just common sense
What is better: using a simple moving average or price action?
- Thread starter padutrader
- Start date
I hope you were not referring to me. I was not trying to mock your idea of using moving averages as I admit that certain lengths of moving averages which are widely followed (50 and 200 day on the SPX) and can be useful. I have just found that they don't seem to add as much as support and resistance levels or trendlines. If you're interested in filters, you might want to look up the work of John Ehlers.
http://www.mesasoftware.com/
Not sure his work provides an edge, but it is interesting. Thanks for sharing and best of luck.
THE BIGGEST EDGE IS IN UNDERSTANDING the market
mocking helps us to check our ideas and keeps us honest.
i am only interested in becoming a traderGOD so that those who mock me today will worship me tomorrow
i am finding that whether it is moving averages, bollinger bands or price action it takes a lot of fine tuning to use anything in the chaos that is the market where conditions are continually changing,where all the movements are meant to deceive and direct you the wrong way: it is an experience like no other in the world.
I LIKE TO USE THE BASIC TECHNIQUE; complicating the technique in the hope of making it work, is a waste of time.
And all techniques whether Bollinger price action or moving averages all are different routes to the SAME destination.
IF YOU KEEP CHANGING THE ROUTE TO THE DESTINATION it is just a waste of time though the scenery may be different, some may be slightly shorter than the other
Volume action is even better.
not for normal people
very tricky to read
there is not much i have not tried
i am the only one who has failed in all
Many years ago i ran some tests on moving averages and lost a lot of faith in them. Well, there will always be one moving average that makes money over the last x periods. But which one is it? The 10 period? The 20 period? You could test for which one it was going back, say, 200 periods, and then, see if that is the one that will perform best over the next 50 periods. Then, every 50 periods, repeat the 200 period curve fitting process, and see, once again, if the curve fitted average performs, with a profit, over the next 50. I never finished this test, so i don't know what the results would be. Generally speaking, in a bull market, your moving average longs will profit, and your shorts will lose. So you will always make less than buy-and-hold, but with the peace of mind you will not take any large losses. And visa versa. But then, what happens in choppy markets? Yes, you will almost always lose...unless you know in advance what would have been the one periodicity that would survive the chop, and/or make profit in the chop.
Having lost faith, i have turned to testing price action with some promising results. I am currently testing an always-in approach, either long or short. In a bull market, it too can make less than buy-and-hold...if the market is heading almost straight up. But if it chops on the way up, it's possible to make more than buy-and-hold. Interestingly, tests indicate that it's also possible to make a little on the short side, even as the market is in a bull. Chop is difficult too but i sense that the price action i'm testing will come through chop with a profit a lot more often than a moving average will come through chop.
I have a lot more testing to do, but this is what i find so far.
I am mainly looking for engulfing (trend following), setting stop at last pivot high/low, and adjusting trail stop after almost every new bar. Do not yet have the ability to explain it to a programmer (me), but am managing to get consistently upward moving profit charts when using the replay feature in Tradingview, that is, when applying some rules, and adjusting stops/reverses, without knowing what the next bar will present. The equity curve of success looks more reliable than profit charts generated by successful moving average algorithms. I would say that during a bull, a long profit factor of 1.8 and a short profit factor of 1.2 is probably better than you could expect using moving averages.
If both longs and shorts have positive profit factors, then there is no reason to ever be out of the market, hence, always in. However, it will be interesting, later on, to see if a moving average, or something else (some other algorithm that relies on indicators), can filter for adjustments in trade size.
Hopefully i'll have some more definite details to report later down the road. It's almost time to walk it forward with $100.
%%
OK by me;
but moving averages are used by so many snake oil vendors\buyers beware.
But,cant really let the snake oil vendors goof up a 200dma ...............................+ price
This is like asking which is better screws or nails. Each has their application and drawbacks. The question you should be asking is: Which to use and when and when not.
For instance short term MAs on a choppy day is pure silliness. And to a lesser extent so is PA.
Trending days both have uses.
But the real question you should be asking HOW do I use them. The grade school method is to link the MA or PA signals directly to trade signals. If it were that one dimensional, it would be 1972. MA and PA tell you about the context, but the actual trading is one of many additional layers atop.
For instance short term MAs on a choppy day is pure silliness. And to a lesser extent so is PA.
Trending days both have uses.
But the real question you should be asking HOW do I use them. The grade school method is to link the MA or PA signals directly to trade signals. If it were that one dimensional, it would be 1972. MA and PA tell you about the context, but the actual trading is one of many additional layers atop.
There are entire threads to search on why moving averages do not work, but the main reason is group delay, more commonly called lag. MAs are really lowpass digital filters applied to price data which can be modeled as a complex waveform.
The only true zero-lag MA is a forward-reverse design. H(z)H(-z) in z notation. The problem with these designs is that they are non-casual, meaning they cannot be used in realtime applications. They can only be used in offline analysis. However, they do make the best filters you will ever come across. I personally prefer exponential MAs for this use case.
The only true zero-lag MA is a forward-reverse design. H(z)H(-z) in z notation. The problem with these designs is that they are non-casual, meaning they cannot be used in realtime applications. They can only be used in offline analysis. However, they do make the best filters you will ever come across. I personally prefer exponential MAs for this use case.