VIX Day-of-Week Seasonality

24Oct11

The next few posts will be about VIX (and volatility ETF) seasonality. I’m not the first to cover the subject (VIX & More has been beating the drum I think the longest), but I’m hoping I can add a twist or two.

First up is maybe the starkest VIX seasonality bias: the day of the week (DOW).

In short, the VIX tends to reach its highest point of the week on Monday, before falling throughout the week to end at its lowest point on Friday.

The graph above shows where the VIX has closed on average, Monday through Friday, relative to the average close of that week, since 1990.

Mondays have on average closed 1.0% higher than the weekly average. The VIX has fallen throughout the week, particularly from Thursday to Friday. Fridays have been the “worst” day of the week for the VIX, closing -1.1% lower than the weekly average.

Numbers for the number lovers…

The stats further confirm Monday’s positive bias and the remainder of the week’s negative bias.

The VIX has averaged a 1.9% gain on Mondays with 65% of Mondays positive. Contrast that with Fridays and an average loss of -0.7% with only 39% of Friday’s positive.

How consistent has this observation been?

In the next graph I’ve made the utterly unrealistic assumption that you could directly trade the VIX. Trader #1 (grey) is long only on Mondays (from Friday’s close). Trader #2 (red) is long every other day of the week.

As the rapidly diverging lines indicate, this has been a very consistent observation.

Two notes:

First, replacing “Mondays” with the “first day of the week” and “Fridays” with the “last day of the week” (since some weeks don’t start/end on Monday/Friday) yields more or less the same results. This observation is not actually tied to the day of the week, but rather the beginning and end of the week.

Second, I haven’t shown it in this post, but there’s some evidence that DOW seasonality might be diminishing. It’s not clear yet, but there are some tells. More on this perhaps in a future post.

In my next post I’ll rerun these same tests on ETFs tracking the VIX to understand whether day of the week seasonality in the VIX carries over. More to follow.

Happy Trading,
ms

P.S. Frank at Trading the Odds also did a nice write up on VIX DOW seasonality here.

. . . . .

To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.



9 Responses to “VIX Day-of-Week Seasonality”

  1. 1 cdi

    This is a calendar vs business day effect. It is a flaw in the way the VIX was constructed as compared to the reality of the options market. Interested to see the backtest with futures and/or ETFs, but its unlikely this effect will be replicated.

    • 2 MarketSci

      RE to CDI: you’re stealing my thunder =) michael

      • I am also interested in the futures/ETF backtest. I have heard of this “weekend effect” on VIX’s reading of the number, not necessarily the underlying IV levels but have not seen it tested.

  2. 4 Charles

    Couple more items to note: (1) VIX can’t be purely traded; (2) VIX ETFs tend to not track the VIX that well so that even if this strategy was real, you wouldn’t be able to replicated it in real life. There has been a decent amount of discussion on this (i.e., trading VIX, etc) on the internet.

    Also, after giving a brief thought on this, I thought this phenomenon was more of a function of time-decay in option pricing but I am not an options person so would defer to an expert on the rationale behind this.

    • 5 MarketSci

      Hello Charles – wowzahs…

      I trade volatility ETFs pretty heavily in my own trading and I’ve talked about what drives the price of these ETFs versus the VIX quite a bit on the blog. I’m well aware of the differences between the two.

      The purpose of the “utterly unrealistic” strategy shown above was to show the consistency of the observation in a way that made sense to readers.

      michael

  3. OK – cdi and Charles said what I was going to say. It would be interesting to see if the market’s tendency to take out the weekend days early is exploitable in some way, perhaps by options getting too cheap or too expensive at predictable points in the week.

  4. 7 MarketSci

    Thinking aloud…

    Readers know I rarely attribute root causes to observations I talk about on this blog.

    I’m a numbers guy and I find the whole process of always trying to attribute a reason to be a fool’s errand – in most cases we’re just guessing.

    But this matter of day-of-week seasonality is a little different in that at least part of the root cause is pretty straight-forward: the mismatch between the fact that the VIX is priced according to calendar days, but only calculated on trading days.

    I’m okay with assuming that that explains the negative/positive bias on the last/first day of the week.

    What I’m having a hard time wrapping my head around is the persistent negative bias on Tuesday – Thursday.

    Basically, ALL of the VIX’s gains over the last 20+ years could have been had on the first day of the week. All other days have exhibited a VERY negative bias.

    No point to this comment other than to wonder out loud THAT root cause.

    michael

  5. 8 Streetlife

    Hey Michael, love the blog. I hate to use the old cliche of long time reader, first time poster but….

    I agree with your statement that finding reasons is sometimes a futile endeavor. However in the case of your VIX seasonality study, I think I have a partial answer for you. I’m entirely an option trader, and when reflecting on my own trading style, an explanation potentially arises.

    As you know, the VIX is only a statistic of S&P option activity. In my trading, I tend to start the week off with a pretty heavily weighted purchase of downside protection, and throughout the rest of the week I tend to sell positions against it.

    If the VIX was entirely comprised of my trading and nobody else’s, we would see the VIX rise on mondays and taper off as the week progresses. It’s just proven to be the most cost effective way for me to slowly build into positions. Buying volatility is best early in the week, and selling volatility is better late in the week. It has to do with maximizing weekend Theta decay. Is it possible that some of the major players in the option market are also doing the same thing?

    It’s an extremely crude explanation I know, but who knows… I look forward to more analysis


  1. 1 Monday links: unpalatables facts | Abnormal Returns

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s