next »

[draco]

I thought it might be interesting to start a topic on Ichimoku mechanical trading systems. I'm aware that everyone here probably considers Ichimoku somewhat of a black art that does not translate easily into a fixed set of rules. Nevertheless, I think it's worth taking a look, if only to see how well IKH performs in relation to other MTS. I think we will never get more than a small subset of IKH into a program, but it would be nice to see how even a "lite" version of IKH performs.

I found a very simple (simplistic) Ichimoku expert advisor for Metatrader 4. It worked well on backtesting, and now I am testing it with live data on the GBPUSD daily chart.

Let me know what you think.

- Draco

[Kumo Trader]

I thought it might be interesting to start a topic on Ichimoku mechanical trading systems. I'm aware that everyone here probably considers Ichimoku somewhat of a black art that does not translate easily into a fixed set of rules. Nevertheless, I think it's worth taking a look, if only to see how well IKH performs in relation to other MTS. I think we will never get more than a small subset of IKH into a program, but it would be nice to see how even a "lite" version of IKH performs.

I found a very simple (simplistic) Ichimoku expert advisor for Metatrader 4. It worked well on backtesting, and now I am testing it with live data on the GBPUSD daily chart.

Let me know what you think.

- Draco

Draco,

I personally do not agree with using Ichimoku Kinko Hyo in a mechanical trading system.  The reason?  Ichimoku was designed to be used "at a glance" from an "entire chart" perspective and is not something we can decompose into simple boolean logic. This means that, while I may have a bullish tenkan sen/kijun sen cross taking place, price may still be trading under the kumo and the chikou span may still be under the price curve.  Would this be a good Long trade?  At a glance, most likely not.  Would a mechanical trading system know this?  Most likely not.

It is conceivable that we could code a mechanical system to perhaps take into account those macro level structures, but then we have the finer details like angle of the kumo, flattening nature of the kijun sen, flat top/bottom kumos, kumo shadows, and myriad other nuances that we, as humans can add into our equation of whether or not to take a trade.  Coding something this complex would require true artificial intelligence.

We also have multiple time frames to consider.  While we can run our mechanical trading system on a single time frame it knows nothing of what is going on the other time frames so it cannot make the same rational choices we do.  It doesn't see that we are trading just below the kumo on the Daily chart - it only sees a Bull setup (in its most crude form) on the 4H and thus would execute a Long trade that we as discerning traders would not since we use MTF before entering our trades.

Be advised as well that backtesting on MetaTrader is of dubious quality, so take what results you get with a grain of salt.  MetaTrader uses extrapolative algorithms that, rather than representing actual price action, provide fuzzy equivalents so this data is not be trusted 100%. 

By trying to apply IKH to a mechanical trading system, you really aren't testing IKH, but rather something different that bears little resemblance to the original, integrated charting system.

- Ian

[GeorgeW]

Hello Draco,

From one side...I agree with Ian.
But from the other... hey, if it makes pips, who cares, put it on plus trade manually.
I know some guys who say they use EA to open trades (when they're not watching), and they use manual to continue/close them. I think this kind of combination is a great idea!

Please, let us know how does it perform. It can always be improved!

Could you actually post this EA here? Or send it to me? I would also like to try it. I know some programmers, if they have time (and willingness) they will be able to look at it and maybe make it better?

Anyway, 2 heads better than 1.

BR

George

[draco]

Ok, that being the case, no problem, let's drop the idea.

George, I will mail you the EA separately, since you are curious.

- Draco

[Kumo Trader]

Ok, that being the case, no problem, let's drop the idea.

George, I will mail you the EA separately, since you are curious.

- Draco

Hey Draco,

I really don't mean to "rain on your parade" by any means, I just want to make sure you understand what I mean when I say that you would not be backtesting true IKH if you turned it into an automated trading system.  You would only be able to utilize a small fraction of IKH's concepts given the limitations of boolean logic and that would, by definition, make the system different than IKH...

Best,

Ian

[draco]

Hi Ian

I think it's your role to define what is relevant to this forum. You say that no MTS could adequately reflect Ichimoku, that an ichimoku-based expert advisor would not actually be doing ichimoku at all but something else , and I think that is a very fair argument.

It doesn't mean I will close this line of inquiry, but if it doesn't belong here, then I won't discuss it here. I work in the IT area, so trying to bring something as complex and nuanced as Ichimoku into an EA is a bit of a challenge for me. If nothing else, I will get a better understanding of how it functions.


Regards,
- Draco

[Kumo Trader]

Hi Ian

I think it's your role to define what is relevant to this forum. You say that no MTS could adequately reflect Ichimoku, that an ichimoku-based expert advisor would not actually be doing ichimoku at all but something else , and I think that is a very fair argument.

It doesn't mean I will close this line of inquiry, but if it doesn't belong here, then I won't discuss it here. I work in the IT area, so trying to bring something as complex and nuanced as Ichimoku into an EA is a bit of a challenge for me. If nothing else, I will get a better understanding of how it functions.


Regards,
- Draco

Very well put, Draco.  I'm glad you are pursuing your line of thinking on this, as your own curiousity is always your best teacher.

Best,

Ian

[Kumo Trader]

Hi All
I hope kumotrader forum survive, this is the best forum about forex. Now something about IKH mechanical trading.
I'am testing EA Ichimoku from http://www.forex-tsd.com/expert-advisors-metatrader-4/210-ichimoku-ea.html
With a little changes I have a great results

initial deposit 300 euro
start: 2007.05.01
stop: 2007.08.26
Result: 382438.36 euro

what do you think about it?
Regards Greg

Hi Greg,

Thanks for your post.  I am not a fan of automating Ichimoku and turning it into a "mechanical system" rather than a discretionary one, as I strongly believe the human mind is much more adept at picking up the nuances of a price move and determining when and if to trade.  This is especially the case with Ichimoku, where you have a number of disparate things to consider before being able to form a complete picture of price action. 

On the other hand, mechanical systems are great for removing emotion from the equation, which is the single biggest killer of trading accounts, so I can see where this type of system has its benefit.  Nevertheless, I have yet to see a mechanical system for Ichimoku that has the requisite intelligence to keep your account in the black over the long term.  Metatrader's backtesting module is notoriously 'buggy' and is generally not to be relied on as a true test of a trading system, so my faith in your results would be suspect at best. 

Have you tried trading using Ichimoku as a discretionary system yet?  That is absolutely necessary before you ever consider creating a mechanical version, in my opinion, though I still believe any automated trading system using Ichimoku concepts is only a vague shadow of the original, comprehensive Ichimoku trading methodology itself.

Best,

Ian

[hardyfoster]

email me at
hardyfoster@gmail.com

put ichimoku robot in the subject line or something similar.

i have programmed an ichimoku robot myself and it works great.

it only opens trades

then i manually close them or trail the stop.  i will eventually have it trail the stop or close the trade.

if you understand programming and how to instruct a computer, an ichimoku ea can be programmed, whether it compares time frames or not.  it's all data.

the only problem i haven encountered which is a lack of programming knowledge is knowing when one of the lines is pointing up from it's previous flat state.  if my programming skills were better, i know it can be done very good as well.

[hardyfoster]

here is a screenshot of my robot i was running on a demo live market.

there are several open trades of the same pair.  i had to fix it so it wouldn't open multiple orders.  either way, if the pair wasn't giving a signal, it wouldn't have opened the order.

heres a link to closed trades:
http://realc4ever.googlepages.com/detailedstatment.html


below were open trades at the time of the picture.

http://realc4ever.googlepages.com/opentrades.jpg

[ezequieldk]

Well, thats Great man !!...
Good job...

I only use the ichimoku-hyo as a "manual" indicator. Yet I failed to generate the rules for buying and selling because there are so many things to take into account.
Currently I use Metastock, so that the formulas of metatreder are different.

I wanted to know if you have the formula of your own robot for metastock, and if it is not so, i wanted to ask which were the rules of your system ...

Many thanks and forgiveness for my bad
English