DSPX measures how much stocks are expected to move apart from the index. Across twelve years it does that job well, and gives you no edge at all on where the market is headed or when it will break.
Dispersion is one of the more elegant ideas in the options world. It measures how much individual stocks are expected to move on their own versus moving together with the index. High dispersion means stocks are trading on their own merits, a stock-picker's market. Low dispersion means everything is moving as a herd, all rising or falling together. The Cboe puts a number on the expectation, the dispersion index, DSPX, and traders use it for the trade it is named after: when dispersion is rich, sell index options and buy single-stock options, and the reverse when it is cheap.
Because dispersion rises and falls with stress, a natural question follows: can it warn you of a crash? We pulled twelve years of DSPX, back to 2014, lined it up against the S&P 500, and ran it through the wringer. The answer is that it does its real job precisely, and it is not a crash detector.
Start with the simplest test. Day to day, DSPX moves in the opposite direction to the S&P about 57% of the time, ticking up when stocks fall, which gives it a faint flavor of fear. But that is a coincidence of the same day, not a forecast of the next one. When you ask what the market does after a high or low DSPX reading, the level tells you nothing.
| DSPX level | Median forward S&P | % of the time up | Chance of a 5% pullback |
|---|---|---|---|
| Lowest fifth (~17) | +1.1% | 67% | 5% |
| Middle (~25) | +2.0% | 71% | 15% |
| Highest fifth (~34) | +1.7% | 66% | 26% |
| Any day (baseline) | +1.5% | 67% | 15% |
Look at the middle two columns. The S&P's forward return is mildly positive at every DSPX level, and it finishes higher about two-thirds of the time whether dispersion is high or low. The next-day numbers are even blunter: the S&P closes up 54% of the time when DSPX is above its median, and 54% when it is below. Identical. DSPX has no say in the market's direction.
Maybe the edge hides at the extremes. It does not. When DSPX is in its top 10% (above 33), the S&P is still up 61% over the next month. When it is in its bottom 10% (below 18), up 64%. There is no DSPX reading, high or low, where the market becomes reliably bullish or reliably bearish. That is exactly what you would expect of a widely watched index: if such a level existed, it would have been arbitraged away long ago.
There is one real pattern in that first table, and it is worth handling carefully. High-DSPX days do carry more forward drawdown risk: a 5% pullback follows 26% of the time from the top fifth of readings, against just 5% from the bottom fifth. At a glance, that looks like detection. It is not. It is the regime. DSPX is elevated precisely when the market is already choppy and correlated stress is in the air, and choppy markets simply have more pullbacks. The dispersion reading is describing the weather you are already in, not forecasting tomorrow's.
The proof is in the worst days themselves. We took the worst 5% of S&P down days and asked where DSPX was.
| When | DSPX percentile |
|---|---|
| Five trading days before the worst days | 66th |
| On the worst days | 68th |
On the market's worst days, DSPX sat around its 68th percentile, moderately elevated, not extreme. And five trading days earlier it was already at its 66th percentile, essentially the same place. It does not ramp up ahead of a bad day, and it barely moves on the day itself, rising about a single point. There is no surge to read, no alarm that goes off in advance. By the time dispersion is elevated, you already knew the tape was nervous.
A high DSPX tells you the market is already in a jumpy, every-stock-for-itself regime, the kind that comes with more pullbacks. It does not tell you a crash is coming. On the day one actually arrives, DSPX looks just like it did a week earlier.
None of this is a knock on the index. DSPX measures expected dispersion, the gap between how much single stocks move and how much the index moves, and it measures it well. That makes it genuinely useful for the job it was built for: dispersion trading, selling index volatility against single-stock volatility when dispersion is rich and the reverse when it is cheap. It is a relative-volatility and correlation gauge. Asking it to call the market's direction or time a crash is asking a thermometer to forecast the weather. The reading is real and useful, just not for that.
Twelve years of DSPX from the Cboe, run against the S&P every way that matters, from direction to drawdowns to streaks.
The S&P closes up 54% of the next days whether DSPX is above or below its median. The level has no say in direction, and forward returns are flat-positive across every band.
Even in DSPX's top or bottom 10%, the market is up 61 to 64% over the next month. No reading turns it reliably bullish or bearish, as an arbitraged index should be.
On the worst S&P days DSPX sat at its 68th percentile, and was already at its 66th five days before. It does not ramp into a drop or spike on it. No advance warning.
DSPX measures expected dispersion precisely, which is exactly what dispersion traders need. It is a relative-volatility tool, not a fear gauge or a market timer.
Why a stress-flavored index still cannot time the market.
An indicator that is elevated during stress is not the same as one that warns of it. DSPX is already up when the tape is rough, but it was up a week earlier too, so there is nothing to act on.
High dispersion comes with more pullbacks because it marks a choppy market, not because it foresees one. The market is still up two-thirds of the time at high DSPX. That is a description of conditions, not a signal.
DSPX is excellent at measuring expected dispersion, which is what dispersion traders need. Stretching it into a crash alarm or a direction call is using a precise instrument for something it was never built to do.
We test popular signals the honest way: every instance counted, every result measured against a plain baseline. See what else held up, and what did not.
Browse Reality Check
Comments
Loading comments…