Volatility-Based Threshold

In previous posts I found that each sector needed to be optimized for lookback period n and threshold X.  Today I am attempting to replace the threshold constant with a volatility level that is automatically adjusted for the industry.

The formula I have been playing around with up until now is as follows:

FMedian("Close(0)",#Industry) / FMedian("Close(n)",#Industry) > X

If you have not been following this blog you should read the last few posts to understand what n and X are.  I found that the optimum value of X varied from sector to sector.  This leads me to believe that a volatility-based threshold may work better than the arbitrarily assigned threshold I have been using.  For today's experiment I decided to use Average True Range for the volatility-based threshold.  Portfolio123 has formulae for ATR(bars,offset) and ATRN(bars,offset).  I decided to use ATRN which is ATR normalized by the last close price and then converted to a percentage value.  I modified my group  formula from above to the following:

FMedian("Close(0)",#Industry)/FMedian("Close(n)",#Industry) >

                    (1+X₁*FMedian("ATRN(X₂,0)",#Industry))

Now you are probably thinking "what the heck is this guy thinking".  He started with one variable X and now has two variables X₁ and X₂.   Well this is true but the idea is to do some testing to see if X₁ and X₂ are robust enough over a wide range of values so that they can be eliminated and replaced by constants.   I did many screener backtests while varying X₁ and X_2.  The screener setup is shown below:

The screener backtest parameters used are shown below:

The results of the backtests are summarized in the EXCEL graph below:

Just a reminder that the screener backtest starts with $100 dollars.  The graph plots the number of dollars at the end of the run for the period 03/01/2002 until 10/29/2008.  From the graph it appears that the results are quite robust and well in the black for all values of X₁ and X₂ that were tried.  I have decided that the longer the period for averaging ATR is best - should give the most robust results.  An X₁ factor of 0.1 appears to suitable and I will work with this value in the future, at least until I repeat some individual sector tests.  So the formula looks like this now:

FMedian("Close(0)",#Industry)/FMedian("Close(n)",#Industry) >

                     (1+0.1*FMedian("ATRN(100,0)",#Industry))

All for now.

Comments

[commodity tips]

To be honest I am just a newbie in the stock market and these graphs and movie usually go over my head but the way you have simplifies and presented is really good…...

Mcx Zinc tips