Nobody knows that. All I can say is that when catching knives its best to wait until they *appear* to have hit some sort of hard ground (vice versa for shorts). When price gets "x" away from my "mean", I don't enter. I enter when the price is coming back towards "x" after being below "x". I don't use a mean in any traditional sense. The market is reverting after it became comfortable after a recent drop/rise. Like I said earlier, I know from statistics beforehand where an entry point needs to be placed *after* I've determined that we've seen a possible bottom.
I also have tough time using volume. I see some relevance to volume on the price axes. I mean price areas where there was cumulative high volume and price areas with cumulative low volume. But as far as PA goes volume is not consistent. R/S support lines can be broken on low volume and on high volume. What is more important is what happens to the cumulative volume after that. I like to see steps. Up accept, Up accept. If I see UP, down accept. Then the UP trend might be over.
Ok. You probably won't believe me, but, after years of trying just about everything one could possibly think of: yesterday's close. Pretty simple, huh? Mike P.S. I'd be happy to challenge the mathematically capable here... it's been a while but I did get through some pretty serious academic bullshit
Your saying an ARMA model is better then a WMA or an EWMA (typical functions in TA software)? But my question is, how can this be the case if the probability distribution or (regime) changes on you? I have back-tested stupid simple WMA strats that perform great on 'select' equities and not so well on others. Why, because its likely these stocks will have a stable distribution unless some fundamental shock or equity curve tells me otherwise. Is that biased test, you bet, but how can an ARMA model perform any better?
Actually, you are correct. One of the best mean levels is the yesterday's close. However, there are better ones.
What about Yesterday's Point of Control (The thickest part of a TPO or Volume at Price profile), The idea is that is where most of the volume was, hence the true value is there.
To be true to the MP teachings. To use POC the day must be balanced. Or use more then one day to achieve the balance. Balance is a bell curve like looking shape where the close was within the tick part . Now if the price moves "to much" from the balanced POC we can start thinking of going against it if...... Or go with it if.....