Skip to content

Improve Your Trading: Heat Zones Classification System Using Statistics

October 7, 2009

Every asset class has a unique “fingerprint” for how it behaves in different types of scenarios. While market behavior changes over time like everything else, these fingerprints generally remain stable for long periods of time. The key is to understand how a given market behaves when it is in certain zones.  This allows you to clearly formulate what approach you should take at this point in time, rather than react and try to process new information as it arrives. Every financial instrument behaves in essentially 6 different “zones” over the longer term. This can be simplified by stating that the market is either trending up, trending down, or trendless, with either high or low volatility for each of these scenarios. The table below shows the historical  magnitude of returns to the S&P500 (SPY) under each scenario, with the zones being color coded from green (best) to red (worst). Each zone can be delineated based on an intermediate r-square of high, low and close data, an intermediate  slope of the regression line reading, and a 30-day historical volatility ranking using percentiles (>.5 is high, <.5 is low). For my own work I use a special indicator called the DVRAC which stands for R-squared and auto-correlation.  Interestingly enough the performance of different indicators  such as simple moving averages and oscillators is radically different in each of these zones. Thus the static use of any indicator is quite silly. Optimal performance can be achieved by using the right indicator the right way, in the right zone. Ironically for the S&P500, the use of a moving average is most effective in the trendless zone, while buying on small dips is more effective in uptrends, and shorting on small rallies is more effective in down trends except when volatility is very high (market returns are slightly positive with much more risk for a huge rally against shorts). For more research on zones and to obtain the DVRAC email us at


Heat Zone Scenario r-square slope
Green Trend Up Low Volatility >.2 positive
Bright Green Trend Up High Volatility >.2 positive
Dark Yellow No Trend Low Volatility <.2>-.2 flat or slightly negative
Yellow No Trend High Volatility <.2>-.2 flat or slightly positive
Red Trend Down Low Volatility  <-.2 negative
Orange Trend Down High Volatility <-.2 negative
7 Comments leave one →
  1. Bill permalink
    October 8, 2009 12:09 am


    any study on month end buying and selling after 5 or 6 days.

    also give some example or spreadsheet of trades which you just talked..will help laymen like me to grasp what u speak..thanks

  2. October 8, 2009 6:32 am

    This sounds similar to the idea of trading using market regimes (which seems to become quite popular..). I understood the idea was to:
    1- Understand the current market conditions (ie as per your classification)
    2- Define and use which strategy (ie mean reversion, trend following, etc.) works under which condition

    ie you would not necessarily change the indicator parameters but instead adapt what trading strategy you use. You could for example trade the SPY differently in a low volatility uptrend than in a volatile trendless period.

    I have also heard of adaptive indicators that apply some logic to vary based on current market dynamics (volatility, trend, etc). One that comes to mind is the Adapative Moving Average by Kaufman. The formula is “static” (ie you dont need to worry about changing parameters yourself) but its behavior/values are adaptive..

    • david varadi permalink*
      October 8, 2009 10:13 am

      Jez, it is comparable to the concept of trading using market regimes, however this is a simpler and easier breakdown as we are not trying to predict what will work in the current environment and whether the market is trending/mean-reverting. This is more intuitive—ie the chart will appear to be trending up/down or consolidating, and either be volatile or quiet. The ability to objectively measure what zone the market is in, is much easier. Ascertaining market regimes is a complicated task. Nonetheless there is a relationship, and both are useful.

  3. Johnny permalink
    October 16, 2009 6:07 pm

    David- could you elaborate on what you mean by “intermediate r square of high, low, close data”? Intermediate in terms of period length? Or do you calculate 3 r squares and take the median figure? Thanks

    • david varadi permalink*
      October 16, 2009 6:10 pm

      hi johnny i recommend using an r-squared of the last 20-40 days of the highs, do the same for the lows and the same for the close. then take the average r-squared from these 3 figures. if you want you can smooth this result over a couple days as well.


  4. eber terandst permalink
    July 2, 2010 3:21 pm

    How is it possible to have a negative R-squared ? By definition, this is always positive. Please clarify.


  1. SPY: When is the right rime for RSI(2) and MR? « Engineering Returns

Leave a Reply

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

You are commenting using your 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: