Average True Range Analysis

Technical Analysis – Average True Range

It’s no secret that, personally, I’m not the biggest fan of technical indicators. That being said, while testing Equities Lab most recent release I seem to have stumbled across an operator that consistently improves the performance of every screen that I use it with.

What exactly is the Average True Range indicator?

previous_period_atr-oct_3_11_05_47You start by finding the average for the day by calculating (daily high – daily low). You find this number for the past 14 days, or 14 periods if you’re basing your analysis on a different time, and you take the average. This becomes the Previous Period Average True Range.
current_true_range-oct_3_11_08_08You then need to simply calculate the high-low for the most recent day in your analysis.

plot_average_true_range-oct_3_11_08_48Now for the magic. At this point you need to take the previous time period and multiply it by the number of days in that period. This ensures that you will always have an updated previous average true range that is updated to one day ago. Once you have that number you simply add today’s range and divide that total number by 14. This will give you the current average true range.

So, when do you buy?

buy-oct_3_11_11_58When looking at the Average True Range you want companies that aren’t extremely volatile. With volatility comes unpredictability. Instead, we want to find companies that are trading on the lower end of a price range that they are statistically likely to stay within. By only purchasing companies with an Average True Range of less than 1, we are effectively cutting out any and all companies that are trading in the top end which may not be able to increase in price much higher.

How well does it work?

page-oct_3_01_37_26Here is a basic screen where we find every company with a share price greater than $40. I went ahead and backtested this “strategy” over the past 16 years and here are the results. This screen did actually beat the overall market since 2000, but not by much.  

page-oct_3_01_38_23Here we’ve backtested the same screen with the added “Average True Range” trading model, and what a difference that trading model makes. We aren’t beating the S&P 500 yet, but we went from keeping up with the S&P 500 at a higher standard deviation, to completely crushing the overall market – returning over 14% annually since 2000. What’s even better is that since this model is based on volatility we are able to drop the risk profile of the strategy overall and trade at a standard deviation of 3.80% compared to the S&P’s 4.65% for the same time period.

What’s even more baffling is that this trading model seems to work across the board. In most every screen I added the model to the returns increased, even if only by a small amount.

Overall, I’d test to see how this along with other volatility models can improve your primary investment strategy.

About tyler information

More Stories

3 Comments

  1. tyler Eric Darnel says:

    This is very intriguing. I too am sceptical of indicators based on price history. However, this increased the annual return of my favorite backtest strategy by 2% and lowered the standard deviation by 1%. Now, to wonder if this is an artifact of data mining. It merits to consider the inverse case of changing your criteria to ‘average true range > 1’ and seeing if the the result is also inverted. Yes! that condition violently lowered the annualised return. Hmm..this is worth thinking about.

  2. tyler Eric Darnel says:

    This effect, while remarkable, seems to have been arbitraged away in recent years.
    I did a variation screening for the <1% rank of average true range and notice something remarkable. The returns go off the chart starting Dec 4, 2013. Then shortly thereafter, starting March 19th, 2014, the returns are asymptotically arbitraged way until today with no apparent return. Perhaps this average true range observation is an interesting case of stock market forensics. Equities Labs seems to say that, 'a few folks noticed something made a lot of money quickly, then no so much'. I only wonder, is there term persistence of this behaviour?

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.