Calculation of Average True Range (ATR)

[REKandler]

Hi

How is IG's Average True Range (ATR) figure averaged over the relevant period?

For instance, is it a simple average, a running average, a weighted average ... or something else?

Many thanks
Richard

[MongiIG]

14 hours ago, REKandler said:

Hi

How is IG's Average True Range (ATR) figure averaged over the relevant period?

For instance, is it a simple average, a running average, a weighted average ... or something else?

Many thanks
Richard

Hi @REKandler

 

It is typically derived from the simple moving average of a series of true range indicators.

 

All the best - MongiIG

[Simon75]

Hi MongiIG,

I know about 10 months has passed since this original thread, but seeing as I am greatly interested in the answer to the original question of the poster, I am posting this late comment.

MongiIG, in answering REKandler's question about how IG's ATR figure is averaged over the relevant period, you have stated "It is typically derived from the simple moving average of a series of true range indicators." This arguably in a sense may be true, for some platforms and traders, but this general statement unfortunately sheds no light on the actual question that was asked, which was essentially how does IG Trading arrive at their figure on their platform?

As a qualified mathematician I can confirm to all that the ATR(14) figure that IG gives for any graph on their platform is not a simple moving average (SMA) of the last 14 true range indicators (TRs), as one could construe is your suggestion. This is why I say the user's question has not been answered.

Any user can confirm the following with their own calculations. IG uses the common "smoothing formula" for ATRs, which is that the ATR at time "t" is:

So in layman's language: For a 14 period ATR (i.e. n = 14) the most recent ATR is calculated by multiplying the previous ATR value by 13 and then adding it to the newest True Range (TR) and then dividing the whole result by 14.

The problem is that in using this "smoothing formula", and not a simple SMA of the last 14 TRs, the ATR value can be quite different depending on how far back in one's data that the formula starts being used.

The "ghost" of the artificial initial values you create for both the first TR (which is simply the High minus Low as there is no previous candle) and the first ATR (which is indeed a simple moving average of the last 14 TRs once you have that many), these 'hang around' in the ATR value for quite some time.

It does not appear to be like the formula for Heiken-Ashi Candles, when the effect of the initial candle (calculated also by irregular means) quickly dissipates or becomes vanishingly small after about 10 or so periods (I can't remember the exact estimate that is often given for Heiken-Ashi candles before they become "as they should be", comparable/equivalent to anyone else's figures).

Because of the nature of the above ATR formula, all values you see for an ATR on any web platform, using the same period (i.e. ATR(14)), for the exact same data, such figures may vary considerably, depending on how far back in the data the ATR first starts being calculated. I have heard that some platforms use 250 periods back, to ensure that the effect of the artificial initial values has somewhat been "nullified".

I have no idea how many periods back that the IG trading platform starts to calculate an ATR(14). It can't be infinite, but I can tell you one thing for a certainty that it is far far greater than 14 periods.

If anyone from IG could tell me the exact number of periods used (even if it is 1000+) I would greatly appreciate it.

Regards

Simon