What the KOSPI did next, by fear & greed band
Across 7,020 trading days from 1998-01-30 to 2026-07-27, every day is grouped by the fear & greed band it fell in, and the KOSPI's subsequent move is recorded. People usually pose this as a backtest — "what if you had bought in extreme fear?" — but no entry or exit is assumed here. This is simply what the market did after each band.
These are not strategy returns. No entry or exit is assumed, transaction costs, taxes and dividends are excluded, and nothing here recommends any action. Past distributions do not forecast future ones.
Median subsequent move
For every day in a band, the KOSPI's percentage change N trading days later, summarised by the median. The median is used instead of the mean because a handful of crashes and melt-ups pull an average a long way.
| Band | 20d | 60d | 120d | 250d |
|---|---|---|---|---|
| Extreme fear | +1.7% | +4.0% | +7.6% | +11.9% |
| Fear | +0.3% | +1.3% | +1.6% | +5.1% |
| Neutral | -0.2% | -0.9% | -1.0% | +1.7% |
| Greed | +0.8% | +1.7% | +1.5% | +4.5% |
| Extreme greed | +1.2% | +4.1% | +7.5% | +8.9% |
| All days (any band) | 0.9% | 2.2% | 3.0% | 6.8% |
Share of observations that rose
The proportion of the same observations where the move N trading days later was above zero. Two bands can share a median and still have very different distributions.
| Band | 20d | 60d | 120d | 250d |
|---|---|---|---|---|
| Extreme fear | 60.4% | 61.8% | 64.7% | 75.3% |
| Fear | 52.8% | 56.2% | 53.7% | 68.5% |
| Neutral | 48.9% | 46.9% | 46.0% | 56.3% |
| Greed | 56.3% | 59.9% | 55.8% | 60.7% |
| Extreme greed | 59.8% | 68.5% | 65.3% | 65.0% |
| All days (any band) | 56.5% | 60.6% | 58.4% | 65.1% |
How wide the outcomes were (60 trading days later)
A median hides dispersion. The quartiles and the extremes show how far apart outcomes ran within the same band.
| Band | Trading days | Episodes | 25th pct | Median | 75th pct | Min | Max |
|---|---|---|---|---|---|---|---|
| Extreme fear | 1,051 | 96 | -4.6% | +4.0% | +9.8% | -40.1% | +84.7% |
| Fear | 1,257 | 201 | -6.4% | +1.3% | +7.0% | -46.3% | +108.9% |
| Neutral | 701 | 213 | -6.6% | -0.9% | +5.6% | -45.4% | +82.9% |
| Greed | 2,084 | 251 | -3.9% | +1.7% | +6.9% | -41.2% | +72.5% |
| Extreme greed | 1,927 | 129 | -1.7% | +4.1% | +12.4% | -32.7% | +49.5% |
What the record shows
- Over 60 trading days, the median move across all days is 2.2%. Each band should be read against that number, not against zero.
- Extreme fear was followed by a median of 4% and extreme greed by 4.1% — both extremes sit above the all-days median. Contrary to the familiar "buy fear, sell greed" framing, greed readings were not followed by declines in this record.
- The weakest band is Neutral, at -0.9%. In this data the dividing line is not fear versus greed, but the extremes versus the middle.
- Dispersion is wide in every band. Large gains and large losses both follow the same readings, so a single median cannot narrow down the next outcome.
How to read this carefully
- The observations overlap. Computing a 60-day forward move on every date means neighbouring observations cover almost the same stretch of market. The observation count is therefore not a count of independent samples; the number of distinct episodes is far closer to the effective sample size.
- Bands with few episodes have medians that hinge on a handful of events. The thinnest here is Extreme fear, with 96.
- Only the KOSPI price index is measured. Dividends, transaction costs and taxes are excluded.
- Band assignment uses the index values this site published, which are never recomputed after publication — the same formula applied consistently across the whole record.
Download the full history (kfg.json) · 1998-01-30 ~ 2026-07-27 · CC BY 4.0