# Volatility-Based Envelopes

I am following an article by Mohamed Elsaiid (MFTA) about Volatility-Based Envelopes - a quite new technical indicator he has introduced, that is being used by Bloomberg. My goal is to get a simple and accurate formula to use.

I have encountered a few issues while trying to formalize the ideas presented in the article.

1. In "Step 1" of the calculations, one should calculate the standard deviation over duration of 21 daily percent change values. I was wondering how percent change values are calculated: according to the change from open to close? From low to high? From close of previous day to close of current day?
2. "Step 2" of the calculations is presented with an example. $\mu$ is said to be the simple average of the percent change (I guess it is over the same period that was used in the first step?), $S$ is said to be the last given price (close?), and $d$ is said to be the standard deviation of the percent change (calculated in "step 1"). Then: the lower and upper row VBE are at $S\cdot (1+\mu-2d)$ and $S\cdot (1+\mu+2d)$ respectively. I guess 2 here is a parameter to be played with?
3. "Step 3" is called "Smooth the raw VBE using weighted moving averages". In the article, Mohamed says we will use "centered weighted moving average (CWMA)", but does not explain any further; in general, weighted moving averages should have some weighing function. If I understand correctly, it is a common abuse of notation in the world of technical analysis to say "weighted moving average" when one really means that the weights decrease in arithmetic progression. I also assume that the arithmetic progression factor is 1. Though, even assuming so I am not sure it is clear to me what does he mean by CWMA - is the decreasing arithmetic progression applied to both sides, and to each with the same arithmetic progression factor (1)?" If that so, and the time period in question is $n$, I guess the system needs $n/2$ values from the future. Though, Mohamed later claims that for $n=21$ there are exactly 5 missing data points.
4. "Step 4" is completely unclear to me, so I am not sure I can even ask a proper question about it. The aim of that step is to provide a method of forecasting the missing values (the ones "from the future"). It starts by calculating the correlation coefficients (?) of percentage change over a few different time spans (21, 17, 13, 9, 5 and 2 days) over the recent 63-actual data points (?). Correlation between what? Without doing the forecast first, how can one even calculate the CWMA, of any period? He keeps doing something a bit clearer with the results later, but I think that this stage was not understood enough to keep writing about it.

I'd appreciate any help in trying to understand this interesting indicator.

Reference: The article

-