Hi,

I have a question for the mathematically 'gifted':

J. Welles Wilder's Average True Range calculation as per his book 'New Concepts In Technical Trading Systems' is as follows:

(Let's assume a 14 day period for the calculation)

ATR(Latest) = ( 13 X ATR(Previous) + True Range(Today) ) / 14

Right??? Common knowledge??? Right???

BUT (here's the problem / question):

Remember that the idea of doing it this way (above) was so that you did not have to keep track of 13 or 14 days worth of data every time (day) you did the calculation (when Wilder devised his ATR calculation the most powerful tool available to him at the time was a calculator) i.e. to simplify things on a daily basis it was (is) easier to just take the previous ATR, add the TR for today, and divide by 14

BUT:

The above DOES NOT give the same result as adding up the TR for the previous 14 days and dividing by 14 every time (day) you do the calculation i.e. every subsequent day that you do his original calculation you are compounding rounding errors and over time the errors become quite 'noticeable' (for want of a better description).

My question is this:

Am I right or wrong? If you think I'm wrong - tell me why.

Also - what do you think of this statement:

By performing the calculation the (his) original way i.e. by taking the previous ATR, adding the current TR, and then dividing by 14 - you are in effect applying 'smoothing' to the equation.

Why is any of this important you may ask???

Well - I've spent a good part of the last month 'programming' his systems into one of my trading platforms and it has become very apparent to me that using the 'standard' ATR / Parabolic SAR / True Range 'indicators' supplied with your (my) trading systems (platforms) the results are quite different from his original work (I have questioned Parabolic SAR at a few brokers for the past year or so). I have three different trading platforms (three different brokers - and one of them is an MT4 broker) and not a single one of the results given by any of them are exactly equal to what the results SHOULD be when manually calculating the ATR (for example) on paper (or using Excel) as per his original work. If you think I'm talking nonsense - have a look see - you'll be 'gobsmacked'!!!

Are these 'variations' or 'errors' 'material' in nature i.e. are they big enough to cause losses / bad trades? My (I) jury is still out on this BUT what if you are relying (for example) on the results of Wilder's ADX or DMI as presented by your trading platform and it's wrong??? What if Parabolic SAR is getting you in too early or out too late because the calculation that has been 'thrown' into your trading platform is incorrect??? Believe me - it's possible!!!

Thanks for reading this far!!!

Dale.

I have a question for the mathematically 'gifted':

J. Welles Wilder's Average True Range calculation as per his book 'New Concepts In Technical Trading Systems' is as follows:

(Let's assume a 14 day period for the calculation)

ATR(Latest) = ( 13 X ATR(Previous) + True Range(Today) ) / 14

Right??? Common knowledge??? Right???

BUT (here's the problem / question):

Remember that the idea of doing it this way (above) was so that you did not have to keep track of 13 or 14 days worth of data every time (day) you did the calculation (when Wilder devised his ATR calculation the most powerful tool available to him at the time was a calculator) i.e. to simplify things on a daily basis it was (is) easier to just take the previous ATR, add the TR for today, and divide by 14

BUT:

The above DOES NOT give the same result as adding up the TR for the previous 14 days and dividing by 14 every time (day) you do the calculation i.e. every subsequent day that you do his original calculation you are compounding rounding errors and over time the errors become quite 'noticeable' (for want of a better description).

My question is this:

Am I right or wrong? If you think I'm wrong - tell me why.

Also - what do you think of this statement:

By performing the calculation the (his) original way i.e. by taking the previous ATR, adding the current TR, and then dividing by 14 - you are in effect applying 'smoothing' to the equation.

Why is any of this important you may ask???

Well - I've spent a good part of the last month 'programming' his systems into one of my trading platforms and it has become very apparent to me that using the 'standard' ATR / Parabolic SAR / True Range 'indicators' supplied with your (my) trading systems (platforms) the results are quite different from his original work (I have questioned Parabolic SAR at a few brokers for the past year or so). I have three different trading platforms (three different brokers - and one of them is an MT4 broker) and not a single one of the results given by any of them are exactly equal to what the results SHOULD be when manually calculating the ATR (for example) on paper (or using Excel) as per his original work. If you think I'm talking nonsense - have a look see - you'll be 'gobsmacked'!!!

Are these 'variations' or 'errors' 'material' in nature i.e. are they big enough to cause losses / bad trades? My (I) jury is still out on this BUT what if you are relying (for example) on the results of Wilder's ADX or DMI as presented by your trading platform and it's wrong??? What if Parabolic SAR is getting you in too early or out too late because the calculation that has been 'thrown' into your trading platform is incorrect??? Believe me - it's possible!!!

Thanks for reading this far!!!

Dale.