Forex Factory (https://www.forexfactory.com/forum.php)
-   Trading Discussion (https://www.forexfactory.com/forumdisplay.php?f=11)
-   -   Why exactly do EMA crossover systems not work? (https://www.forexfactory.com/showthread.php?t=17340)

aicccia Feb 9, 2007 4:04pm | Post# 1

Why exactly do EMA crossover systems not work?
 
Is it because ema's inherently lag and this kills them in ranging markets?

FXEZ Feb 9, 2007 4:18pm | Post# 2

Is it because ema's inherently lag and this kills them in ranging markets?
Look at the question from another angle. For an EMA crossover system to "work" or in other words to have a long term profitable expectation certain conditions need to exist. The current price vs. the EMA would need to be an effective predictive tool for future price direction. You really have two variables - the current price and an averaged historical price that are being compared with each other. When they "cross over" this is supposed to have some directional predictive value. However if you study it out price tends to cross over historic levels more than it tends to expand away from historic levels. This leads to multiple false signals (losses) in advance of the big profitable "trend" that is captured by the EMA cross.

Price has the tendency to revert to the mean, or in other words to return to its historic price levels or its averages. This is why so many whipsaws (losses) occur, because price is reverting more often than it is extending away from the (mean) average. I would suggest that you would have more luck with a system that looks for extensions from the average greater than the norm (as measured statistically), and counts on reversions to the mean or on continued extension away from the mean. But these strategies aren't without pitfalls either as trends do tend to persist and it isn't possible to know how far it will go ahead of time. I'm not using a method like these though I have explored them in some detail a while back.

One other thing that I ran upon a while ago has to do with the nature of averages with relation to price series. I read (I think it was in some engineering literature) that averages work best on price series that are relativey stable, and don't tend to be a very good fit for dynamic price series, in which category price data tends to fall.

pipple Feb 9, 2007 4:19pm | Post# 3

Is it because ema's inherently lag and this kills them in ranging markets?
\

i dont know, but they work for me, i think you need to anticipate the crossover, and just watch it closely.

i have faith in them, thats just my opinion.

im sure some negative schmuck will post that i am crazy, but they do work
as far as the anticipation, i dont see that as a negative thing, because with any system or trade, you are anticipating to an extent anyway.
i mean we do take risks in this market right
when a cross looks like it is about to happen, i will take the trade if it is withing enough cushion of a support and resistance.
sure they are lagging, if you wait for them

i think watching them closely and figuring where they will cross is the best way, sure ive gotten burned befor this way, but it has rewarded me way way more than it has hurt me.

mwilkinw Feb 9, 2007 4:30pm | Post# 4

\

i dont know, but they work for me, i think you need to anticipate the crossover, and just watch it closely.

i have faith in them, thats just my opinion.

im sure some negative schmuck will post that i am crazy, but they do work
as far as the anticipation, i dont see that as a negative thing, because with any system or trade, you are anticipating to an extent anyway.
i mean we do take risks in this market right
when a cross looks like it is about to happen, i will take the trade if it is withing enough cushion of a support and resistance.
sure they are lagging, if you wait for them

i think watching them closely and figuring where they will cross is the best way, sure ive gotten burned befor this way, but it has rewarded me way way more than it has hurt me.

Hey Pipple! Don't worry I'm not the negative schmuck!! As a mathematician.. I can't deny MA's in general. I do think that stand alone, they are a bit overratted. However, some time series are better than others as far determining support or resistance in a trend. I think coupled with a momentum indicator and price action you can increase your winning percentages.

Coupling those you have: lagging indicators and one real-time indicator (price action). This will increase your chances of predictibility.

But Money Management is where it's at!

pipple Feb 9, 2007 4:39pm | Post# 5

Hey Pipple! Don't worry I'm not the negative schmuck!! As a mathematician.. I can't deny MA's in general. I do think that stand alone, they are a bit overratted. However, some time series are better than others as far determining support or resistance in a trend. I think coupled with a momentum indicator and price action you can increase your winning percentages.

Coupling those you have: lagging indicators and one real-time indicator (price action). This will increase your chances of predictibility.

But Money Management is where it's at!

you just hit the nail on the head for me.
my "system" if you can call it that, includes a realtime indicator to verify the validity of my crosses.
i love trading this way
and absolutely, MM is key
glad to see someone agrees
good luck to you have a nice weekend

StockKJay Feb 9, 2007 4:40pm | Post# 6

Just dont use MA crossovers by themselves. I really like MAs, I dont know why people are always saying they do not work. If they get you into the market, it is all about MM from there on. I am sure there are other technicals that work better, I just have yet to find them.

Craig Feb 9, 2007 4:47pm | Post# 7

They don't work because past averages have nothing to do with future averages.

pipple Feb 9, 2007 4:51pm | Post# 8

They don't work because past averages have nothing to do with future averages.
i dont know
i have built a business based on them, i trade for all my clients based on them as well as a real time indicator or two
they work for me
maybe not for others
i think before you write them off maybe experiment more with them because im telling you, they work man

and theyre easy to use

StockKJay Feb 9, 2007 5:23pm | Post# 9

They don't work because past averages have nothing to do with future averages.
I think you are biased because you have worked, discovered, and eliminated them as a tool. That isnt a negative remark, it is just the reality of someone who has been around the markets longer like you may have been. The truth is though, and tell me if you agree, just because something "doesnt work" doesnt mean it cant be used to make profits. It is only after the trader trades for a longer bit of time he will realize that it wasnt ever the MAs at all that made him profit. Instead it was likely insightful interpretation and anticipation along with good MM.

I use MA crosses and make good returns. They may be my training wheels, and I will give them up in a year or so. Though why give up on something that is helping to make profits.

EDIT: You say that they do not work because they have nothing to do with the future... do you have a tool that tells 80% of the time what the future will do?

Mr Trend Feb 9, 2007 5:44pm | Post# 10

do you have a tool that tells 80% of the time what the future will do?
I do. My cat.

jatotek Feb 9, 2007 5:45pm | Post# 11

price cross the EMA
 
I usuallly pay more mind to the Price moving past a ema versus watching 2 ema's cross. Like look at the eur/jpy with a 50 ema on the daily and the price. Starts to show its personality. As of late the price likes to visit the 50 ema just to say hello and check to see if its a bull market. So I pay more mind to price crossing a ema than 2 ema's crossing.

Price is the only thing that is truely exponential on the chart.

fritz Feb 9, 2007 5:46pm | Post# 12

The very idea that something doesn't work as a black and white statement is often times blown to hell. Life is subjective, trading is subjective, if the subject/the trader can make it work for them that's all that matters. If something doesn't work for you as a trader it doesn't mean it can't work for others, this is like saying that nobody can eat peanuts because they are deadly allergic to some people but several billion other people have no problem ingesting peanuts. Human Beings Are Not Generic, Everyone Is Different, If We Were All The Same Life Would Be One Big Boring Place, Diversity Is A Truism, Always Has Been And Always Will Be.

mr.marketz Feb 9, 2007 6:26pm | Post# 13

I'm with Pipple on this one
 
MA's work - I don't use them personally, but they work. Why would any one say they don't? Is it because they put on a trade too big in leveraged size and leave a 15 pip stoploss... just to watch the price retract and take them out.

Don't enter the cross that way. If you see a cross occur try to wait for a retraction to the slower MA. If your a greedy and impatient 'F-er' like myself, then buy 1/2 of your intended contract size after the cross and double up on the unrealistic 15 pip stop - to 30 maybe. If it goes your way then great. If it ends up retracting to the slower MA buy the other 1/2 there and place your stop at the same level as the first SL. If it goes against you again then you lose money... simple. I would like to post some examples for this wordy explanation, but I'm not going to. Everything works if you find a way to make it work.

MM

Turveyd Feb 9, 2007 6:45pm | Post# 14

MA's work for sure, try a 5min chart, 288SMA (24hour) GBBUSD, notice how twice the price has bounced lower from this MA.

I also recently using a EMA 4 , 9 cross on a 1min chart, when I think it's Long time I don't go long till this crosses up, reduces my impatient trades that then get stopped out.

Ofcourse wip sawed totally, using ema 4,9 as a system on it's own.

Thats why I also use :-

1min, EMA 4, EMA 9, SMA 30
2min, SMA 288, SMA 144, SMA 24

To base my trades around.

Craig Feb 9, 2007 7:36pm | Post# 15

I think you are biased because you have worked, discovered, and eliminated them as a tool. That isnt a negative remark, it is just the reality of someone who has been around the markets longer like you may have been. The truth is though, and tell me if you agree, just because something "doesnt work" doesnt mean it cant be used to make profits. It is only after the trader trades for a longer bit of time he will realize that it wasnt ever the MAs at all that made him profit. Instead it was likely insightful interpretation and anticipation along with good MM.

I use MA crosses and make good returns. They may be my training wheels, and I will give them up in a year or so. Though why give up on something that is helping to make profits.

EDIT: You say that they do not work because they have nothing to do with the future... do you have a tool that tells 80% of the time what the future will do?
I think that the people who are making money with MA's are probably using good money mangement.

Maybe I should clarify my point a bit... There is no doubt that sometimes a MA system will catch a big trend and everybody is happy, but in my experence there is just too much chop inbetween, only MM will see you though that. It seems that to use MA's successfully you need to also be aware of support & resistance etc. For me, I got to a point and thought 'why bother with the MA stuff (and indicators in general), just look at support & resistance and use good MM' they just didn't seem to add anything to my strike rate. What I tried to do as I was learning (still learning), was to eliminate what was not working, even if everybody else was going in a different direction. In backtesting, MA systems proved worse than random due to top & bottom picking.

Now before everybody starts tearing off my head, this is what works for me, I'm not trying to state my opinion as fact. I have read here and elsewhere that people use MA cross systems for managing other peoples money, so good for them, I'm all ears if somebody has a good system because when it comes down to it I would rather let an EA do the work!

By drawing support & trendlines on charts I have a hit rate of about 70%, I just could not get this with any form of indicator based system, so I thought 'why bother...'

So to answer your question, they don't work for me, I simply think the forex market is simply to volatile for any form of mathamatical smoothing, and in my opinion a low freqency past does not tell you anything about market structure (this is the important thing). BUT this does not mean that people can't make them profitable...I just can't see how it would work in the long run.

Bemac Feb 9, 2007 8:38pm | Post# 16

Just my opinion.

I'm not sure I like the idea of calculating an MA based on 100% of the available historical data.

Moods change {Market Sentiment}. Especially if you are looking at a Daily with 3,000 bars on it. Do you really think that what was going on 10 years ago is actually being considered today? {Now THAT'S a Lagging Indicator}

Which is why I would rate MA's in this order... 1) being the better of the 3)
1) WMA
2) SMA
3) EMA

But it's only my opinion.

StockKJay Feb 9, 2007 8:41pm | Post# 17

I think that the people who are making money with MA's are probably using good money mangement.

Maybe I should clarify my point a bit... There is no doubt that sometimes a MA system will catch a big trend and everybody is happy, but in my experence there is just too much chop inbetween, only MM will see you though that. It seems that to use MA's successfully you need to also be aware of support & resistance etc. For me, I got to a point and thought 'why bother with the MA stuff (and indicators in general), just look at support & resistance and use good MM' they just didn't seem to add anything to my strike rate. What I tried to do as I was learning (still learning), was to eliminate what was not working, even if everybody else was going in a different direction. In backtesting, MA systems proved worse than random due to top & bottom picking.

Now before everybody starts tearing off my head, this is what works for me, I'm not trying to state my opinion as fact. I have read here and elsewhere that people use MA cross systems for managing other peoples money, so good for them, I'm all ears if somebody has a good system because when it comes down to it I would rather let an EA do the work!

By drawing support & trendlines on charts I have a hit rate of about 70%, I just could not get this with any form of indicator based system, so I thought 'why bother...'

So to answer your question, they don't work for me, I simply think the forex market is simply to volatile for any form of mathamatical smoothing, and in my opinion a low freqency past does not tell you anything about market structure (this is the important thing). BUT this does not mean that people can't make them profitable...I just can't see how it would work in the long run.
Thank you for your response. I do agree with you about MM. I would say the MAs I use, if used alone, would only give about a 50% profitable rate. It is my job as a trader to make that into something that will work consistantly.

Good trading!

StockKJay Feb 9, 2007 8:43pm | Post# 18

I do. My cat.
LOL! I know someone that has his grand daughter (is 3) tell him what the market is doing. "Diane, which way is the market going just right here?" - "It is going up PAPA" He reaches his 40 pip TP everytime :

Darkstar Feb 9, 2007 11:07pm | Post# 19

The more interesting question is, why "Should" EMAs work?

Keys Feb 9, 2007 11:27pm | Post# 20

The reason crossovers don't always work is because you are only looking at 2 or 3 MAs. What if the MAs you are looking at signal a buy and another trader using different MAs sees a sell? One of you will be wrong. I find it more useful to look at many MAs like GuppyMMA and look for the squeeze, not the cross over. When there is a MMA squeeze, many traders looking at different MAs are all on the same page.


© Forex Factory