Trading similar currency pairs?

If you trade similar currency pairs, eg GBPUSD, GBPJPY, ie the GBP part.
Should all your trades be in the same direction?
More often than not, they move the same way, if GBP gets stronger, then it moves up in all currency pairs.

There just seems to be a lot of correlation between currency pairs where one currency is the same. So much so, that it is pointless trading similar currencies as you increase your risk.
?

Yeah, you'd think you could just do the cancelling out and it should 'work', but it doesn't.

For instance, a Long GBPUSD and a Short GBPJPY doesn't give you a Short USDJPY other than in symbolic equivalence. Your net position will not necessarily move anything like USDJPY.

There's a lot pf 'stuff' that goes on at the macro-economic level that 'means' things to real world users of the underlying (not just in currencies) which precludes any sensible comparison of synthetic postions with their logical equivalents.

So 'no' is my answer to your question. If you have good technical and/or fundamental reasons to be trading an apparently self-cancelling set of FX pairs - go ahead anyway as it wont make a great deal of difference.

As GJ says if you've got exposure to more than once currency across a number of pairs you are going to get the big moves potentially across a number of pairs - but that works both ways - literally.

GammaJammer said:

Actually Tony, provided that the trade sizes are the same in GBP notional, sorry to disagree mate but that's exactly what you have. Otherwise the arbers would step in and close any gaps.

Click to expand...

Then I must be doing something wrong as I looked at arbing using precisely these synthetic pairs and found the price action was so far out of whack that no opportunity presented itself.

Are you saying if you bought GBPUSD and sold GBPJPY your performance profile would EXACTLY mirror a short USDJPY?

TheBramble said:

Yeah, you'd think you could just do the cancelling out and it should 'work', but it doesn't.

For instance, a Long GBPUSD and a Short GBPJPY doesn't give you a Short USDJPY other than in symbolic equivalence. Your net position will not necessarily move anything like USDJPY.

Click to expand...

This is a synthetic pair, and by definition must be identical to USDJPY, for the reasons GJ has mentioned, as well as being the most basic of mathematical transformations.

You're paying twice the commish mind, but it's still exactly the same trade.

a picture speaks a thousand words.........

it's used mainly by traders whose brokers don't offer the more exotic pairs 'as is', as I understand it.

the method of creating a synthetic pair. Guy wants to buy GBPCHF, broker doesn't offer it so he buys GBPUSD and USDCHF, in equal amounts of notional USD.

Obviously anyone who is able to trade GBPCHF outright wouldn't do this because of execution risks and the fact that you're paying the spread twice, so I can't think of any other reason you'd use a synthetic pair. That said, I don't trade FX so I wouldn't know.

True enough; like I said, I don't trade FX and I don't trade retail so I wouldn't really see a purpose for it, I was more interested in the point that it has to the be the same theoretical trade, since it's just a substitution between ratios.

Does liquidity enter the picture?

I would imagine it is easier to get off USD/CHF and USD/JPY trades in size rather than CHF/JPY...

Skill Leverage said:

This is a synthetic pair, and by definition must be identical to USDJPY, for the reasons GJ has mentioned, as well as being the most basic of mathematical transformations.

You're paying twice the commish mind, but it's still exactly the same trade.

Click to expand...

I appreciate the points you and GJ are making, but while the theoretical synthetic position should mirror the primary pair, I’ve rarely found them to do so. If it were simply the spreads involved then I would have spotted that, but I don’t think it is.


I’ll give some thought to how to capture the divergences between these two net positions and post it up here. If I’m making a mistake somewhere in my calculations (always possible), I’m sure you guys will be able to point me in the right direction.

I already have that GJ, which is what led me to believe there wasn't a consistent edge to it. I'm still working on something I can post up here as I'm increasingly certain I've foobar'd the calculations somewhere along the line. Double-checking my formulas before going public in an effort to reduce the shame and pain.

I don't trade the data for which I have the DDE link, but that's largely irrelevant for our purposes here.

GammaJammer said:

PM me if you want to keep it out of the public domain - happy to try and check your calculations if you like.

Click to expand...

Nah. Thanks anyway. I need a public spanking now and again...see below.

Just to check my reasoning and math before I post up the data, if we assume there's no spread and we take a single $100K lot on the vanilla side of the trade and hedge it by constructing a synthetic position on the other side consisting of two legs of appropriate size which in theory is identically opposite to the vanilla side.

For instance.

Take a Long AUDJPY and hedge all exposure by taking a Short EURJPY and a Long EURAUD in appropriate proportion.

AUD JPY EUR
vanilla L AUDJPY +1 -1 0

syn.leg1 S EURJPY 0 +1 -1
syn.leg2 L EURAUD -1 0 +1

NET Pos 0 0 0


So in theory, it shouldn’t matter what happens (remember we’re spread-free in this example) and the combination should maintain a flat position. Is this correct?

In order to remain size-neutral we need to buy and sell the synthetics in proportion to the vanilla. So a 1 pip move on the vanilla must be matched by an equal and opposite 1 pip move IN TOTAL by the synthetics.

Using the pairs above as our example, with AUDJPY, EURJPY and EURAUD trading at $10.18, $10.17 and $7.41 per pip per $100K size, we would take a 100K in the vanilla AUDJPY, 50K in the EURJPY and 68K in the EURAUD. Thus a 1 pip move in the AUDJPY should be offset by an equal an opposite TOTAL of a 1 pip move in the two synthetics. Is that correct?

Final part of this post.

I am taking it as mathematically equivalent to replace the relative position sizes of the two synthetics by the same ratio applied to their respective pip moves. So a 1 pip move on the 50K EURJPY is mathematically equivalent to a 0.5 pip move on a 100K position. And a 1 pip move on the 68K EURAUD is mathematically equivalent to a 0.5 pip move on a 100K position. Is this a valid transform?

This might illustrate things for you:

http://www.forexfactory.com/attachment.php?attachmentid=30955&stc=1&thumb=1&d=1189881017

Right, now these data are pretty much what you would expect - and not what I'm getting.

Can you spot any obvious flaws in the assumptions I'm stating above?

MrGecko said:

Does liquidity enter the picture?

I would imagine it is easier to get off USD/CHF and USD/JPY trades in size rather than CHF/JPY...

Click to expand...

I think that is a very good point, for instance Aud Nzd can be very thin at times, so If you want to get some size on it can be better to trade Aud Usd and Nzd Usd against each other.

Erm just realised that there are three more pages of this thread that I hadn't read. My bad...

GammaJammer said:

If you did the same EUR amounts in your EUR/JPY and EUR/AUD and set it so that the AUD amount satisfied the size requirements for your AUD/JPY position you are 100% fully hedged mate - just have some P/L sitting in JPY depending on where you did all the legs.

Click to expand...

Rather than ask you to do all the work and explain through a specific example, how about…

Say I want to hedge a long AUDJPY $100K position. I decide to do a Long EURAUD and a Short EURJPY.

The EURAUD and EURJPY synthetic need to be bought and sold in quantities which will match the primary pair, pip for pip, in total.

A 1 pip plus move in the primary will provide a +$10.12 move in the primary position value. So I need to ensure I buy/sell USDCHF/GBPCHF so that their combined move of -1 pip equates to a -$10.12 move in the synthetic position value. Yielding a net move for the entire hedge of Zero. A position size of $68K for EURAUD and $50K for the EURJPY will provide the theoretically perfect hedge. Yes?

If I had done this at 08:15:46 this morning (which is when I time-stamped the data), as I write (15:40) the primary Long AUDJPY is standing right now at +77.1 pips. The synthetic short is standing at -106.7. An overall -29.6 loss on the position. Had I taken the opposite strategy, obviously this would have yielded a +29.6 profit. But this is beside the point. If my thinking and calculations are correct (which is where I suspect the problem lays), a theoretical hedge such as this does not perform as the simple math transform suggests it should.

GammaJammer said:

I think you are over-complicating it a bit.

Click to expand...

Generally a sign of the lack of total comprehension or the confusion which precedes it. LOL.

GJ there is a discrepancy with what I thought I would have expected from a hedge like this, what everyone else thinks we will get from a hedge like this, and the actual results I’m getting. I’m trying to find where my error is.

The attached shows two tables. The upper is the multiplier I’m using for each of the synthetic pairs (row 4) for any given primary pair (column B) based upon the current $/pip/$100K move. The second table shows the position size required for to hedge a nominal $100K size on the primary.

Where am I going wrong?