System to hedge every major dollar pair for profit / low DD

Hi Croupier,

Yes it's mean't to bridge the gap between the programs to call the Rterminal. You can locate it on 7bit website, in his words "This is my solution for the problem of mql4 programs not being sufficiently able to communicate with the outside world. It adds the ability to run arbitrary python code (even multi-threaded) directly from within an mql4 script, expert advisor or indicator". Still its frustrating not being able to make the program work.

G5supra.

See this post and please use the mt4R R Project thread for technical support.
http://www.forexfactory.com/showthre...08#post6545008

FXEZ, do you think the below idea is worth testing for an EA?
Constantly scan target pairs and open/close positions based upon the ADF test threshold?

BTW, I got 7bit's arbmat and trendmat working after following your instruction of "check the enable EA in the MT4", both are good starting point for a pairtrading EA, I am also trying to transform my offline SVM prediction testing result in R into an online EA.

You could certainly try it. Just keep in mind the caveat posted by saussiche that having a theme to the selection may make things more stable. Also, recall that the ADF or any other statistical test essentially just gives you confidence that the model or the hypothesis is sound on the data tested. However, the more important question is whether the hypothesis or model makes money out of sample (on new data). So I would put more weight in the out of sample test than on the statistical test. I'm not sure if ADF will tell you in a timely manner when to close out from a particular model but it possibly might depending on your specific set of rules and window size.

A bit of advice: Treat the testing as generalized framework testing rather than whether any specific system is sound. You are developing a framework for system development using certain tools. Test a group of individual models selected using your method on out of sample data over many different sliding window samples to get a feel for how long you can expect the models on average to continue to perform in similar stable fashion into the future. If you follow the same testing methodology with your framework in the future, you should get similar results.

(purely hypothetical) For instance, let's say you follow your selection strategy and apply it to 2000 bars of data from the M15 time frame. This is about 500 hours or 21 days of look back period. You determine from your OOS testing that you can effectively run the average model from your selection scheme forward for 500 bars before the OOS performance starts dropping off. So you get about 5 good days on a particular model then you need to reoptimize. This is what I mean in the previous paragraph. You will then need to determine if the 5 days of life of each model provides enough opportunity to be worth the effort.

Glad to hear you got arbomat going. Sometimes it is the simplest things that get us hung up.

for those of you trading this system. if you just use eurusd and gbpusd, hedge. it'll just end up exactly with eurgbp. From OANDA exposure tab, you could see that.

am i wrong? and the graph just show the same as eurgbp. why not just trade the eurgbp?

Like trader101 system, it is actually heavily rely on chfjpy and eurjpy pair.

I also test the 7 pair basket, it end up with either long eurchf or short eurchf.

anybody else noticed this ?

Cointegration techniques have been criticized as being confusing, unreliable, and dangerous. from the wiki

but i'm gonna try this method in real oanda account

I'm not sure if anyone is still reading this thread but I get the sense that those that were leading the charge on finding the new system either gave up or went off in their own new direction. Such is common with cointegration and even correlation threads throughout most forums.

I recently started trading cointegration but experimented for quite some time with correlation. I'm versed in VBA and MQL and learning R. My results have been mixed. There are good weeks and there are bad weeks. Just like any other system it is not fool proof. The major issue, just like all other trading system, is proper money management.

Having said that, I'm wondering if anyone out there has mixed cointegration with rebalancing algorithms successfully? A general rebalancing algorithm paired with what 7Bit defined here has potential.

How did it go?

Hi,

I'm relatively new to this but looks good.

I would like to know how can i compute sizes. For example for the setup below :

AUDUSD -4 (0)
EURUSD 100 (0)
GBPUSD -4 (0)
NZDUSD -5 (0)
USDCAD -7 (0)
USDCHF 100 (0)
USDJPY -16 (0)

Are these sizes below are a valid option ?

AUDUSD 0.04 Lots
EURUSD 1 Lots
GBPUSD -0.04 Lots
NZDUSD -0.05 Lots
USDCAD -0.07 Lots
USDCHF 1 Lots
USDJPY -0.16 Lots

Yes the sizes look correct, but I've never seen two currencies with 100 hedge ratios. I don't know why but it doesn't look right to me.

Does Arb-o-Mat works 4u at MT4 update (600)? Because mine has some problems. What have you guys done, do I have to update anything? Many thanks 4 reply.

Rgs FxJ

Have You got ARBOMAT working on MT4 update (600) ? Only 1 of my versions is seen by MT4 in NAVIGATOR WINDOW but it's still not working :/

EA doesn't work because of errors during compilation of mt4R.mqh and common_functions.mqh. I've come back to my work on this subject but earlier there had been no problem with running this. Is there a possibility, the syntax have been changed? What else may couse the problem?

I havent got any progress with set up the chart called by R. I guess mt4R.mqh and common_functions.mqh been out-dated! Right?

back bars 3500


back bars 2000

Hello,

I am currently testing this system? Does it burn your account in long term or does it work?

I would like to know if someone has the same problem as me

The R graphics popup and close 1 sec later, how can i fix this problem?

Thanks for helping

Have a good day

I also encountered the same problem, seek to explain

Hey Guys

I'm having a trouble getting the R charts going. I followed DirtyBrown instructions. when I go to attach to chart nothing is happening. Im not sure if its because everything is outdated since MT4 is now a new built. Anyone installed and using R lately? and with updated R 3.2.2 ?

Any help would be appreciated.

Thanks
Spaz

Have you downloaded the updated DLL and MQL files from here? https://www.mql5.com/en/code/11112

Have you followed FXEZ' instructions http://www.forexfactory.com/showthre...08#post6545008 ?