This page looks at correlation between the VIX and S&P500 (SPX) indices. It presents correlation statistics for the entire available history since 1990 and how the relationship changed throughout the years. It also looks at seasonality – correlation by month and weekday.
Relationship between VIX and S&P500
It is well known that the VIX, often nicknamed the fear index, tends to move in the opposite direction to the broad stock market. It typically spikes when stocks fall and declines when stocks rise.
However, this relationship is far from universal. In about one in five trading days, the VIX and S&P500 indices move in the same direction.
On this page we will look at correlation statistics, based on daily changes of the VIX and SPX indices. Obviously, the results are also applicable to S&P500 derivatives such as ES futures or the SPY ETF.
The current version of this page covers data from 2 January 1990 (start of available VIX history) to 21 July 2022.
All-Time VIX-SPX Correlation (1990-2022)
For the entire available history, the correlation between daily percentage changes of the S&P500 and the VIX index has been -0.70. This confirms a strong relationship and tendency to move in opposite directions.
The correlation coefficient is even more negative when we use absolute point changes instead of percentage changes for the VIX: -0.79. (For S&P500 we always use percentage changes, as stock indices generally drift to the upside in the long run, making absolute point changes incomparable).
The table below presents a few other interesting figures.
all days corr_pct corr_pts vix_up spx_up same_dir avg_vix med_vix
8200 -0.70 -0.79 46% 53% 21% 19.60 17.70
The period covers 8200 day-to-day changes (days). Out of these, the VIX index has been up (day-to-day change greater than 0) on 46% days (vix_up), while the S&P500 has been up on 53% days (spx_up).
On 21% of days (same_dir), the two indices moved in the same direction.
The average VIX closing value was 19.60 (avg_vix). The median was 17.70 (med_vix).
VIX in the 1990's vs. Now
When making conclusions from aggregate market statistics over a long period, we must always keep in mind that the environment and operations of the financial markets have changed significantly in the last decades (globalization, technology, and financial innovation are just some of the main factors).
This is particularly important when analyzing the history of the VIX and its derivatives. In the 1990's, the VIX was just an interesting market statistic, known by only a few specialists. Nowadays it regularly makes headlines in mainstream financial media and it is underlying to heavily traded derivatives such as VIX futures (introduced in 2004), options, ETFs/ETNs, and many other instruments.
Therefore, the relationship between the VIX and S&P500 may have changed throughout the years. Below we will look at a chart and then the correlation statistics grouped by individual years.
VIX-SPX Correlation Chart
The chart below shows correlation between the S&P500 (daily percentage changes) and the VIX (point changes) over a rolling window of 252 trading days (which is approximately one year).
The relationship appears to have been weaker in the first years (first half of 1990's), but then it is relatively stationary and oscillates around the -0.80 level.
The sharp downward spike in early 2020 was caused by large market moves at start of the covid pandemic. The day with strongest 252-day rolling correlation was 16 March 2020 with -0.96 (this is also the day with all-time greatest day-to-day VIX spike). We must keep in mind that a small number of outliers can push the correlation coefficient towards the extreme.
On the opposite side, rolling correlation weakened (upward spike) to -0.58 on 17 March 2021, as the previous year's volatility fell out of the rolling window.
VIX-SPX Correlation by Year
year days corr_pct corr_pts vix_up spx_up same_dir avg_vix med_vix 1990 252 -0.53 -0.60 48% 53% 28% 23.09 22.58 1991 252 -0.55 -0.57 45% 49% 26% 18.37 17.43 1992 254 -0.55 -0.56 46% 51% 31% 15.45 15.36 1993 253 -0.52 -0.50 49% 51% 32% 12.69 12.43 1994 252 -0.72 -0.72 50% 52% 25% 13.93 13.86 1995 252 -0.45 -0.45 48% 61% 34% 12.39 12.29 1996 254 -0.69 -0.69 50% 54% 28% 16.44 16.24 1997 251 -0.70 -0.71 51% 54% 21% 22.36 20.95 1998 252 -0.82 -0.86 46% 55% 14% 25.60 23.14 1999 252 -0.80 -0.82 49% 51% 16% 24.37 24.10 2000 252 -0.78 -0.79 48% 47% 18% 23.31 23.24 2001 248 -0.81 -0.82 45% 47% 20% 25.75 24.26 2002 252 -0.81 -0.84 50% 44% 19% 27.29 26.39 2003 252 -0.64 -0.67 46% 54% 25% 21.98 19.86 2004 252 -0.76 -0.76 42% 55% 18% 15.48 15.32 2005 252 -0.83 -0.83 44% 55% 17% 12.81 12.52 2006 251 -0.82 -0.80 47% 56% 25% 12.81 12.00 2007 251 -0.84 -0.87 48% 54% 18% 17.54 16.43 2008 253 -0.84 -0.88 47% 49% 11% 32.69 25.10 2009 252 -0.75 -0.78 41% 55% 19% 31.48 28.57 2010 252 -0.84 -0.83 44% 57% 18% 22.55 21.72 2011 252 -0.86 -0.89 44% 54% 17% 24.20 20.71 2012 250 -0.76 -0.77 49% 52% 23% 17.80 17.52 2013 252 -0.82 -0.83 45% 58% 19% 14.23 13.75 2014 252 -0.85 -0.85 44% 57% 17% 14.18 13.67 2015 252 -0.87 -0.86 48% 47% 13% 16.67 15.31 2016 252 -0.81 -0.83 48% 51% 22% 15.83 14.31 2017 251 -0.74 -0.74 46% 56% 25% 11.09 10.85 2018 251 -0.76 -0.81 44% 52% 20% 16.64 15.49 2019 252 -0.83 -0.85 45% 59% 19% 15.39 14.87 2020 253 -0.72 -0.79 43% 57% 18% 29.25 26.70 2021 251 -0.84 -0.84 44% 56% 17% 19.67 18.69 2022 141 -0.83 -0.85 39% 42% 24% 26.34 26.67
Looking at the last 20 years, the ones with strongest negative correlation between SPX and VIX have been 2007, 2008, and 2011. Interestingly, all these years also have higher average VIX closing value than their preceding years (and 2008 has highest average VIX of all years at 32.69).
On the contrary, the weakest correlation occurred in 2003, 2004, and 2017. Average VIX value in these years was considerably lower than in their preceding years (the average of 2017 is all-time lowest at 11.09).
From this it could seem than the correlation is stronger when VIX is rising / stocks falling, and vice-versa.
However, there are other years with sharp increase in the VIX and weaker correlation (e.g. the first covid year 2020) and also those with lower average VIX and strong correlation (2005, 2014, 2019).
The number of years is too small to drive significant conclusions about the relationship between VIX level and VIX-SPX correlation. Furthermore, in some of the years the "market regime" significantly changed at some point during the year. It might be useful to divide the history into periods of "similar market regime" rather than calendar years, with possible boundaries including summer 2007 (start of financial crisis), spring 2009 (market bottom), March 2020 (covid outbreak) etc.
One conclusion we can make is that correlation between the VIX and S&P500 has been relatively stable in the last 20 or so years, the changes between individual years have been rather small. The correlation continues being strong throughout all different market conditions.
VIX-SPX Correlation by Month
The table below looks at individual months to see whether there is any seasonality in the SPX-VIX correlation.
month days corr_pct corr_pts vix_up spx_up same_dir avg_vix med_vix 1 675 -0.75 -0.79 46% 53% 21% 19.45 18.68 2 632 -0.68 -0.75 45% 53% 23% 19.82 18.95 3 722 -0.68 -0.77 46% 51% 21% 20.45 18.02 4 681 -0.69 -0.75 44% 55% 19% 19.05 17.02 5 698 -0.71 -0.74 46% 53% 21% 18.66 17.19 6 706 -0.74 -0.78 46% 51% 23% 18.65 17.43 7 693 -0.67 -0.74 45% 55% 21% 17.99 17.01 8 708 -0.77 -0.82 46% 53% 21% 19.19 17.10 9 649 -0.75 -0.84 48% 51% 20% 20.44 17.85 10 707 -0.73 -0.84 48% 54% 18% 21.65 18.05 11 653 -0.67 -0.81 46% 56% 24% 20.36 17.65 12 676 -0.67 -0.78 49% 53% 22% 19.51 18.13
One thing that stands out is stronger correlation in the autumn months – the four months when the correlation (using VIX point changes, corr_pts) is stronger than -0.80 are August to November, and the top two are September and October. These are also months with rising or above average VIX value (avg_vix), although this may have been distorted by extremely high VIX values in autumn 2008 (see median in med_vix).
Conversely, there is notable period of relatively lower VIX and slightly weaker correlation in May to July.
Nevertheless, like with individual years, the differences between individual months are relatively small and big portion of them could possibly be attributed to one-off events in individual years rather than seasonality.
VIX-SPX Correlation by Weekday
The following table shows how the correlation varies between weekdays (1=Monday, 5=Friday).
wkday days corr_pct corr_pts vix_up spx_up same_dir avg_vix med_vix 1 1550 -0.68 -0.79 62% 53% 26% 19.87 17.89 2 1681 -0.69 -0.78 45% 51% 21% 19.62 17.86 3 1680 -0.72 -0.81 44% 54% 20% 19.56 17.83 4 1653 -0.76 -0.83 45% 53% 18% 19.54 17.56 5 1636 -0.71 -0.76 37% 54% 22% 19.41 17.41
We can see strongest correlation on Thursdays, followed by Wednesdays. These are also two weekdays with smallest percentage of days when the VIX and S&P500 move in the same direction (same_dir).
This weekly seasonality appears to be affected by the weekend effect on the VIX – the tendency of the VIX to decline towards the close of Friday trading, as option prices reflect the upcoming weekend break. This technicality may dilute the traditional relationship between S&P500 and the VIX, resulting in slightly weaker correlation.
The weekend effect is clearly visible in the percentage of VIX up days (vix_up), where the VIX tends to decrease more often on Fridays (37% days up) and increase on Mondays (62% days up). Monday and Friday are also days when the VIX and SPX move in the same direction most often (26% and 22%, respectively).
Smaller number of days for Monday is due to the fact that US market holidays often fall to Monday. Such holiday weeks may (to some extent) explain the relatively weaker correlation and higher percentage of same direction days on Tuesdays (as the Monday weekend effect end shifts to Tuesday), although no such effect is visible on the percentage of VIX up days (vix_up).
Summary
The correlation between the VIX and S&P500 has been relatively stable in the last 20 years, oscillating around -0.80.
It confirms the tendency of the two indices moving in opposite directions most of the time.
The correlation stays relatively stable throughout different market conditions (bull/bear market, expansion/recession).
Seasonality-wise, the strongest correlation has been observed in the autumn months and the weakest in summer. However, these seasonal differences are very small and may be mostly due to one-off events.
During the week, the relationship between VIX and SPX may be affected by the weekend effect (weaker on Fridays and Mondays), although its effect on the correlation figures is also very small.