Skip to content

Random Regime Musings

August 4, 2010

It is now well-known that historical volatility is a significant moderator of mean-reversion performance. This is just common sense, and is strongly supported by the research. Another good recap (and great visual) was done by Jeff Pietsch of the cool new site ETF Prophet  http://etfprophet.com/relative-volatility-comparative-mean-reversion-strategy-performance/and formerly of Market Rewind fame. In this post Jeff shows how a 3-year relative ranking of 100-day historical volatility had a substantial impact on a basic RSI2 strategy. Another series of posts that break things down in a more detailed and simple manner that is easy to follow was done by Woodshedder http://ibankcoin.com/woodshedderblog/2010/07/19/part-4-the-secret-ingredient/. The use of shorter or more complex transforms/measurements of volatility can boost separation in mean-reversion returns even more. But that is not the point—in fact, the point is that we are using historical volatility in some way or another to determine what to do right now with our trading.

Now here I am going to connect some dots:

Fact 1: We can measure how 30-day volatility has historically affected a simple mean-reversion strategy by separating the readings into different quartiles or quintiles. These can represent different volatility regimes.

Fact 2: Using this regime information we can actually improve strategy performance out of sample by moderating bet size or exposure. However, there is inherent lag in responding to indicators that measure current historical volatility.

Fact 3: Volatility–lets say 30-day volatility– can be predicted far more reliably than market prices using even simple projection models. The use of volatility projection models (EGARCH etc) is well established in academia and considered to be fairly robust.

Conclusion: Using predicted or projected volatility will substantially aid in regime-based strategy exposure models. In English, by using forward estimates we can respond more quickly to changes in volatility and how they will impact our mean-reversion strategies.

5 Comments leave one →
  1. August 4, 2010 12:27 pm

    Hi David,
    I like this good summary of volatility impclications and the convincing conclusions. The other day I was skimming through Tony Cooper’s winning paper of the 2010 NAAIM conference (http://www.naaim.org/default.aspx) I remember that the subject was on volatility of volatility and how predictable it is. I think it may be interesting to bring this in. Will look into this as well.
    Regards, Michel

    • david varadi permalink*
      August 5, 2010 2:02 am

      Hi Michel, thanks very much and welcome to the blogosphere! some good work so far.
      best
      david

  2. August 5, 2010 8:47 am

    true that volatility is interesting because of its higher predictability, which I guess makes for a nice(r) piece of data for regime switching..
    Thanks for the post, that’s given me some motivation to do some analysis on impact of volatility on Trend Following returns.

    @Quanting Dutchman: cool new blog indeed! Just added you to my blogroll – hope that helps…

  3. Mrkt_Rwnd permalink
    August 5, 2010 8:19 pm

    Hi, I don’t think I wrote that those were the HV parameters. In fact, I didnt’ mention it at all!

    😉

    Rechecking my code, it was in fact an HV of 100 normalized using a 252-period percent rank.

    Cheers, J ~~

Trackbacks

  1. Volatility Distribution Analysis « Quantum Financier

Leave a Reply

Fill in your details below or click an icon to log in:

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

%d bloggers like this: