Triangular Arbitrage

The arbitrage gap is filled.

I liked your idea so i made an EA for a long term backtest to get results and optimize settings, for now, 26 years backtests results in a loss in all cases:
- If you don't set an SL, in some point one pair will explode and the other 1-2 will be ranging so the system breaks.
- If you set an SL it has to be BIG to manage temporal drawdown.
I'm currently exiting when the PnL is >= a % of used margin, but i'll add other methods after the initial backtests.

Will be backk with more data ASAP.

Let me clarify. In my opinion, your information provides a generalised analysis, but what about specifics? Which of the three currency pairs should be distanced from the moving average? How far from the moving average and under what conditions do we fix or understand that it is necessary to close trades?

Currency pairs do not have a single fixed price. The price you see on the screen is the instantaneous market price. The EMA, on the other hand, represents the average price over a defined number of bars (x periods).
Instant prices are temporary and often distorted by short-term order flow. That is why they tend to correct or react.
If you attempt triangular arbitrage based purely on live spot prices, you may capture brief one-second inefficiencies. However, this approach is generally not favored by brokers, and profits from such trades are often disputed or not honored.
This is not true triangular arbitrage.
To consistently profit from arbitrage, you must first accept the principle that:
3X = 0
In other words, the three legs must be in equilibrium.
Then, your task is to identify where the imbalance—the deviation from zero—exists within that equation.

How do you find the error in the 3X = 0 equilibrium?
Let’s revisit last week’s chart.
Even if you simply look at the 100 EMA in these 3X pairs, you can already see the imbalance. However, arbitrage requires looking at the equation from the correct perspective. Those who work on this equation often spend more than 10 years studying it. With each step forward, they feel they are on the right path — yet they still cannot find the answer. Because their focus remains on price.
Just to shed some light on the logic, let me demonstrate using the EMA example and give a case.
In the image:

1. EURCAD is below the EMA → EUR (–), CAD (+)
2. CADJPY is above the EMA → CAD (+), JPY (–)
3. EURJPY is below the EMA → EUR (+), JPY (–)

So the question is:
Who is imbalanced in the market?

If these three pairs form a zero-equilibrium structure, then when we analyze them, we see that CAD is positive in both pairs and JPY is negative in both pairs, while EUR is positive in one pair and negative in the other.
That means the imbalance is in EUR.
According to the 3X arbitrage rule, the combinations must be either buy–buy–sell or sell–sell–buy.
So you would place a buy on the fallen EURCAD and a sell on the risen EURJPY. As a result, by definition of the equation, your structure becomes BUY BUY SELL.
From that point on, the rest is simply waiting for the market to close the imbalance.

3 pairs arbitrage is impossible in any broker. The inefficiencies are not often, nor significant , therefore if you find any, by the time you Triger entries , it is gone, or, the spread and commission eats the imbalance.
In my past life I had a very good arbitrage software. But just for singular pairs, different brokers. Once in a while, retail brokers quote different prices, say eurusd.
So the software shorted the high price, and long the low price. then waited for a reverse and close both trades, usually you come up with a net profit. Problem was, in one broker you had profits and in the other losses. If in one broker you were losing all the time, that account started to get low in funds, replenishing the low account was always expensive, specially if needed to withdraw from the winner brokers. A hassle. Ended up being a breakeven at best.

3X gaps must eventually be filled.
However, when people see the profit and loss figures in example 3X charts, everyone starts chasing the “big fish” — and they get lost along the way.
That path requires much deeper mathematics and knowledge.
Whereas in reality, arbitrage profits are sufficient to turn the capital and time that most forex traders waste into substantial gains.
Buy–buy–sell or sell–sell–buy is the safer path.
As long as you have not identified the correct solution, any other combinations can be unforgiving.

Because this is an a*b=c equation based on the price. In this case, you can no longer collect your winnings once you've won.

I read from the start of the thread, and realize the first post is not visible, but the second post mentions pair equation. Yes, 3 pairs equations produces possibility for arbitrage, but not worth to spend time on it. From there on in the thread, it seems became a salad of different concepts. Currency strength and many others.
Galip, I do not understand "you can no longer collect your winnings once you've won." To me, impossible, to win period.

In any case, some other ideas exposed here are worth a second look.
I notice people mention not able to back-test some ideas. But most mentioned are possible to backtest in Jforex platform. I have done some strategies that pick up prices from other pairs . Data in Dukascopy is in ticks, with timeStamp, so, that is not a problem.

If anybody explain with clarity an idea, then, maybe I can help. So far nothing is clear in this thread :-) :-) . To many topics

3X is a reduced form of the zero-equilibrium equation of the eight major currencies in forex. It is a complex subject, but it represents the pinnacle of forex trading.
3X is an advanced framework that can simultaneously provide arbitrage opportunities across three pairs, two pairs, and even identify a single target pair in advance. When you reach this level, it allows you to trade the target pair and its cross at the same time, effectively executing a double-position strategy.
Let’s take a live market example:
FOR TONIGHT:
In the GBPAUD – AUDCHF – GBPCHF equation, the structurally imbalanced currency is CHF. However, over the past two days, GBP has experienced a rapid distortion. That imbalance must be corrected first. After that, CHF will adjust.
Instant trade example:
GBPAUD — BUY
AUDCHF — SELL
GBPCHF — SELL
This setup is expected to correct first. In this structure, you are effectively selling AUDCHF both directly and through its cross exposure. We will review the outcome tomorrow evening.

The main target is AUDCHF; he sold both himself and his cross. Sometimes he wins more, sometimes his cross wins more. But they both win.

With the GBP gap close approaching, shifts toward CHF may begin.

Gbpcad buy
cadjpy sell
gbpjpy sell

If you can explain your method i can code it into an EA

mate you can not open same size for all 3 trades. each size should be calculated accordingly with the correlated prices.

and if you do that, then it makes no sense.

Good luck anyways

After all, it doesn't matter, i coded an AE using the 100 EMA to "detect" the imbalance, it works sometimes, but in a long run (15+ years) all backtests die, no matter if you correctly calculate the lots or you use same lot size, 2 things can happen, either you don't use SL and you blow your account because one pair just goes crazy it's own way, or you use a drawdown % SL and you just keep losing because the same scenario.
I always do LONG term backtests (got tick data for most pairs since year 2000) because most "holy grail" strategies just works for a defined period.

Your results was predictable GotyMetal.
In essence, there is no such imbalance in tri-equation pairs . (it may be, but for some milliseconds only) Regardless of EMAs .

For example:
from the photo of post 275.
Gbpcad buy 1,82491
cadjpy sell 115,233
gbpjpy sel 210,289

1.82491 * 115,233= 210,289 That is a perfect balance. And those prices will continue to fulfill the equation. Therefore, if the lots traded are perfectly balance, then impossible to make/lose money.
Now, if lots are not balanced (Mentioned by mrBrown) then you might have profit/loss, but minimal, due to the imbalance. But in reality, would be better just to enter with the net imbalance to avoid spread costs

uff. I just realize that
Gbpcad buy
cadjpy sell
gbpjpy sell

is not hedged. The whole thing is a short CADJPY. So, if CADJPY goes down, profit, if goes up Loss.
If that is the intended purpose, then is fine, but why trade the other 2 pairs?