Feature request: deviation for EMA

[dimitriyfomin]

Standard deviation doesn't work well for EMA. It's better to use deviation by the same way as EMA is calculated.

my suggestion...
var emaVariance = function(vector, options, callback) {
// Single numeric argument defining the smoothing period
if (typeof options === 'number') {
options = {
period: options,
ratio: function(n) { return 2 / (n + 1); }
};
}
var variance = vector.slice(0, options.period).toVector().variance(),
ratio = options.ratio(options.period),
emaVariance = [variance].toVector();
emaVariance.divide(options.period / (options.period - 1));
for (var i = 0, len = vector.length - options.period; i < len; i++) {
emaVariance.push(
ratio * Math.pow(vector[i + options.period], 2) + (1 - ratio) * emaVariance[i]
);
}
return result(emaVariance, callback);
};
var emaDev = function(vector, options, callback) {
return result(_(emaVariance(vector, options)).map(function (v) { return Math.sqrt(v) }), callback);
};