Suppose you have a 9 day moving average in a daily chart, you average over nine days of which one is today, then 4 days are past and 4 days are yet to come. So you cannot know the 9-day moving average of today until 4 days from now. Hence the 9-day moving average can only be drawn into the chart up to 4 days ago. All those nice chart tools ignore this, and put the data point on today instead of 5 days ago. You can actually see this easily in the chart, as the line clearly lags behind the un-averaged data.
Conclusion, a 9-day average lags behind 4 days, and a 5 days average lags behind 2 days. Therefor, you can't compare them meaningfully unless you shift them to the right place in time (so 4 days to the left for the 9-day average, and 2 days to the left for the 5-day average).
I looked if you can shift the moving average lines to their proper place in bitcoinwisdom, but didn't find such option, unfortunately.
That's an objection to the use of averages in TA I've heard several times, and it's valid in principle, but:
a) It assumes you are looking at it from a signal processing perspective, where using averages like this way would simply be wrong. Instead, chartists know (intuitively at least, I believe) what is going on, and accept that the m day average from days n-m to n plotted to day n is a
lagging indicator, and then try to work around this inherent lag.
Do you think TA should be time invariant? I.e. should it matter if we are talking about now, two months ago, or next month? Should it matter if things develop at half or double the speed? Should it matter if things go in reverse (that the principles involved in a decent in price are basically the same as the ones involved in an ascent)?
