One pair or multiple pairs??

[CHRISTO9HER]

Hi guys,

I have recently been working on a forex methodology that I am going to start trading pretty soon once the uni exams are over.

It uses 5min bars and trades roughly 2-5 times per day per pair during London session and a couple of hours either side. Average (successful) trade duration about 2-4 hours. I have developed this from scratch and I would say that it is probably 80% mechanical and 20% discretionary.

Anyway, my question is if I am going to start trading a system that trades with this frequency, am I better off trading it on only 1 pair? I origninally developed it to be used on cable, but with a little tweaking (only adjusting for the different ATR), it appears to work equally well on EURUSD and USDJPY.

I was origninally going to be risking 0.5% per trade when it was going to be only for cable, but I could do it over all three and risk around 0.2% per trade.

As far as I can see, the advantages of using all 3 are pretty obvious in terms of being more likely to catch a nice move, so what are the disadvantages? Should I stick to one pair?

Thanks in advance guys

Chris

 

[rathcoole_exile]

as they say "don't put all your eggs in one basket."
I think you may be correct in looking at spreading risk, and also opportunity.

Go for the 3 pairs, but keep accurate comparative notes so you can maybe drop the underperforming pair every quarter and try out a new pair ?
How does that sound ?

 

[rathcoole_exile]

by the way, does the system work on currency futures too, have you looked at GBP futures ?
May be another way to diversify risk ? okay, they run almost in parallel with the spot pairs, but not always and not exactly.

 

[MrGecko]

First and foremost, chris, don't let trading **** up your exams. Seriously mate, no matter how good your system is, it won't make up for a desmond and it won't get you a job.

Moving on, I should admit that am not a user (nor fan TBH) of mechanical systems; anyway, you are right to say that different currency pairs behave in slightly different ways - the obvious suggestion is to backtest / forwardtest your results. I am sure you have the quantitative and technical skills to adjust your model appropriately, and of course 3 pairs is likely to give you more opportunities, both for a profit and a loss.

 

[nine]

The downside of your proposal is that all 3 have USD as one component.

Why not try gbp.jpy, aud.usd, jpy.eur and eur.usd ... but never take more than one position with usd in it. Perhaps not the best combination but you should get what I mean.

 

[Rhody Trader]

nine hits the point. If you're trading EUR/USD, GBP/USD, and USD/JPY you run this risk of them all moving in the same direction at the same time rather frequently. If you're idea is to generate more trades per day (and if the system really works well that should be your objective) then you would probably be best off looking at a few relatively uncorellated cross pairs.

 

[grantx]

Chris,

Agree with Nine. However, you can look at all three but take a position only in one pair at a time.

To exapand on John's point re non-correlataion, you could also look at correlated pairs where the correlation breaks down.

Grant.

 

[CHRISTO9HER]

Thanks for the replies guys, as always you have raised some excellent points.

First and foremost, chris, don't let trading **** up your exams. Seriously mate, no matter how good your system is, it won't make up for a desmond and it won't get you a job.

Ok mum Actually MrGecko, you are absolutely right. I am sitting Mathematical Economics 2 on friday and I am fine tuning the trading system when I should be fine tuning my constrained maximisation of the utility function!!

The downside of your proposal is that all 3 have USD as one component.

Why not try gbp.jpy, aud.usd, jpy.eur and eur.usd ... but never take more than one position with usd in it. Perhaps not the best combination but you should get what I mean.

and

nine hits the point. If you're trading EUR/USD, GBP/USD, and USD/JPY you run this risk of them all moving in the same direction at the same time rather frequently. If you're idea is to generate more trades per day (and if the system really works well that should be your objective) then you would probably be best off looking at a few relatively uncorellated cross pairs.

I think you are both right. As Rathcoole said, "dont put all your eggs in one basket", but if all the pairs have USD then its not a whole lot better than one basket!

Chris,

Agree with Nine. However, you can look at all three but take a position only in one pair at a time.

To exapand on John's point re non-correlataion, you could also look at correlated pairs where the correlation breaks down.

Grant.

Thanks for this Grant, that is a very interesting idea to look at all three and then take the position in only one. It would be quite easy to come up with some quite objective criteria for which one is showing the most potential for the trade. If I do go for these 3 pairs, this may be the way to do it.

I also think that looking for breakdowns in correlation is certainly an interesting area but may be a little beyond what I am trying to do, at least to begin with.

OK, so the next step for me is to quantify the correlations between all of the pairs. I will run the data for 5min and 1hour periods through some of the econometric software on campus and come up with a correlation table for all main pairs. I will post this tomorrow and hopefully we will be able to see which pairs may work well together.

Again, thanks guys you are all stars

Chris

 

[MrGecko]

Out of interest, which econometric software do you use? Have you come across Oxmetrics?

 

[CHRISTO9HER]

Out of interest, which econometric software do you use? Have you come across Oxmetrics?

I havnt used that one. I have always used either e-views or stata, but I prefer eviews. Its one of those things where whichever one you learned for the first time on, you will always think it is the best.