Can you share how you are calculating the standard deviation for use with the delta throttle? In my model, I show the same values for the ATR, the fast and slow ATR, but my standard deviation calculations are somewhat off which wouldn't have shown a delta throttle for today. Any help that you can provide would be greatly appreciated. Thanks.

To see this working, head to your live site.

5 Comments

Thanks for the feedback. I think it is pretty clear. I clearly state that it is the STD off the Fast MA.

When the Fast ATR Moving Average exceeds the Slow ATR Moving Average by more than 1.5 standard deviations (calculated on the fast moving average)

I'm not really catering to people who want to re-create my dashboard, if they choose too they can, but I'm not writing my about page to really detail every calc in minute detail. It's more to give people an overall sense of what the dashboard is doing. If they want to re-create it they can try and ask for help but that's not my purpose to cater to helping others re-create my work.

Your language would benefit from some clarity on the strategy details page.

There are two ways to potentially calculate the stddev of the average true range, which is what the strategy is described as doing.

Since a single's ATR score is indeed an average of the previous 10 days true ranges, that average has an stddev. Using that calculation one will get some massive deviations that clearly do not fit the model.

The second way, which is not intuitive from your language, is to take individual 10-day ATRs from the past 10 days, and take the stddev of those values. I would refer to this as a second order standard deviation or utilize some other clearer verbage. It is technically the standard deviation of the 10-day SMA of the 10-day ATR that is being calculated here.

The second method seems to recreate the values on the dashboard although I have some small differences, probably since I am using TOS chart studies to do the calculation for me rather than excel.

Yep! Glad you got it. Takes STD of the fast MA to smooth out the STD a bit so its not as jerky.

Ok good that you were able to match the Fast ATR and Slow ATR.

For the 1std I use excel's =stdev.s( ) formula calculated on the

for the past 10 days.Fast ATRThen for 1.5 I just multiple the 1std * 1.5

For 2.0std I juust multiple the 1std * 2

Then I have have an if statement that basically says if it's Fast ATR > Slow ATR + more than 1.5std of the FAST ATR, then throttle 25% and if Fast ATR > Slow ATR + 2 std of the FAST ATR then throttle to 0%.

You might have been adding the STD of the SLow onto the Slow. I am adding the STD of the 10day FAST onto the Slow and comparing to the Fast. A bit of nuance there. It's just what tested better. I think the more reactionary Fast ATR STD gave the system a small marginal boost to be a bit more reactionary.

Let me know if that gets it for ya!