What type of smoothing is applied in builtin ADXDMI study?

[olegpc]

What type of smoothing is applied in builtin ADX/DM study in the eSignal platform for formation of DMI and ADX values?

I know, that standard to smoothing DMI and ADX values is exponential smoothing. I have made such initial ADX/DM study wth EFS, however its values differ from the builtin ADX/DM study though the general forms of curves are close enough. My conclusion: or for builtin ADX/DM uses non-exponential smoothing, or smoothing is exponential, but non-standard parameters are used, for example, in expression dPercent = (2.0 / (nLength+1.0)) other parameters are used not 2.0 and 1.0. And whether double smoothing is applied to ADX values in the builtin ADX/DM study?

I would like to find out precisely what smoothing is applied in the builtin ADX/DM study and with what internal parameters.

I shall be very grateful for the help.

[RMartin59]

ADX Computation

I also can not seem to confirm the ADX computations. If no smoothing is applied, the pure daily computations are correct. As soon as any smoothing is applied, the values do not match up. I've tried both simple and exponential of varying lengths, but haven't once had the computations match up.

[ACM]

olegpc
As far as I know eSignal uses the same smoothing implemented by Wilder (and described in his book "New Concepts in Technical Trading Systems") which is based on an accumulation technique rather than a traditional moving average calculation ie
ADX(x) = ((ADX(x-1) * (n - 1)) + DX(x)) / n
where DX = 100*(ABS(DIplus-DIminus)/(DIplus+DIminus)) and n is the period over which the study is calculated.
For the calculations of DIplus and DIminus you may want to refer to Wilder's book
Alex