Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Triangular Arbitrage 251 replies

Nerr Smart Trader - Triangular Arbitrage Trading System 17 replies

Triangular Arbitrage and carry trade 6 replies

Triangular Arbitrage 7 replies

Question: Trade Mechanics of Triangular Arbitrage 6 replies

Attachments: Triangular Arbitrage
Exit Attachments

Triangular Arbitrage

  • Post #1
  • Quote
  • First Post: Edited Jan 2, 2008 5:07pm Dec 28, 2007 7:47pm | Edited Jan 2, 2008 5:07pm
Is going
long 1 lot of eur/usd +
long 1 lot of usd/jpy = long 1 lot of eur/jpy ?
(is the ratio of the number of lots correct for a perfect hedge?)

If so, when you hedge them, do they always revert to some mean P/L?
What is the range of +/- P/L like for 1 lot of each?

Which combination of major & cross pairs and brokers give
most discrepencies and thus better for arb opportunities?
It seems better to long the majors and short the cross pairs ...is that always so?

Why does the net P/L fluctuates so much even though the hedge between major and
cross pairs is a standard 1:1:1?

I found some articles on FPI but I am looking for some practical examples.
Are there any EAs written for Triangular Arbitrage besides FPI ones?

anyone else doing triangular arb here care to share more of their experiences?
I implemented some forward testing of FPI using Metatrader.

First of all you need to calculate the correct trade sizes, these will be fractional so you need a broker that provides small lot sizes.

Second, the FPI relies on a detected arbitrage opportunity between 3 (or more) symbols, with a Metatrader EA you can only monitor one symbol in real time, you have to then query the current values of the other symbols, this leads to calculated FPI values which are either out of date or just wrong. (There may be some clever way to do this that I'm not aware of)

Thirdly, Metatrader order execution is hardly what one would call 'razor sharp' and the spreads are usually quite high for some of the symbols in the ring.

So in summary, I found that FPI was pretty much a non-starter on Metatrader.
The breaking of a wave cannot explain the whole sea.
My question is this:

If I long 5 lot of eur/usd, long 5 lots of usd/jpy &
short 5 lot of eur/jpy, why doesn't my net P/L
revert to some mean value if it is essentially hedged to
nothing...Instead it widens to either more positive or negative?
Hello scalpmaster,

The reason is that your exposure is not zero.
With those equal lot sizes, you are still very exposed to USD/JPY.
You are effectively still short USD/JPY 2 lots.

You would need to buy 735,000 USDJPY to be flat

Coder
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Exposure.gif Size: 58 KB
Click to Enlarge Name: Flat.jpg Size: 194 KB
  • Post #6
  • Quote
  • Edited 3:22am Jan 3, 2008 3:06am | Edited 3:22am
Quoting Coder
Hello scalpmaster,
The reason is that your exposure is not zero.
With those equal lot sizes, you are still very exposed to USD/JPY.
You are effectively still short USD/JPY 2 lots.
You would need to buy 735,000 USDJPY to be flat
Coder
Thanks, your figure should be right but why 5eur/usd x 5usd/jpy
is not directly equal to 5 eur/jpy as the multiplication should give?

Anyway, where can you find the exact numbers to be flat for
other currency pairs and lots ratio as well?
Is there a table that gives the correct ratio to have a equivalent hedge
for the different majors and their crosses?
Do you have to trial & error on this basket tester each time for the right figures?
Where can I get this basket tester?
Hello scalpmaster,

The currency amounts to create a hedge are determined by the
currency values at the time that the hedge is created.

In the example pictures, it shows the following:

1 EUR = 1.4708 USD
1 USD = 109.53 JPY
1 EUR = 161.05 JPY

1 standard lot = 100,000 therefore:

100,000 EUR = 147080 USD

If we are long EUR/USD, we are buying 100,000 EUR & selling -147080 USD.
If we are short EUR/USD, we are selling 100,000 EUR & buying 147,080 USD.

So from this, you should be able to see that as the price of the EUR/USD
moves up and down, that it will be affecting the amount of USD that each
EUR can buy or sell.

Coder
Just need to know is there a table somewhere that shows the ratios to create an exact equivalent hedge between majors and crosses?

e.g. eur/usd x usd/chf = ? eur/chf

gbp/usd x usd/jpy = ? gbp/jpy

eur/usd / gbp/usd = ? eur/gbp

...etc
Hello scalpmaster,

You can use the basket tester or the Oanda platform has an
exposure tab, or just create a spreadsheet to do the calculations.

There is no table that I'm aware of. Remember that the amounts change
every time the values change, so you really need a calculator, not
a static table.

Coder
so the ratio between major and cross pairs is a Dynamic one to create a zero net exposure? i.e. the spreadsheet needs livefeed? or you need to change the lots input constantly in the backtester to maintain zero exposure?

Have anyone done an EA to display live this dynamic zero exposure ratio for different majors and their crosses?

Does this mean all FPI stuff is not correct too since there is a 'dynamic' factor involved in executing a triangular arbitrage?
Yes, it is dynamic, but not enough to be significant.
If the price moved > 1000pips, then you would want to think
about adjusting the hedge.

But if you want to be technical, the amount to hedge would change
every time the prices move.

PS.

Forget the FPI stuff. A lot of people have wasted a lot of time
chasing something that brokers don't want to give you.

PPS.

Infact, for get the whole triangular arbitrage idea.
It does not work with retail forex brokers, and you will find that
if you pursue it to ECN brokers that the slippage will kill any profits that
you may be lucky enough to make.


Coder
Any updates on this Topic? I think its quite possible
Quoting johnny122
Any updates on this Topic? I think its quite possible
Possible in theory, but automated systems at large institutions and banks completely dominate this strategy and crush out any opportunities faster and more efficiently at the tier of trading (trading direct between banks and large institutions)... by the time quotes get down to the retail level, with all the markups in spread and commissions, and additional latency, and less efficient execution paths... triangular arbitrage becomes pretty much impossible to do well.

A retail trader would have to get really creative in how they employ such a strategy if they are to find a way that works profitably..
FXGears.com
hi,

I code a EA use to analysis feasibility of Triangular Arbitrage in Metatrader. As Craig said, it's too hard.

Attached Image (click to enlarge)
Click to Enlarge Name: 3928120 FxPro - MetaTrader 4 - Demo Account - [EURUSD,M1]_2012-11-28_15-30-30.png Size: 100 KB


Attached Image
Take, for example, the GBPUSD - USDJPY - GBPJPY triangle
we calculate the price of GBPUSD = GBPJPY / USDJPY, we got 157.387 / 114.017 = 1.38038, and the real price of GBPUSD at the moment = 1.38048. That is, 10 points higher than the calculated one! i.e., in theory, after selling GBPUSD, we should get 10 p.
We also make calculations for all three currencies and open 3 sell sell buy transactions - for all three / Now we are guaranteed that wherever the price goes, the drawdown will be within the total spread.
That's just the lot is the same for everyone, or do you suggest calculating it somehow depending on the tick value?
Quoting VladimirH
Take, for example, the GBPUSD - USDJPY - GBPJPY triangle we calculate the price of GBPUSD = GBPJPY / USDJPY, we got 157.387 / 114.017 = 1.38038, and the real price of GBPUSD at the moment = 1.38048. That is, 10 points higher than the calculated one! i.e., in theory, after selling GBPUSD, we should get 10 p. We also make calculations for all three currencies and open 3 sell sell buy transactions - for all three / Now we are guaranteed that wherever the price goes, the drawdown will be within the total spread. That's just the lot is the same for everyone,...
10 POINT=1PIP. Now consider the spread. which minimum is 2 pips for each pair. It does not reach us. Sorry for my google english
  • Post #17
  • Quote
  • Last Post: May 4, 2022 6:34am May 4, 2022 6:34am
Is this really useful and profitable?
0 traders viewing now