Open up a standard template SD indicator in most any popular trading package and they often look just 14 bars back to calculate a SD. Is it valid to use such a small sample size when calculating SD in trading? Say you're measuring bar range. Wouldn't it be better to use as many bars possible to get a broader sample of the true SD of the bar range? Wouldn't a larger sample provide a better estimate?
You can make the StdDev. to whatever suits you. For example, with moving averages, some employ 14, 20, 28, 50, 200 etc periods or anything.... for that matter. I use Amibroker and it has a feature called 'Parameters' which allows one to put up an indicator on the price chart and then using a small pop up window, slide the period up and down the scale so one can visually see ( curve fit) an indicator to the best apparent period. The shorter the time frame the more noise so one will tend to get stopped in and out more frequently. Your period time frame should suit your trading style, eg day trader or longer term holder.