Instrument choice (focused on FX, here)
What follows is a “brief” note on some ground rules that traders should be aware of when choosing pairs. Oversimplified as usual. Apologies to the expert traders.
The most important traits I am looking for in a pair are: volatility, liquidity and correlation.
Volatility because we trade a momentum/breakout system. Liquidity because I need low spread (remember: spreads are costs, and the lower your timeframes, usually the higher your trading frequency, so, the higher your costs.) Correlation …..where do I start?
So, simply put, I look at the average excursion (range) in pips (ADR, or ATR, or whatever: you pick yours) and the lowest spread and then I check the correlation.
This is usually a compromise. For example, EU and UY are extremely liquid, but they are way down the volatility list. Why is this important? Because, in systems like this, having an instrument that on average moves 100 points is better than having one that moves 30. Depending on the risk management method you use, it may seem to make no difference: if you use fixed fractional it is always (say) 1%. Right? It is a bit trickier than that. I do not have time here to go deep into this, but think about, for instance, broker’’s fees. 1% risk is 1% risk, but if on one instrument you can “only” trade 5 lots and on another you can trade 15 lots, other things equal, the latter is 3 times more expensive. Let’s stop here for now.
Once you have identified a bunch of suitable instruments, you have to analyse their correlation.
The importance of correlation would deserve several chapters in any trading book. I won’t try it here. Let’s look at an example. According to one measurement of volatility (and ignoring liquidity,) in one of my tables I have these pairs on top: GBPNZD, GBPAUD, EURNZD, GBPCAD, GBPJPY. Notice anything peculiar? Yeup! GBP is in almost all of them. Now if you were using these instruments, you would have an exposure to GBP that is multiplied by 4 (unless you divide the regular risk on these pair by four!) If the Governor of the BoE comes out with a “strange statement” and the GBP moves a lot, 4 of your 5 pairs will be going in exactly the same direction. This makes no sense in risk management terms.
Another thing to remember is that the higher the time frame, the higher the correlation. For example, on Daily charts AUDUSD and NZDUSD have a correlation of 94 (they are basically the “same thing”.) If you look at the 15m correlation, it is 74. Still very high, but a quarter less. It is even less on 5 and 1m.
So, let’s have a look at “my 3 pairs”: GU/UC have a correlation of -58 on 15 m and -10 on 1H, GU/EA is 7 for 15m and 58 for 1H, UC/EA is -20 for 15m and 5 for 1H. You can do worse than this.
This is just scraping the surface, as usual.
I am sorry to complicate things further, but I can’t resist! For the more advanced traders, I should add that there are situations where using correlated pairs may make sense. I am tempted to ask if somebody wishes to venture an explanation of why this may be, but I have a feeling there would be no takers....
What follows is a “brief” note on some ground rules that traders should be aware of when choosing pairs. Oversimplified as usual. Apologies to the expert traders.
The most important traits I am looking for in a pair are: volatility, liquidity and correlation.
Volatility because we trade a momentum/breakout system. Liquidity because I need low spread (remember: spreads are costs, and the lower your timeframes, usually the higher your trading frequency, so, the higher your costs.) Correlation …..where do I start?
So, simply put, I look at the average excursion (range) in pips (ADR, or ATR, or whatever: you pick yours) and the lowest spread and then I check the correlation.
This is usually a compromise. For example, EU and UY are extremely liquid, but they are way down the volatility list. Why is this important? Because, in systems like this, having an instrument that on average moves 100 points is better than having one that moves 30. Depending on the risk management method you use, it may seem to make no difference: if you use fixed fractional it is always (say) 1%. Right? It is a bit trickier than that. I do not have time here to go deep into this, but think about, for instance, broker’’s fees. 1% risk is 1% risk, but if on one instrument you can “only” trade 5 lots and on another you can trade 15 lots, other things equal, the latter is 3 times more expensive. Let’s stop here for now.
Once you have identified a bunch of suitable instruments, you have to analyse their correlation.
The importance of correlation would deserve several chapters in any trading book. I won’t try it here. Let’s look at an example. According to one measurement of volatility (and ignoring liquidity,) in one of my tables I have these pairs on top: GBPNZD, GBPAUD, EURNZD, GBPCAD, GBPJPY. Notice anything peculiar? Yeup! GBP is in almost all of them. Now if you were using these instruments, you would have an exposure to GBP that is multiplied by 4 (unless you divide the regular risk on these pair by four!) If the Governor of the BoE comes out with a “strange statement” and the GBP moves a lot, 4 of your 5 pairs will be going in exactly the same direction. This makes no sense in risk management terms.
Another thing to remember is that the higher the time frame, the higher the correlation. For example, on Daily charts AUDUSD and NZDUSD have a correlation of 94 (they are basically the “same thing”.) If you look at the 15m correlation, it is 74. Still very high, but a quarter less. It is even less on 5 and 1m.
So, let’s have a look at “my 3 pairs”: GU/UC have a correlation of -58 on 15 m and -10 on 1H, GU/EA is 7 for 15m and 58 for 1H, UC/EA is -20 for 15m and 5 for 1H. You can do worse than this.
This is just scraping the surface, as usual.
I am sorry to complicate things further, but I can’t resist! For the more advanced traders, I should add that there are situations where using correlated pairs may make sense. I am tempted to ask if somebody wishes to venture an explanation of why this may be, but I have a feeling there would be no takers....