More Stats on the Indicator that Beats Buy and Hold Returns

This excerpt is for educational purposes only and is not to be interpreted as trading or investment advice.  See Terms Of Use here.

Last week I posted a video on a Rate of Change indicator (ROC) that showed a lot of promise in timing the overall market, on an Index like the S&P 500.

Ideas like this are typically used as an “Index Filter“,  where a trader might stay out of the market unless the overall Index meets the criteria he or she prefers.  And that is how we tested it – by buying a basket of random stocks each time the Index met that certain criteria.  We tested 1000 of these random portfolios (more commonly called Monte Carlo testing), so we could all see the range of returns, and the most common returns.

Well that article certainly generated a lot of discussion – some were excited by it, some were appalled by it, others simply looked on with interest.  I had the benefit of speaking with a trader and statistician in Canada, who recommended a few things when testing, and a book by Robert Pardo called “The Evaluation and Optimisation of Trading Systems”.  Here is what came out of it:

  • A trading system should be tested over at least 30 signals where possible

The test I conducted last week took only four signals – so we’re certainly a bit off the recommended 30 there.  I have extended the test time window in the new stats below.

In the book, Robert Pardo also mentions that with longer term trading systems, sometimes 30 signals may not be possible.  Under those circumstances, we can:

  • Test different markets
  • Test different timeframes on those markets

He also talks about degrees of freedom, meaning we need a large enough sample size to accommodate the time it takes to generate one signal.  But more on that later.

The great thing about having this system coded in Amibroker already was that I could easily extend the test timeframe, and easily change the test market as well.  So extending the tests of the ROC Index timer, I’ve included:

  • The S&P 500 from 1995 to 2015 (20 years)
  • The Russell 3000 from 1995 to 2015 (20 years)
  • The All Ordinaries, Australia, from  1995 to 2015 (20 years)

Each different market took eight signals over the 20 years, giving us 24 signals in total.  Much closer to the recommended 30, but overall still not quite there.  So, as with anything, take them with the appropriate grain of salt.

Here are the results below, using the Index ROC Indicator and 1000 random stocks each time via a Monte Carlo test (click the images to enlarge):

ROC Indicator S&P 500 Timing Returns:

Timeframe: 1995 – 2015 (8 signals)

  • 16% – 17% were the most common annual returns
  • With a range of 13% to 22% per year
  • 30% – 40% was the most common drawdown range

1995-2015_SP500 ROC Index  ROC Indicator Russell 3000 Timing Returns:

Timeframe: 1995 – 2015 (8 signals)

  • 15% was the most common annual return
  • With a range from 12% to 22% per year
  • 30% – 40% was the most common drawdown range

1995-2015_Russell3000 ROC Index

ROC Indicator All Ordinaries Australia Timing Returns:

Timeframe: 1995 – 2015 (8 signals)

  • 17% was the most common annual return
  • With a range from 12% to 22% per year
  • 35% – 70% was the most common drawdown range

1995-2015_All Ords_ROC Index

If you want to compare this against a benchmark, I’ve done the Monte Carlo results for a simple Buy and Hold without the Indicator as well.  It is also very interesting to compare the equity curves with, and without the indicator.  As you can see, using the indicator smooths the returns much more than a buy and hold approach.

Click the pictures for each index to enlarge!

Buy and Hold Monte Carlo tests, 1995 – 2015:

1995-2015_SP500 BnH        1995-2015_Russell 3000 BnH

1995-2015_All Ords BnH

I hope you’ve enjoyed this article.  Please leave a comment below.

Happy trending!

– Dave McLachlan

Other Market Research articles:

  1. Stock Market Research: The REAL Effect of “Buy and Hold”
  2. This Simple Indicator SMASHES “Buy and Hold” Returns!
  3. More Stats on the Indicator that Beats Buy and Hold Returns
  4. New Mean Reversion Trading System (Stocks Over 50 day MA) Tested in Amibroker
  5. How This Random Entry Beat The Market (The Tom Basso Coin Flip Proven & Explained)
  6. Does the “Golden Cross” Outperform Buy and Hold? Market Timing, Real Results
  7. The Trading System from “Reminiscences of a Stock Operator”, Tested Over 116 Years
  8. The Three Billion Dollar Day Trading System Revealed and Tested

December 13, 2015  Tags: , , , , , , , ,   Posted in: Stock Market Research

Leave a Reply