Invitation to BSD to prove MACD entry/exit alone can work on long time scales

[montmorencyt2w]

Invitation to BSD to prove MACD entry/exit alone can work on long time scales

Background: This is a spin-off from this thread:

http://www.trade2win.com/boards/gen...0-random-entry-perception-17.html#post1148992

That was in reply to my posting here:
http://www.trade2win.com/boards/gen...0-random-entry-perception-16.html#post1148842

Which in turn came from this:

http://www.trade2win.com/boards/gen...0-random-entry-perception-10.html#post1147418

and my post here:
http://www.trade2win.com/boards/gen...0-random-entry-perception-10.html#post1147414

Markus,

How about (probably in another thread) you doing a little test:

See if you can grow some arbitrary starting amount sixfold using only entries and exits based on MACD.

You can choose the starting amount, but it should be a real money account.
Fair enough?

Markus had said:
http://www.trade2win.com/boards/gen...70-random-entry-perception-9.html#post1147318
(partial quotes)

]
Toast says:

Regular/Regurgitated Technical Analysis

You know MACD divergence, MA xovers, Stochastics - all of those common entry techniques often discussed in books on trading despite the fact they don't work.

Course MACD don't work tho according to the knowledgable Toast right !

Good for the guys who grew their money 6 fold that they didn't read Toasts musings.

OK, so that's the approximate background. Roughly speaking, DT says that MACD (and similar) is rubbish. Markus says, oh no it isn't. Here is a fund run by Germany's 2nd largest bank, Commerzbank, that only uses MACD for entries and exits, and it has done very nicely thankyou over many years. I then asked Markus if he could reproduce this success. After all, if it is as simple as he implies (and he is always saying trading is simple), it should be easy to prove, although it might take a long time. I suggested 6 months. At no time did I mention "intraday". Actually, I don't think 6 months would be enough, since from what I can gather, even Commerzbank say you might not get more than one signal a year. Anyway, this was Markus's most recent reply to me (I said I would not post again on that thread on this subject, so here we are):

http://www.trade2win.com/boards/gen...0-random-entry-perception-17.html#post1148992

Lol I sure as heck aren't gonna waste my time trading a long term daily or weekly or whatever it is system to assuage your belief or disbelief in a fund being marketed by Germanys second bank, you have gotta be joking !

And that IS the kinda time frame you need to get a classic trend following system to work.

Intraday one just gets chopped up with such systems, do you really not know that ?!?

No offense, but are there only two people on here who have ever backtested anything, or any more than 4 or 5 who actually make a living from this ?

This thread with all it's nonsense and that is predicated on nothing but a ludicrous lie that Tom Basso fiddled with data on a fact finding mission intended only for him is really the best example of why I'm getting bored more and more with boards.

Anyway, Commerzbanks system was widely reported in the media, and MACD is ALL that drives it.

Lol have you ever heard of the Turtles and their incredibly simple system and the unbelievable money it made them ?

Bill Dunn by his admission, another great trend following CTA with a decades long long history is another trend follower employing all of two parameters driving his fund. (Trader Daily, june / july issue 2006)

Trend following ain't about complex methods lol, it's about catching trends and riding em until they bend.

@Markus,

First of all, just to clear one thing up, I have no axe to grind regarding DT's original thread/posting, and have a completely open mind on that. I have no idea what he was trying to prove really, or really why you took against it so. It would seem there is "history" between you, which is a shame, but nothing to do with me.

I happen to share some of DT's scepticism about TA in general, but it doesn't mean I take his side in all arguments. I don't even understand a lot of what he says to be honest, but I'm still learning this business, I freely admit (for about 2.5 years now). No, I am not making a living trading, but I have made more by trading on the side than my official other income so far this year. It remains to be seen if that can continue.

No I don't backtest. That is not where I choose to put my energies. I think that is for yet another thread though.

Back to MACD and Commerzbank. I never mentioned "intraday"; you did. Of course it needs a long time-frame; that's why I (tentatively) suggested 6 months, but of course longer is better. Sorry, but I didn't read the media coverage of that Commerzbank fund. I am slightly surprised though that you are using one of their advertising pieces to prove a point about trading. By the way, before we go much further, since you appear to reside in Germany, you don't actually work for Commerzbank do you? Maybe you have some inside knowledge of this fund? That would of course alter the whole discussion.

But assuming that you don't, and your only knowledge of this fund is what has been published, then I would repeat: I do not believe that this fund is only traded using MACD. As you say, they are Germany's 2nd largest bank and have all the analysts they need at their disposal. Why would I believe what is published in the media or in an ad? Why would they give away all their secrets? Of course they would not. (Not that size of bank is much to go on; Goldmann Sachs, for example, has lost plenty of money for its clients if Zerohedge is to be believed).

And yes, of course I have heard of the Turtles; we have discussed this in the past. How Richard Dennis eventually blew up and had to stop trading clients' funds. Yes, I know about the Turtles.

But this thread is not about the Turtles, or about Bill Dunn. It is about MACD, and how it and it alone can apparently be used for buy and sell signals over a long time-frame and make money. You claim that this particular fund does exactly that and nothing else. But then you say you would not use just MACD. Why not? If it's good enough for Commerzbank, why isn't it good enough for you?

If it works, it should be demonstrable in actual trading for real money (not back-testing).

I therefore repeat my invitation:

Trade for 6 months (or longer if you wish) with an arbitrary starting amount of real money (as small as you like) using only MACD for entry and exit signals, publishing here signals and trades taken, and running balance. I have suggested using spread-betting because I presume that is the cheapest way of doing it, but if there is another way, fine.


Surely you don't want me to go away thinking that what you said about MACD was just bluster in the heat of an argument, and that you cannot prove what you have asserted?

Well, if you won't do it, then perhaps someone else here who thinks it can be done might?
Feel free as far as I am concerned.

 

[BSD]

My reply here really says it all I gave you yesterday:

I sure as heck aren't gonna waste my time trading a long term daily or weekly system to assuage your belief or disbelief in the track record of a fund being marketed by Germanys second largest bank.

And that IS the kinda time frame you need to get a classic trend following system to work, meaning you'd have to wait the next thirteen years to see if you can achieve the same returns Commerzbank did, and for a long term trend following system a random 6 month period is exactly that, random with no predictive qualities.

Intraday one just gets chopped up with such systems.

No offense, but are there only two people on here who have ever actually backtested anything ?

Anyway, Commerzbanks system was widely reported in the media, and MACD is ALL that drives it.

Have you actually ever heard of the Turtles and their incredibly simple system and the unbelievable money it made them ?

Collectively they made Billions, Dennis and all his offshoot, many of whom are still very successful trend following CTA's to this very day !

Bill Dunn by his admission, another great trend following CTA with a decades long long history is another trend follower employing all of two parameters driving his fund that is up and running to this very day. (Trader Daily, june / july issue 2006)

Trend following ain't about complex methods lol, it's about catching trends and riding em until they bend.

Here is one of Germanys leading finacial magazines that compared several long term investing methods, and here is their finding on a MACD ONLY strategy:

http://www.capital.de/finanzen/100025900.html

In short, since 1989 an annualized 11,2 % p.a.

And that annualized 11.2 since 1989 would have resulted in a compounded gain of 798% with a max drawdown of 12% !!!

The difference between trend following funds - and this and Commerzbank -, is essentially down to nothing but leverage, the bigger you position size the bigger your returns AND drawdowns.

The main reason why most fail at trading / investing is not becaus it's rocket science which very clearly it's not, no, it's because most want a guarantee that they'll make money, they can't live with the uncertainty of trading / investing being nothing but a probability game, as complicated it sure isn't !

I'm not gonna provide proof for you in the form of trading this live the next 13 or 21 years lol mate lol !

You wanna make money you need to expose yourself to risk, and live with handling that and drawdowns and uncertainty and just basically all the things that constitute the hard part about otherwise pretty simple trading.

You have the same info now a guy like Larry Hite had when he had backtested his simple trend following system.

He didn't wait for confirmation by having that forward tested by a buddy the next decades, no, he exposed himself to risk - the ONLY way to get rich - partnered with MAN and because of his success turned MAN into the giant - largest money manager in the world - they are today. (Trader Daily, october 2008)

What do want, a written, notarized guarantee from the great spaghetti monster in the sky that in ten years you'll be rolling ?

Or is it enough for you to accept that you have a probable edge, and then implement, expose yourself to risk, and do what needs to be done to get where you want yourself ?

Can't delegate that if you wanna reach the top, check out the hedge fund thread here now or the wasp saga.

 

[Splitlink]

Statements like that never cease to surprise me. They are mindless assunptions made to influence the opinions of other readers and are not based on any personal knowledge of what the object knows.

This site is full of them.

 

[bevok]

I would actually not mind the challenge if my funds weren't already committed - certainly historically this type of system works in the circumstances mentioned - diverse portfolio, long term, big pot. The results of my testing show trade hold times on average of around seven months, so if anyone's up for it, they'll need to bookmark this page and come back in five years or so! Thats my problem with the system - it tends to tie up capital too much. It would be profitable though as long as there are enough big longterm trends to overcome the chop. Another year like '08 - huge. A few years like '04 - not so good.

 

[DionysusToast]

There is no need to wait 13 years.

If someone wants to provide the rules, I will commit my time to coding them into Tradestation and applying them to any US market stock/futures or forex markets that we deem appropriate.

I can't do European markets as I don't subscribe to the data.

 

[timsk]

There is no need to wait 13 years.

If someone wants to provide the rules, I will commit my time to coding them into Tradestation and applying them to any US market stock/futures or forex markets that we deem appropriate.

I can't do European markets as I don't subscribe to the data.

Hi DT,
Do your skills - or the limitations of TS software - allow you to code and produce tests for histogram divergencies? What I'm thinkling is, if you could generate a test based on that study you kindly produced for me a while back, I think the results are likely to be more favourable/profitable than just taking regular MACD crossovers. The problem here is that you'll be working to produce tests that prove the opposite of what you believe. I have to say, the prospect of it generating a profit tickles me!

Tim.

 

[Black Swan]

There is no need to wait 13 years.

If someone wants to provide the rules, I will commit my time to coding them into Tradestation and applying them to any US market stock/futures or forex markets that we deem appropriate.

I can't do European markets as I don't subscribe to the data.

Dead easy bud...EUR/USD; standard MACD settings, 4 hr TF. As for stop 100 pip, as for take profits presumably we'll simply swing, either way baby, on the MACD x-over... It wont *work* IMO (in as much as you have to have descretion esp. in terms of take profits)...Test time 3 months...

 

[Black Swan]

Hi DT,
Do your skills - or the limitations of TS software - allow you to code and produce tests for histogram divergencies? What I'm thinkling is, if you could generate a test based on that study you kindly produced for me a while back, I think the results are likely to be more favourable/profitable than just taking regular MACD crossovers. The problem here is that you'll be working to produce tests that prove the opposite of what you believe. I have to say, the prospect of it generating a profit tickles me!

Tim.

Nailed on certainty Tim...but IMO it can't be done robotically...you have to still be light on your toes even on 3-4hr TFs..

 

[BSD]

I would actually not mind the challenge if my funds weren't already committed - certainly historically this type of system works in the circumstances mentioned - diverse portfolio, long term, big pot. The results of my testing show trade hold times on average of around seven months, so if anyone's up for it, they'll need to bookmark this page and come back in five years or so! Thats my problem with the system - it tends to tie up capital too much. It would be profitable though as long as there are enough big longterm trends to overcome the chop. Another year like '08 - huge. A few years like '04 - not so good.

That's exactly the thing.

Here is one of Germanys leading finacial magazines that compared several long term investing methods, and here is their finding on a MACD ONLY strategy:

http://www.capital.de/finanzen/100025900.html

In short, since 1989 an annualized 11,2 % p.a.

And that annualized 11.2 since 1989 would have resulted in a compounded gain of 798% with a max drawdown of 12% !!!

The difference between trend following funds - and this and Commerzbank -, is essentially down to nothing but leverage, the bigger you position size the bigger your returns AND drawdowns.

Eg this system had six down - ie negative, minus - years in the 21 year period of it's existence, meaning that depending on when you started trading it your ability to carry on could have been quite different, irrespective of end results that averaged out at 11,2% p.a.

Thats' why trading is first and foremost a mind game and nothing else, as the basics really aren't rocket science.

That's the reason why I would never trade such a system myself.

As an open minded person I am fully able to accept the success that trend trading can provide, but personally I much prefer a trading style that is more sustainable in choppy, ranging markets as well as in trending markets, which buying pullbacks in the direction of a trend on a higher time frame and not strength can provide, in addition to much superior risk / reward ratios through far tighter stops.

 

[Xeno]

Makes me laugh. While the TA doesn't work crowd are posting hundreds of lines on here, most of the few members who do use it exclusively to make money sit back and spend their time actually doing it. And every time they get challenged to 'prove it', they say why should I? And why should they? Put the time and effort in and potentially reveal proprietary stuff that has taken a long time to develop? I really don't think they care whether it's proved or not.

I guess they probably think the chances of every person who has ever said they've done it, here and in books, being a liar is really rather small.

And even if you did put all the time and effort in, you'd just end up arguing about the proof. What would the MACD challenge prove? Nothing, since I imagine we don't know exactly how Commerzbank used it. For all we know, they changed the parameters to MACD every 4 hours based on market conditions.

 

[BSD]

Makes me laugh. While the TA doesn't work crowd are posting hundreds of lines on here, most of the few members who do use it exclusively to make money sit back and spend their time actually doing it.

Haha lovely I couldn't agree more.

Trading really is nothing but a mind game, just like success anywhere else in life.

It's not really the idea or innovation that determines your success, it's having first of all the guts to go ahead while accepting that there can never be a guarantee, and then above all the implementation, doing it well, being able to hold on through the tough times etc.

As William of Ockham very rightly observed in what came to be known as Occams Razor, "All other things being equal, the simplest solution is the best."

And all other things are very equal in trading, but by absolutely no means only there.

Most things in life are really simple enough - if not maybe easy always -, once you succeed in seperating the noise from the few, truly success relevant factors.

It's almost always lack of understanding or a need to stoke a bruised ego that leads to undue complexity.

Ego driven efforts at problem solving only make you want to be right over wanting to be successful.

As amazing as it is, but a clear majority of people prefer being right over having success.

A very bad sign is when people aren't able to explain in a few words what constitutes the relevant essence of what they are on about, what the 20% of their effort that generates 80% of their results is.

Jack Welch said it and did it, Business is simple, undue complexity is almost always nothing than an attempt to camouflage lack of true understanding.

Harvards department of psychology did a famous complexity experiment where two groups of students had to come up with explanations to simple problems. The first group got the correct evaluation from the professors, ie if they had come up with the correct explanation they received a "correct", if not they got a "wrong", while the second group got random evaluations, so that even if they were right they might have received a "wrong" and vice versa.

The first groups solutions were all admirably simple, while the second groups explanations became increasingly complex as they tried desperately to force the inexplicable facts to fit their ever wilder theories.

KISS: Keep It Simple, Stupid !

 

[DionysusToast]

Hi DT,
Do your skills - or the limitations of TS software - allow you to code and produce tests for histogram divergencies? What I'm thinkling is, if you could generate a test based on that study you kindly produced for me a while back, I think the results are likely to be more favourable/profitable than just taking regular MACD crossovers. The problem here is that you'll be working to produce tests that prove the opposite of what you believe. I have to say, the prospect of it generating a profit tickles me!

Tim.

Tim - anything can be coded but the rules need to be clearly defined.

There is actually a divergence function in EasyLanguage but I'm not sure I'd trust it. Divergence with a histogram is probably easier than with an oscillator. You need 2 peaks/troughs in your histogram to compare with price.

Now - see if you can write down on paper what constitutes a peak in the histogram. For instance you might look at a histogram and see 2 clear peaks. Just before the second peak, on the way up, you had a slightly lower histogram bar. Is this now 3 peaks ? We can't just say it's succesive higher bars followed by lower bars - you need to filter out some of the noise potentially. The peaks are not just places where you go from low-high-low histo, it's potentially more subtle and a little more complex. Still it can be coded.

You'd need to clearly define how to find a peak/trough to compare against each other and then what to diverge against and whether that is swing highs/lows or just price at last peak vs price at this peak. I suspect the latter would be inadequate. Potentially you need peaks in the histo vs swings in the price. Probably also need to know how far away from the histo's peak a swing can be for it to count...

Anyway - define the rules & ze code is a easy. I suspect you need DT to help you turn your idea into a ruleset too...

 

[Black Swan]

'kin 'ell DT, regretting I even offered up my two penneth, just fookin do it, or shut the fook up eh? Not rocket science is it...

 

[DionysusToast]

Well BS - I always say, if you can't blind 'em with science - baffle 'em with....

The MACD system as per BSDs spec will be done over the next few days - I think it's actually probably already on my PC - probably one of the canned strategies.

 

[redart]

So you're gonna program and test the system as per BSD spec and see if it actually makes money or not? Cool

Ps - DT how old are you, and how long did it take you to learn to code?

 

[BSD]

I didn't actually include any specs so far so no idea what everybody is talking about or what they want to base their backtests on.

Here is one of Germanys leading finacial magazines that compared several long term investing methods, and here is their finding on a MACD ONLY strategy:

http://www.capital.de/finanzen/100025900.html

In short, since 1989 an annualized 11,2 % p.a.

And that annualized 11.2 since 1989 would have resulted in a compounded gain of 798% with a max drawdown of 12% !!!

But I can say that Capital.de 's results were based on the DAX and off of weekly charts, and they don't say what MACD settings they used, but anyone who wants to find out can contact them.

 

[timsk]

. . . . You'd need to clearly define how to find a peak/trough to compare against each other and then what to diverge against and whether that is swing highs/lows or just price at last peak vs price at this peak. I suspect the latter would be inadequate. Potentially you need peaks in the histo vs swings in the price. Probably also need to know how far away from the histo's peak a swing can be for it to count...

Anyway - define the rules & ze code is a easy. I suspect you need DT to help you turn your idea into a ruleset too...

Hi DT,
I take your point!
That's why I'm in Black Swan's camp and I'm a discretionary trader. I can see that producing a ruleset is more complex than it may first appear to peeps like me who know little about mechanical back testing.

I also agree with zeno's point that the exercise probably won't prove anything very useful or concrete. However, as a bit of fun, it might be interesting?
Tim.

 

[MrGecko]

Ok, how about this*

Daily Timeframe, all asset classes (though I guess you would have to test them seperately?).

ENTRY = 10 SMA crosses above 100 SMA - reverse for shorts

STOP = For longs, below the lowest low over the last 10 bars - reverse for shorts

EXIT = 50 SMA crosses below 100 SMA

I have no experience of system design but I guess MACD are just Moving averages anyway - that is, you could explain the above system in terms of MACD crossovers as well.

Run it over 10 years of data, say the standard 1% position size per trade for the sake of argument.

 

[BSD]

Btw:

"Capital hat zusammen mit der HypoVereinsbank elf bekannte Methoden für den deutschen Aktienmarkt über 20 Jahre getestet, Rendite und Risiko bewertet. Ergebnis: Alle elf Methoden schlagen den Dax."
http://www.capital.de/finanzen/100025900.html

CAPITAL.de only reported the findings, the actual tests of the 11 methods described were done by this bank:

http://www.hypovereinsbank.de

So should contact the Hypo bank for details, not the magazine.

 

[DionysusToast]

So you're gonna program and test the system as per BSD spec and see if it actually makes money or not? Cool

Ps - DT how old are you, and how long did it take you to learn to code?

I'm 40 and I learnt to code at 13 - it didn't take long but then the computers didn't do much then either !

Do we have a spec or not ???