Dear brethren: I have some questions about moving averages. Which would be the most sound way to optimize a moving average to a single market and time frame? Is there a way to or are we dealing just with cognitive illusions? I mean, whichever "optimal" moving average we use would be subject to cycles of high performance and cycles of continued low, even slightly negative performance, randomly. Otherwise it would appear that there are certain periods which suit best the characteristics of a given market and time frame. Perhaps the only determinant factor would the moving average time lag in comparison with the price action. Then we adjust the periods according to our preferences. And what about simple vs. weighted vs. exponential moving averages? Would the choices be subjective too? cpo
Depends on what you want the moving average to do. Few people understand that a moving average in and of itself has no magical qualities. It is, basically, a moving trendline, that is, it is a line, it indicates trend, and it moves. There is nothing, in other words, that a moving average can do that a trendline can't, so any concerns about weighting or about how to figure exponentiality or whatever are extraneous to the issue of finding something to help you determine the presence or not of trend, the direction, changes in trend, and defining trend reversal. If you want to keep things simple, just use trendlines. If you want to use their slope as a measure of strength of trend, you can do that as well. --Db
"Few people understand that a moving average in and of itself has no magical qualities. It is, basically, a moving trendline, that is, it is a line, it indicates trend, and it moves." Well-said!!!
if prices go up then they go down then they go back up again, how can I use a trendline to determine if on average, prices have gone up or down? Let's just say I'm buying milk, and I want to buy it at or below the average price for the last 20 days. Now I have 3 trend lines, one going up, one going down and one going up. Well, I can add 2 up minus 1 down,...no that won't work. Well, they are lower now than they were 5 days ago, does that answer the question? No? Well, don't bother me with moving averages, because I can figure it all out with just these trendlines. There was never any need to invent moving averages in the first place. It is all just a bunch of new fangled fooey. Ok, I measure the first trendline, and then I measure the second trendline, then I measure the third trend line, then I divide that by 3...what was it you wanted to know? I can do it all with nothing but these little old trendlines. Besides, you should just be able to tell if you are getting a good price. If anyone asks you why you bought so much milk at that price, just tell them, well it seemed low. or just tell them, how should I know? Prices are always moving, what am I supposed to do, keep a moving average?
optimising a moving average is sort of like putting an air foil and heavy duty shocks on a minivan. If you have to optimize it, you might just be trying to use it for something it was never designed to do in the first place.
An uptrend, for example, is defined by higher highs and higher lows. Therefore, whether your prices have gone "up" or "down" depends on the relationship of the ups. If the ups are "stepping", i.e., each is higher than the last, then prices are trending upward. If, however, the most recent "up" is lower than the previous "up", then you may be topping, or, at the very least, getting ready to form a coil. You can apply a moving average to all this if you like, but if you have higher highs, what's the point? And if you aren't, what's the point? --Db
becaue I'm not trying to determine if prices are trending. If I wanted to do that, I would just draw a trendline. That's what trendlines are for. I want to determine if the current price is higher or lower than the average price has been. That's what moving averages tell you. It's not my fault trendlines don't tell you the average price and averages don't tell you if the highs are lower, I didn't invent either trendlines or moving averages. I just try to use the right tool for the job.
And why do you want to know that? In other words, once you know, what do you plan to do with the information? --Db
Knowing the past average price probably has less value to me than studying algebra did in high school. But just because I had no future plans for it's use, it was not declared an irrelevant subject. And I think to this day they still teach it. My life is already simple enough. I am not worried that acquiring useless information will make it too complicated. There is still plenty of empty space in my head, so why not use the idle hours to fill it up with something that may come in handy someday? It's only useless if I never use it. I have no plans for a lot of information. That doesn't mean the information is useless. It just means I have no plans.
I'm afraid you're losing me then. The thread is about trend. If you're not interested in trend but rather in how the current price compares to an average price over a certain number of days, then all you have to do is use an MA that matches however many days you want to include in your average. Or just add them up and divide by the number of days. --Db