Clearly if you are measuring the correlation in real time then there is no point opening 4 trades (you are wasting money on excess spreads)...just open one pairs of trades when you expect the correlation to shrink and try to get your 10-20 pips then. What point is there in building up a winning pair of trades and a losing pair of trades...they simply cancel out and waste spreads.
Perhaps a simple way to measure correlation in real time is to track what the hypothetical 4 trades would be doing (without opening them) by storing the values in memory variables...then when you reach the point of less correlation (a big win and a big loss) you have reached the point at which to enter 2 real trades in the hope of capturing a pullback towards correlation. That would have the same effect but you'd pay only half the spread costs.
Another idea, if from coding standpoint this is too hard or too complex, open your 4 hedged position on a very very small scale (think 1 micro lot) and use that relation as the indicator to trigger your trade.
Also on the topic of correlation, it is very hard to distinguish when a true shift of correlation has occur and when it is just a temporary deviation. In terms of tradable correlations deviations, I have found they are infrequent in occurrence.
Due to the infrequence of appearance, and the always present possibility that the deviation is a new shift, I have found it very hard to solely base a consistent trading strategy off this.
With this being said, correlation trading is one of my many tools I employ when trading, however I have found it best to only be a part of the whole puzzle that is trading.
Your strategy sounds good.
Unfortunately, it's impossible to backtester your EA because of error 4107.
Any idea ?
Secondly when mt4 closes and re-opens ,the EA triggers new trades.
Could you introduce a feature to avoid that ?
In my opinion, your EA would need to only trigger the trades according a level of correlation indy in negative area for example.
Congratulations for your genious idea.
Hi cacus and fellow traders..
im a forex newbie.
Can u please advise me..
1. on what tf should we use the ea?
2. Will the ea open the position automatically or we need to manually open the trade?
Thank you sir..
Sorry guys, i was really busy this past few days.
Here's what i got:
Currently running QuadHedge 2.5 at 60/20, 0.02 lots with 5 cycles completed and 18usd in gain just in 5 days of trading.
Seems nice, let's see if this still in profit during a month.
About you questions:
PopsIsDead: Whatever TF, EA does this all automatically.
madscalp: it's not possible to backtest the EA with MT4, that's why i'm forward testing this strat to see if this seems as good as in paper.
PipALot and Styex: yes, i'll see if I can get some spare days soon and get my hands on this kind of "correlation measure" entry signals. But i thinks this needs a lot of previous analysis. PipAlot, I also attach you the indicators you asked for.
Bobster: have you reached any improvements with your idea? As I can see you mt4live page shows a bad buy open trade left... Something gone wrong?
Also I attach my inverse QH EU-UC, it's working with inverse correlation, currently started to test this one with 30/10 as sets. Will keep you updated.
Could you explain why sometimes the first couple is closed in loss.
If it's not possible, so there is a bug
And have you tested another couple according this link :
Currency Correlation | OANDA FXTrade InfoCenter
madscalp, it IS/MUST BE a bug, it's impossible (at least never happened to me till now) the EA to close the first couple in negative. At least not with a PipsTrigger value higher than 0. Which version are you running?
Would you please be kind to report your statements so i can take a look to the possible problem?
QuadHedge 2.5 (60/20), 6 cycles completed, $21,78 net gain, 6 days running. (0.02 lots) (cacus1.mt4stats.com)
QuadHedge 2.5 Inverted (30/10), 1 cycle completed, $2.96 net gain, 1 day running. (0.05 lots)(cacus.mt4stats.com)
EDIT: i forgot to answer you question about other pairs, yes, i'm starting to test EU-UC and see how this performs (cacus.mt4stats.com). Will see some other possibilities soon.
I'm making many tests and i destroyed the statement but it's occured with 2.5 one or two times i think.
If it's occurs again, I'll post the statement.
Some questions please :
Your settings are 60/20 in cacus1.mt4stats.com. Could you explain in the first cycle where are the 60 pips and how is calculated the 20 pips ? I don't understand.
What floating maxi did you get during these 6 days ?
Do you think for an improvement about an entry at a good moment according to a correlation indy ?
Here is a statement with couple EURUSD - EURJPY.
Yesterday this couple worked greatly.
This morning, I set 20/10 , 1 lot and only one cycle.
At 9:04, the first couple closed in gain ...ok ...good.
I checked the second couple during a few minutes and at 10:00 it was lightly in gain but did not close. At this moment, my balance was in gain about $ 80.
At 10:53, the second couple closed ....in loss .
Result of the cycle: negative !!!
How do you explain that ?
Maybe could you introduce in your EA an equity profit ie for example at 2 % profit the EA closes all the trades ?
Thanks for sharing your idea!
Now I have tried the v2.5 with 1 lot and 30/10 settings, default EU/GU pairs.
The first pair (buy GU / sell EU) closed with profit (+84 / -33) but with the second pair (buy EU / sell GU) made a loss overall (+22 / -91). 84-33+22-91= -18 instead of +10
What could be the reason of this? A price spike? Server lag? (I guess the EA can do the math. )