How do you actually decide which forex pairs are worth trading?

New comment
 
I’ve been thinking about this a lot lately and wanted to hear how others approach it.

There are so many pairs available, but realistically we can’t trade everything.

Do you stick to the same pairs every day, or do you adjust based on what the market is doing?

I’ve noticed that sometimes I take trades that look good on the chart, but the move just doesn’t follow through.

It made me wonder if I’m missing something at a bigger picture level, like how the overall market is behaving or which currencies are actually strong vs weak.

Would really appreciate hearing how you approach this, especially if you’ve already figured out a consistent way to choose good pairs.
 
I think part of the issue is that a lot of us focus too much on individual setups without considering what the rest of the market is doing.

I used to take trades that looked perfect on one pair, but they wouldn’t go anywhere because there wasn’t real momentum behind that currency overall.

Once I started paying attention to which currencies were actually moving across multiple pairs, it became easier to filter out those weaker setups.
 
Charly Diaz:
Would really appreciate hearing how you approach this, especially if you’ve already figured out a consistent way to choose good pairs.

Long-term volatility coupled with good liquidity is beneficial to any kind of trading other than short-term scalping. This is based on the premise that if the market doesn't move much, then it's tough to catch a swing.

Have a look at the free myfxbook volatility table. You can sort by pair and timeframe. Hint: XAUUSD is #1 (almost always), and GBPJPY is #2 (most of the time).

 

For EAs what have worked for me is backtesting each pair individually over a large span of time and check the stats.

Then build a portfolio of Pairs and apply the EA to it.

 

Yeah that makes sense, especially about volatility and liquidity.

I’ve been noticing something similar — like even if a pair looks good technically, it doesn’t always move if there isn’t real strength behind that currency overall.

Starting to think it’s not just about the pair itself, but more about which currencies are actually strong vs weak across the market at that time.

Do you look at that as well or mostly stick to volatility?

 
Charly Diaz #:

Starting to think it’s not just about the pair itself, but more about which currencies are actually strong vs weak across the market at that time.

Do you look at that as well or mostly stick to volatility?

In my style of trading, any particular time (or any particular trade for that matter) is irrelevant. A high statistical level of price movement of a particular instrument that tends to continuously stay high over extended time is key. I generally look, in that way, at the daily timeframe volatility numbers at myfxbook and then settle on an instrument. XAUUSD is a prime example. I've never seen it leave the top of the volatility rankings. Then I apply a "less than daily" chart structure to that instrument─such as M20 or 10 big point/pip Renko, etc. and add trading logic. I refer to this as "dissecting the day." Again, the underlying premise is that an instrument can't very well sit still intraday if its daily volatility is high. And again, liquidity is obviously important, so I stay away from any exotics.

I guess that you could say that this is my version of strong versus weak but, no, I don't evaluate price direction at the level of instrument selection.

 

If I am building a general strategy like I just had an idea and I wanted to test it out, than I would probably backtest it on all of the pairs. But that doesn't mean that I pick most of the pairs. If all major pairs are loosing in my backtest and only 1 survives, that tells me I have messed up somewhere because the strategy should generalize. It's a sign of overfiting. So the answer is not that simple, it depends on the context. 


Otherwise, most of my strategies are built around a specific instrument. And that makes it really easy to choose what kind of pair I want to trade it on. Otherwise, if you just look at the stats of each pair and see why your strategy does well in a specific pair, you can figure out which pairs your strategy could perform well. 


On the other hand: There's a concept called Network Momentum. It's very technical but is the closest thing you will get to cross-market relationships that you are saying. It's a much more robust way to analyze "strengths" and "weakness" as it's a well known fact of the market that cross-market relationships exists. And you can take advantage of it with a specific technique called Network Momentum. It's not necessarily strength and weakness, but it's much better than you are looking for. 

 
Muhammad Minhas Qamar #:
If all major pairs are loosing in my backtest and only 1 survives, that tells me I have messed up somewhere because the strategy should generalize. It's a sign of overfitting.

IMHO, you're being a bit "loose" with the definition of overfitting in that part of your post. Overfitting merely connotes inadequate data sampling for an algorithm's intended purpose─compounded by an overly complex algorithm that's tailored to it. One of the most widespread fallacies is the belief that any system or strategy should be profitable when applied to any instrument. If that were true, wouldn't the same principle apply across all markets as well? Equities/stocks? Futures? Crypto? It doesn't. Even within the FX market, apply the average XAUUSD strategy to EURUSD and watch what happens.

Muhammad Minhas Qamar #:
Otherwise, most of my strategies are built around a specific instrument.

Correct.

 
Ryan L Johnson #:

IMHO, you're being a bit "loose" with the definition of overfitting in that part of your post. Overfitting merely connotes inadequate data sampling for an algorithm's intended purpose─compounded by an overly complex algorithm that's tailored to it. One of the most widespread fallacies is the belief that any system or strategy should be profitable when applied to any instrument. If that were true, wouldn't the same principle apply across all markets as well? Equities/stocks? Futures? Crypto? It doesn't. Even within the FX market, apply the average XAUUSD strategy to EURUSD and watch what happens.

Correct.


inadequate data sampling? No my friend that is not overfitting. And you are correct I was being loose in a casual way. In my context, I was talking about parameter overfitting. Which mainly refers to drastic change of behavior due to slight changes to the parameter(s). 


You are correct though that not every system can be profitable in other markets aswell. But profitability is not the sole metric when looking for generalization. As I said, it depends on the context of what you are trying to achieve. Markets like EURUSD and GBPUSD who are highly correlated, I would expect a strategy to work in both. I would not expect it to work on equities for instance. imho

 
Muhammad Minhas Qamar #:
inadequate data sampling? No my friend that is not overfitting.
Consider the situation where a trader has quality data going back to 1970 (the max in MT5). How exactly could that trader overfit code to a whopping 56 years of data? 
Muhammad Minhas Qamar #:
And you are correct I was being loose in a casual way. In my context, I was talking about parameter overfitting. Which mainly refers to drastic change of behavior due to slight changes to the parameter(s)

Ok, but slight changes to parameters is merely the process of optimization. Extensive changes to parameters (using multiple agents in Tester, for example) is over-optimization. A primer on overfitting is linked below.

Muhammad Minhas Qamar #:
Markets like EURUSD and GBPUSD who are highly correlated...

Instruments having different underlying assets are correlated until they're not. For example, GBPJPY was highly correlated with the S&P 500... until the 2008 Recession. To this day, GBPJPY has never fully recovered to the extent that the S&P 500 has recovered.

If you want to exploit correlation, check out EURUSD spot against EURUSD futures.

What is Overfitting? | IBM
What is Overfitting? | IBM
  • www.ibm.com
Overfitting occurs when an algorithm fits too closely to its training data, resulting in a model that can’t make accurate predictions or conclusions.
 
Ryan L Johnson #:
Consider the situation where a trader has quality data going back to 1970 (the max in MT5). How exactly could that trader overfit code to a whopping 56 years of data? 
What you are talking about is related to training sample. It's a workflow for ML. Overfiting is generally used to describe a model that does not generalize well. Whether or not you have more or less training sample will indirectly affect this, so it's not what overfitting is and by the way, you can pretty easily overfit "code" to even a century of data. Consider a modern ML architecture for transformers. It requires millions if not tens of millions of data points for it to reliably perform on out-of-sample tests. So unless you are training on ticks, there's a high chance of overfiting. 
Ryan L Johnson #:
Ok, but slight changes to parameters is merely the process of optimization.

Yes, but those slight changes should not affect drastic change in behavior. That's the point. 

Ryan L Johnson #:
Instruments having different underlying assets are correlated until they're not.
You are correct. With my example for EURUSD and GBPUSD the biggest example of this was during Brexit. But again my point was that a strategy should work well enough on pairs that behave the same. An event like Brexit can't be predicted anyways. 
12
New comment