as thunderdog noted, SMA handles moves coming in and then out. So if you have a huge spike today day your 12 day SMA will be big and 12 days later when that spike fallas off your SMA will drop further than if you had used EMA
Actually, I agree, but in the sense that a shorter lookback period SMA can give similar performance to a slightly longer EMA. Obviously when keeping the lookback periods the same, the EMA is more responsive to current data.
Any one type of MA will lack a feature or 'ability' that some other MA has. They cancel each other out. One will get you in early at the expense of giving you false moves. The other will keep you out of some moves but will be late to the party on quick market rips. There's no holy grail here. Just pick one type of MA you like for subjective reasons and stick with it. In the end it all evens out. opm8
A Bollinger band system might use a simple moving average because it first calculates an average value then calculates the standard deviation. Since the average value is already known it might be used to calculate the exit signal. I observe a three month trendline often approximates a 40 day moving average curve. Sometimes it is difficult for me to draw a trend line and the 40 day moving average curve appears a good substitute. There are lots of ways to compute moving average values. You can calculate some of them yourself using Microsoft Works or Excel spreadsheets. I am reading about a volume weighted exponential moving average at http://www.gummy-stuff.org/EMA.htm. You might try different formulas and observe the trading results of your modeling studies. Different moving averages might show different results.
I would vote for ema's simply because that is what most people are using and the major use of ma's, imho, is to anticipate what the people who are using them are going to do.