Seems like a simple thing but I'm left stumped sorta as what to do. When you have your charts show data outside regular trading hours (pre/post market) your moving averages will drastically change. Most of my time I have always had pre/post market data on my charts and thus my moving averages reflect that. But I've been seeing/watching traders who dont do/show pre/post market data and their different moving averages. So I'm wondering which is the "most accurate" or proper reflection that most people use/follow? It seems sometimes one set of moving averages is accurate and another time others are. I know there will probably not be a definitive answer, but I'd like to hear your thoughts. Do you show pre-post market data on your charts and have your moving averages reflect that?
It is a logical dilemma. On one hand there's the fact that it's a continuous market and holding a position overnight does experience the effects of price change. On the other hand, the volume to price relationship is characteristically distinct between RTH and extended hours. One's methodology and trading timeframe would make the difference which to pay attention to and when.
Well that adds the extra complication/twist if you were to consider using Exponential Moving Averages. But the question remains...lagging as they may be...Moving Averages are still used by a great many people as baseline support and resistance points, self fulfilling as they may be. So the question remains...yes or no to outside market hours data? I dunno.
I think you're right (that there isn't a definitive answer) but my suggested answer is to include everything but use constant-volume candles always, rather than timed or tick charts, and the indicators will probably be more reliable that way than any other way (I can't, of course, prove it, but that's my story and I'm sticking to it!).
I'm a systematic trader and so when I'm faced with this type of question I just go and test it. How does my model do with RTH only, how does it do with RTH+extended and, if there are differences, I dig deeper. I stick to RTH for most of my models, but not all of them. I have one that trades an equity on Market A but that is related to what happens in Market B that closes at a different time. So, for that model I always go into the extended trading hours on Market A to match the end of RTH on Market B. This might not be relevant to your particular issue. For what it's worth, I will sometimes define the end of my trading day based on volume and bid/ask spreads.... so on some days the "trading hours" end when the regular session ends and on other days the "trading hours" continue into the extended hours for a certain amount of time. Perhaps this sparks some ideas? I think the "deep" answer is to think about your system and why it has an edge (assuming it does). If you can understand the "why" behind it, you should be able to reason out which data logically makes sense to include. Then I would personally be strongly inclined to test my hypothesis using historical data.
Pick your chart for a FOMC release on any Wednesday at 2. Changes in liquidity affect indicators but don't necessarily invalidate them. The goal isn't to discount some data as irrelevant because of its timing or condition, but understand how that input affects your (lagging) output....and how to use it.
If you are not finding any statistical edge on the Moving Average periods that you are testing whether on the day or night or combined sessions I would suggest just try using VWAP, and as Xela suggested using constant range or volume bars if your charting package facilitates it. I would think there is more money trading off of VWAP then any fixed period MA.