Trading the cointegration | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Day Trading vs. Swing Trading vs. Position Trading 86 replies

Cointegration of currency pairs 62 replies

Synthetic hedges, cointegration, mean reversion and similar stuff 419 replies

Cointegration, Synthetic hedges, mean reversion in R, Tech Thread 0 replies

Attachments: Trading the cointegration
Exit Attachments

Trading the cointegration

  • #21
  • Jun 17, 2012 7:27am Jun 17, 2012 7:27am
Quoting eremix
Ok,but i dont understand how this is bad for our trading.Since we can find coefficients that produce short term cointegration,why cant we exploit that?
Anyway,i've also tested the exact same strategy over a period of 5 years in pairs that are not cointegrated at all.Same results,nice profit.
overfilting doesn't mean you found a relation - it just fits the data for something you are looking

just like looking that - at the same time the bakery in Venice drops their price in bread, Fb stocks drop also; It could happen at one time, doesn't mean it will happen again
  • #22
  • Jun 17, 2012 10:17am Jun 17, 2012 10:17am
Quoting Rikers
overfilting doesn't mean you found a relation - it just fits the data for something you are looking

just like looking that - at the same time the bakery in Venice drops their price in bread, Fb stocks drop also; It could happen at one time, doesn't mean it will happen again
Hmmm got it.You are probably right and thanks for pointing that out.
I never said though that cointegraton trading is risk-free trading.
You are betting that the cointegration will hold and that the residuals will get back to their mean.
It's like taking a trade after the price has bounced off a S/R level 3 times.You are betting that it will bounce one more time,noone guarantees .
Heck,that's a great signal for techincal traders though.And many have made a fortune off of it.
Anyway,i agree with you but it's trading.You have to take risks.And it's the best i ve tried so far..
  • #23
  • Jul 4, 2012 6:31am Jul 4, 2012 6:31am
I recieved some pms the past week and i thought i should answer in public.
I was on vacation for the past couple weeks and i havent traded in a while that's why i stoped posting.
I was asked the following questions:
1. Is it a good software based on your experience?

2. Do you trade live with it? How's the result?

3. Your demo account looks REALLY good too. Are those trades done using analysis from arbmaker?

4.i just found your thread about cointegration using ArbMaker, are you still using it? because im considering ArbMaker too. is it worth the cost ?

I have used arbmaker for about 3-4 months.It is a new software.Based on my experience i have to say that it is a very god and reliable way to trade.Dont expect to win every day but if you set some rules and do some backtesting you'll realize that it can be very profitable.Read some fundamentals,take in account S/R levels,use the software and you have yourself a very profitable way to trade.My demo account is a demo account.I take trades and risks that i dont want to take in a live account.It may look good but my live account looks even better.
Is it worth it?A lifetime license costs about 2000$.You'll have to use a live feed provider as well,which costs about 80$/month.Based on my experience and the excessive backtesting i ve made i would say that you can easily come up with a 10% profit/2 weeks.That means that if you use a 10,000$ live account,you'll cover your costs in about a month.Having said that, if you take trading seriously,it is a very powerful tool in your arsenal and it is definetely worth it.
  • #24
  • Jul 12, 2012 2:41pm Jul 12, 2012 2:41pm
For everyone following this thread sorry i havent posted for a while.It's just that i am rarely at home in summer.Here are the pairs i am trading at the moment.You ll probably notice two things.Firstly many of them are equities and secondly many of them are not cointegrated.The thing with equities is that they have much better statistics than currency pairs and they can be much more cointegrated. The non cointegrated trades have been taken in a demo account. I ve found that trading non cointegrated pairs can also be profitable.Its just much more risky.
I also noticed about two days ago that the team behind arbmaker has launched a referral program http://arb-maker.com/referrals .
Anyone interested can send me a pm and i ll be the referrer.

http://img31.imageshack.us/img31/2441/34303515.png
  • #25
  • Jul 12, 2012 6:35pm Jul 12, 2012 6:35pm
From what I can see, you are more or less taking USD out of the equation and trading GBP/CAD.


Quoting eremix
Ok,i 've decided to close my trade because of the upcoming greek elections.
Z=0.2 and my profit is 150 pips at the moment.I am not willing to have open trades this Monday.Take a look at the huge move GBPUSD made....like magic.

http://img85.imageshack.us/img85/1543/exitz.png
http://img829.imageshack.us/img829/8...pusdusdcad.png
  • #26
  • Jul 13, 2012 8:32am Jul 13, 2012 8:32am
Quoting fierceman
From what I can see, you are more or less taking USD out of the equation and trading GBP/CAD.
Yes,exactly.I just showed an example of a trade.Didnt want to make it complicated.
  • #27
  • Jul 14, 2012 8:55am Jul 14, 2012 8:55am
I had to do some coding in order to to combinr matlab to mt4. what about coding your strategy to make it automatic, so we could run some backtest on it ?
  • #28
  • Jul 14, 2012 9:03am Jul 14, 2012 9:03am
Quoting Rikers
overfilting doesn't mean you found a relation - it just fits the data for something you are looking

just like looking that - at the same time the bakery in Venice drops their price in bread, Fb stocks drop also; It could happen at one time, doesn't mean it will happen again
but it look like the boy is actually making money out of his cointegration story...

Rikers did you already try it in the past with unsuccessful results ?
  • #29
  • Jul 16, 2012 2:21pm Jul 16, 2012 2:21pm
Quoting dinhero
I had to do some coding in order to to combinr matlab to mt4. what about coding your strategy to make it automatic, so we could run some backtest on it ?
Arbmaker has a backtester where you can backtest any cointegation strategy you want.It can also do automatic trading if yo have an account with Interactive Brokers.I dont.So i do everything manually.

Quote
but it look like the boy is actually making money out of his cointegration story...

Rikers did you already try it in the past with unsuccessful results ?
The question isnt whether it is profitable or not but how proitable it is.
It has been proved profitable since the majority of hedge funds use it.
  • #30
  • Feb 12, 2013 4:51pm Feb 12, 2013 4:51pm
Did you managed to set up MT4 feed ?? I followed the instruction from manual but when i try to test it by clicking MT4 icon, MT4 starts correctly but i'm getting error 'No connection could be made because the target machine actively refused it 127.0.0.1:2007'

Krzysztof
  • #31
  • Apr 15, 2013 10:36pm Apr 15, 2013 10:36pm
Hi, I am a MATLAB Expert (not a Trade Expert) and am in the process of doing a major back testing on the co-integration strategy to see how it works on real data...The data is expensive but I have to do it to get/see the results. My code will find co-integration relation among a huge pile of dataset.
Infact, I have already implemented the co-integration and the generation of trade signals, you can see that in the figure
The top chart has the closing rates in red and blue color. The middle chart has the residual and the trading signals based on the residuals.
..... but there are some questions un-answered regarding trading......
I made these chart from historical data (e.g 14 values only). The trade signals were generated from historical data, but how to do it for current data...... e.g Suppose we have 2 shares which are showing co-integration in history/past..... How can we trade in present basing on this simple historical knowledge. e.g I know that USD and GBP had shown co-integration since last 1 month, then what next ???
Attached Image (click to enlarge)
Click to Enlarge Name: coint.jpg Size: 79 KB
  • #32
  • Last Post: Edited 10:00pm Apr 16, 2013 2:05am | Edited 10:00pm
Quoting Hkc
Hi, I am a MATLAB Expert (not a Trade Expert) and am in the process of doing a major back testing on the co-integration strategy to see how it works on real data...The data is expensive but I have to do it to get/see the results. My code will find co-integration relation among a huge pile of dataset. Infact, I have already implemented the co-integration and the generation of trade signals, you can see that in the figure The top chart has the closing rates in red and blue color. The middle chart has the residual and the trading signals based on the...
Does your backtesting include out of sample testing? It seems to me if you know enough to run a backtest on your cointegration model, then you already have an explicit strategy. It looks like your spread line is plotted in green. You know what you are buying and what you are selling and so it's unclear what you might be wondering about. Instead of historical prices, you feed live data into your model and you should be able to replicate similar results to you are getting now, assuming you are doing proper out of sample testing.
0 traders viewing now