Your screen is red. The market has fallen for several sessions, the headlines have discovered the word “turmoil,” and someone on television has drawn a line pointing down. What is happening? A pullback? A correction? A crash? The honest answer is irritating: you can measure the loss so far, but you cannot read the final label from an unfinished path.
Suppose you see a 5% decline. One historical example here was already near its ultimate bottom and would regain the old peak two days later. Another decline kept going until it exceeded 33%. The first number did not contain the ending. It only described the journey up to that moment.
Markets move in real time. Their categories are written in the past tense.
Four completed paths—now stripped of suspense
All four use selected local closing-price peaks and daily closes. The set was chosen to show different depths and clocks; it is not a random sample, frequency estimate or claim that the episodes began alike.
Make the call before the reveal
Hide the dates for a moment. Case A bottoms after four calendar days. Case B falls for 13. Case C for 24. Case D for 33. Before seeing the final loss, which one would you call a crash? If your answer depends on how the story ends, you have discovered the problem with using the label as a live decision rule.
At what point would you change your plan?
Each answer uses a different kind of evidence. Percentage thresholds can be observed, but they say nothing about what comes next. Consecutive losses depend on the chosen interval. A rebound can fail. Full recovery is knowable only after much of the climb has happened. None is secretly “the bottom.”
What the familiar thresholds actually say
Market commentary commonly calls a fall of at least 10% from a high a correction and a fall of at least 20% a bear market. Stockspanic uses those depth bands to organize its historical picker: below 10% is a pullback, 10% to below 20% a correction, and 20% or more a bear market. The word “crash” has no equally tidy universal percentage rule.
These depth bands are descriptive conventions, not diagnoses. Crossing 10% does not start a different financial machine. Crossing 20% does not reveal how much farther the market will fall, how long it will remain underwater or whether an investor should sell. A rapid bear-market decline can recover sooner than a milder correction. Depth and time are separate coordinates.
| Episode | Band after completion | Fall | To bottom | Bottom → recovery | Total |
|---|---|---|---|---|---|
| Brexit vote | Pullback | −5.62% | 4 days | 2 days | 6 days |
| February 2018 | Correction | −10.16% | 13 days | 197 days | 210 days |
| India election 2004 | Bear-market depth | −23.97% | 24 days | 175 days | 199 days |
| US COVID shock | Bear-market depth | −33.92% | 33 days | 148 days | 181 days |
A deeper fall can finish its cycle sooner
The February 2018 S&P 500 correction lost 10.16%. Its old January closing peak was regained 210 calendar days after the selected peak. The 2020 COVID shock lost more than three times as much, yet regained its old February closing peak in 181 days total. The deeper episode completed this particular peak-to-recovery measure 29 days sooner.
That observation is memorable, but it is not a law. It does not prove severe crashes recover quickly. It proves only that severity alone did not order these two completed histories by recovery time. Japan’s decades-long underwater cycle exists in the same research library. So does the six-day Brexit path. A useful scenario set contains both.
Why this research uses closing prices
Intraday records make dramatic headlines, but they can describe a price that existed briefly and was unavailable to many decisions. Stockspanic defines peaks, bottoms and recoveries using daily closing index levels. It does not include intraday-only crashes, and it does not mix an intraday low with a closing recovery.
This matters especially for India on 17 May 2004, when the intraday move was larger than the closing loss. The selected Sensex path uses the closing level. That choice is not inherently superior for every question; it is consistent with the calculator’s historical replay and makes the four statistics comparable by observation type.
The name can change your behavior
Calling a decline a “crash” can make action feel urgent. Calling it a “correction” can make inaction feel sophisticated. Neither word knows your time horizon, savings buffer, portfolio composition or capacity to tolerate a loss. The label can become emotional permission for a decision that was never written down.
A more useful prompt is concrete: “I will panic if the market falls more than 5%.” That sentence does not pretend the threshold predicts the ending. It admits where your behavior may change. You can then test the same response against a path that stopped at 5.62%, one that crossed 10%, and two that crossed 20%.
Keep the response fixed as you switch histories. If you move everything to savings, state how you return: after a chosen drop, the first rebound, full recovery or a number of months. The exercise is not to win each historical replay. It is to see which assumptions your future self would have to execute while the category was still unknown.
What this does not prove
- The four selected episodes are not statistically representative and cannot estimate how often a small drop becomes a crash.
- The examples did not share identical early paths, news, valuation, policy response or market structure.
- Recovery means the benchmark’s first close back at the selected old peak—not a personal portfolio’s inflation-adjusted break-even.
- All four are price indices here, so dividends are excluded. Currency conversion, tax, fees and contributions are also excluded from these headline figures.
- The 10% and 20% bands are descriptive conventions, not trading instructions or forecasts.
Method and sources
Statistics use daily closing levels and elapsed calendar days. Exact source URLs and captured-response hashes remain in the downloadable research record.
- FTSE 100 historical data and the Reserve Bank of Australia’s August 2016 market context.
- S&P 500 historical data for February 2018 and COVID-19.
- BSE Sensex historical data and the Reserve Bank of India 2003–04 annual report.
- Yardeni Research’s S&P/Haver correction and bear-market table for conventional US depth bands.
- Stockspanic selection limits, definitions and reproducibility notes.