-30 · 8 min read

A decline that stopped short of a name

Between September 20 and December 24 of 2018, the S&P 500 fell 19.8 percent from its closing high. Between February 19 and April 8 of 2025, it fell 18.9 percent. Neither episode was a bear market.

That is not a judgment about how either one felt. It is a definition. The convention is twenty percent, and both declines stopped a fraction short of it, so most published lists of bear markets do not contain them.

Two of the four largest declines in the US stock market over the past decade missed the label by roughly one percentage point each. Once you notice that, the interesting question is no longer what a bear market is. It is what the number twenty is doing in the sentence at all.

• 19.8% the late 2018 decline, which stopped short of the bear market line
• 33 days how long the 2020 decline took to reach its low
• 746 days from the January 2022 peak to a full recovery

Where twenty percent comes from

Nowhere in particular. That is the honest answer.

No statute defines it, no regulator enforces it, and no piece of financial theory derives it. Twenty is a round number that became a convention through repetition, and the convention is now settled enough that official sources state it plainly. The SEC investor glossary says a bear market generally occurs when a broad market index falls by 20 percent or more over at least a two-month period, and a bull market is the mirror image, a rise of 20 percent or more over at least two months.

Read that second clause again, because it is almost always dropped. The definition carries a duration requirement as well as a depth requirement.

Now apply it. The 2020 decline ran from a closing peak on February 19 to a closing low on March 23, which is 33 calendar days. It fell 33.9 percent, clearing the depth test easily, and it failed the duration test by nearly a month. The fastest bear market anyone alive has traded through does not satisfy the clause in the SEC glossary. Nobody minds, because the label describes rather than regulates, and that is exactly the point. A term with no enforcement has no reason to be applied consistently.

Why two sources can date the same episode differently

There are at least three defensible ways to measure one fall, and they disagree.

You can measure closing price to closing price, which is the common convention and the basis of every figure in this article. You can measure intraday extremes, the highest tick to the lowest tick, which always produces a deeper number. Measured that way, the late 2018 decline did cross twenty percent, which is why some accounts call it a bear market and others do not. Both are reporting honestly about different measurements.

You can also measure total return instead of price. The index levels everyone quotes exclude dividends, so an investor who held through a decline was in a slightly shallower hole than the price index shows.

None of the three is wrong. They simply produce different answers, and hardly any chart tells you which one it used.

A word on what that chart does not contain. The daily S&P 500 series published by FRED carries ten years of history, so those four episodes are every decline of more than fifteen percent inside that window rather than the largest on record. The declines of 2000 to 2002 and of 2007 to 2009 were both considerably deeper, and the collapse that began in 1929 was deeper again. Those figures are well documented elsewhere. They are not printed here because they fall outside the series this article can actually check.

How long the falls and the recoveries took

This is the part worth committing to memory, and it is not a definition at all.

In all four episodes the climb back took longer than the fall. In 2020 the index took 33 days to shed a third of its value and 148 days from the low to make it back, so roughly half a year from peak to recovery. The 2022 decline was shallower at 25.4 percent but took 282 days to reach bottom and another 464 days to recover, which is 746 days, a little over two years, from the January 2022 high to the next one.

That shape is the useful generalization. Declines tend to be fast and recoveries tend to be slow, which means the stretch that actually tests an investor is not the crash. It is the long flat period afterwards, when nothing dramatic is happening and the account is still below where it started.

Falls are events. Recoveries are weather.

The arithmetic that makes a deep loss expensive

One piece of hard math sits underneath all of this, and unlike everything else in the article it involves no convention and no judgment.

A loss and the gain that reverses it are not the same size. If a $100 position falls 20 percent it is worth $80, and getting from $80 back to $100 means earning $20 on a base of $80, which is a 25 percent gain. Fall 50 percent and you hold $50, and $50 back to $100 is a 100 percent gain. The base shrank along with the price, so every percentage point of recovery is doing less work than the percentage point of loss that came before it.

That asymmetry is why depth matters more than frequency. It is also why the gap between a 19 percent decline and a 35 percent one is far wider than the headline difference suggests. The first needs a 23.5 percent gain to undo. The second needs 53.8 percent.

Correction and bear market are one thing at two depths

A correction is conventionally a decline of at least ten percent and a bear market at least twenty. Nothing mechanical changes at the boundary. The market does not behave differently at negative 20.1 percent than at negative 19.9 percent.

The two words describe the same process sampled at two arbitrary depths, and the second word only becomes available once the first has already been passed. Every bear market was a correction first, and most corrections never become one. The upgrade happens on the way down, which makes it a report on where you have been.

What the label is actually good for

Very little, as an input to a decision. It is genuinely useful for three narrower things.

• It gives people a shared vocabulary, so that saying the market is in a bear market conveys roughly the right magnitude without a table of numbers.
• It provides a way to compare episodes across decades, as long as everyone measures the same way.
• It offers a rehearsal. Knowing that falls of 20 to 35 percent have happened repeatedly, and that the recovery has historically taken longer than the fall, is the difference between an expected event and a shock.

What it cannot do is say anything about what happens next. The classification is computed entirely from prices that have already printed. A measure built from the past has no forward content, however dramatic it sounds when it is announced.

Reading the next one

You will meet this vocabulary during the next decline, delivered with confidence. Three questions keep it in proportion.

Ask what the measurement window is. Closing prices or intraday, price or total return, which index. A change in any one of those moves the number by percentage points.

Ask whether the figure is the peak to trough depth or the decline from today's level, because coverage slides between the two without saying so.

Ask what the recovery arithmetic implies. A reported 30 percent decline is a 42.9 percent climb back to even, and that sentence carries more information than the label attached to it.

Euphoria's market lessons hand you the series and let you find the peaks and troughs yourself. It is a faster route to the asymmetry than reading about it, because a number you had to compute stops being a headline.