Understanding Bear Markets: What 100+ Years of S&P 500 Data Reveals

A Word Everyone Uses, Few Define Precisely

"Bear market" is one of those terms financial media uses constantly, but the definition has drifted over the decades. The most widely cited rule of thumb — a 20% decline from a recent peak — is a journalistic convention, not a statistical law. It is a useful threshold, but it is also arbitrary: a 19.5% drop and a 20.5% drop are economically indistinguishable, yet only the latter gets the "bear" headline.

A more honest framing is to ignore the threshold and look at the distribution of drawdowns directly. Over more than a century of S&P 500 (and predecessor index) data, large drawdowns are not rare anomalies — they are a recurring feature of equity markets. Anyone who plans to hold stocks for thirty years will live through several.

This post walks through what the data actually shows. All figures here are approximate, based on publicly available S&P 500 (and S&P Composite predecessor) price data from roughly 1900 to 2024.

Table of Contents

• The 20% rule and its limits
• The shape of the drawdown distribution
• Average duration, depth, and recovery
• The top ten bear markets at a glance
• Why drawdowns feel worse than the numbers say
• How Dipsern frames bear markets
• Key takeaways

The 20% Rule and Its Limits

The convention treats a 20% peak-to-trough decline in a major index as a bear market. By that bar, the U.S. equity market has seen roughly a dozen bear markets since 1900, depending on how you count overlapping episodes. The convention has two problems.

First, it is binary. A 22% decline followed by a multi-year recovery is grouped with a 50% collapse followed by a similar recovery, even though the lived experience is wildly different.

Second, it is path-dependent. A market that grinds down 25% over two years is psychologically very different from a market that drops 25% in five weeks. Yet the 20% rule treats them identically.

A more useful approach: look at all drawdowns greater than, say, 10%, and study the distribution. That is closer to what an actual investor experiences year to year.

The Shape of the Drawdown Distribution

If you plot a histogram of every distinct S&P 500 drawdown over the last 100+ years, three features stand out:

1. Small drawdowns are extremely common. Pullbacks of 5-10% happen on average more than once per year.
2. Mid-sized drawdowns (10-20%) occur every few years. They are not unusual; they are baseline noise on a long enough timeline.
3. The right tail (30%+) is fat. Catastrophic drawdowns are not as rare as a normal distribution would suggest. They cluster around macroeconomic dislocations: 1929-32, 1937-38, 1973-74, 2000-02, 2007-09, and 2020.

This fat-tail property is the single most important feature of equity risk. It is also the reason why naive volatility measures (such as standard deviation) systematically understate downside risk.

Average Duration, Depth, and Recovery

Across U.S. bear markets since 1900 — using a 20%+ peak-to-trough rule, and treating overlapping declines as a single episode — three rough averages stand out:

• Average depth: approximately -36% peak-to-trough.
• Average duration (peak to trough): approximately 14 months.
• Average time to full recovery (trough back to prior peak): approximately 24 additional months.

Total round trip: roughly 38 months, or just over three years, on average. The median, however, is shorter than the mean because the 1929-1954 episode (which took roughly 25 years to fully recover in nominal terms, longer in real terms) skews the average dramatically.

Strip out 1929, and the average round trip drops to closer to 30 months.

The Top Ten Bear Markets at a Glance

The table below lists ten of the most cited U.S. equity bear markets since 1929. Figures are approximate and based on monthly closing values of the S&P 500 (and its S&P 90/Composite predecessor). Recovery duration is measured from trough back to the prior peak in nominal terms.

| # | Start (peak) | Trough | Approx. depth | Approx. peak-to-trough | Approx. recovery | |---|---|---|---|---|---| | 1 | Sep 1929 | Jun 1932 | -86% | ~34 months | ~25 years | | 2 | Mar 1937 | Mar 1938 | -54% | ~12 months | ~9 years | | 3 | May 1946 | May 1947 | -28% | ~12 months | ~3 years | | 4 | Dec 1961 | Jun 1962 | -28% | ~6 months | ~1.5 years | | 5 | Nov 1968 | May 1970 | -36% | ~18 months | ~3 years | | 6 | Jan 1973 | Oct 1974 | -48% | ~21 months | ~7.5 years | | 7 | Aug 1987 | Dec 1987 | -34% | ~3 months | ~2 years | | 8 | Mar 2000 | Oct 2002 | -49% | ~30 months | ~7 years | | 9 | Oct 2007 | Mar 2009 | -57% | ~17 months | ~5.5 years | | 10 | Feb 2020 | Mar 2020 | -34% | ~1 month | ~5 months |

A few patterns jump out. Recoveries are not symmetric with declines: it almost always takes longer to climb back than it took to fall. The 2020 COVID drawdown is a striking outlier — both the fastest crash and the fastest recovery on the list. The 1987 crash was sharp but, in retrospect, relatively brief. And the 1929 episode is in a category of its own.

Why Drawdowns Feel Worse Than the Numbers Say

Two structural reasons explain why investors consistently report that bear markets feel worse than the historical statistics suggest.

First, time dilation. When a portfolio is rising, weeks blur into months. When it is falling, every daily print is remembered. A 14-month decline can feel like five years lived in real time. Memory of bull markets compresses; memory of bear markets stretches.

Second, the recovery is invisible from the bottom. When the S&P 500 dropped 49% in 2000-02, no one knew in October 2002 that they were standing at the trough. They knew only that they had lost half their equity, that the news was uniformly grim, and that there was no statistical signal telling them the decline was over.

A retrospective chart with a clearly labeled "trough" obscures the lived ambiguity of the moment. Every bear market, in real time, looks like it might keep going.

This is why disciplined frameworks — rules-based rebalancing, written investment policies, pre-committed allocation rules — outperform discretionary judgment so consistently during drawdowns. The frameworks bypass the cognitive distortion.

How Dipsern Frames Bear Markets

Dipsern does not predict whether the current drawdown will deepen or recover. Nobody can reliably do that. What Dipsern does is answer a more tractable question: for an asset currently down X% from its all-time high, what has historically happened over the next 90 calendar days?

The engine segments the entire drawdown range [-100%, 0%] into equal-width buckets and, for each bucket, computes a rolling median of forward returns and a win rate (percentage of historical observations that ended positive). For the S&P 500 in deep drawdown territory (-30% or more), the median 90-day forward return has historically been positive and the win rate has been well above 50%. That is not a prediction — it is a statistical observation about a base rate.

The point is not to time the bottom. The point is to know what the historical distribution looks like before making a decision, rather than reacting to headlines.

Key Takeaways

• Bear markets are a recurring feature of equity investing, not anomalies — expect several over a thirty-year horizon.
• The "20% rule" is a journalistic convention, not a statistical fact. Look at the full drawdown distribution instead.
• Average U.S. bear market: roughly -36% deep, ~14 months from peak to trough, ~24 additional months back to prior peak.
• Recoveries are not symmetric with declines. The climb back almost always takes longer than the fall.
• Drawdowns feel worse than statistics suggest because of time dilation and the invisibility of the trough in real time.
• Base-rate thinking — what has historically happened from this level? — is more useful than headline-driven timing.

A Note on International Context

U.S. equities are the cleanest dataset for bear-market analysis, but they are also unusually well-behaved by global standards. Japan's Nikkei 225 peaked in late 1989 and did not regain that level for over three decades. European indices have had bears as deep as the U.S., often with slower recoveries. Emerging markets routinely produce 50%+ drawdowns in single calendar years.

The implication: when you study U.S. bear-market statistics, you are looking at the best-recovering major market in modern history. That is informative — most readers do hold U.S.-tilted portfolios — but it is not the global norm. If your portfolio has meaningful international exposure, the "average" recovery profile in this post is optimistic for your situation.

This is one of the reasons Dipsern computes drawdown statistics per asset, not just at the index level. The base rate for SPY's recovery from a -25% drawdown is not the same as the base rate for ASML's, or for a Brazilian large-cap's. The right reference distribution is asset-specific.

Try It Yourself

If you want to see the historical distribution of forward returns for the S&P 500 at its current drawdown level, run the analysis on SPY or the underlying index. Compare it with international or sector ETFs to see how bear-market behavior varies across asset classes.

Educational content. Past performance does not guarantee future results. This is not financial advice.