# Part I. Intramarket Correlations

Commander in Pips: Once we’ve discussed different relations between the forex market and other markets – such as financial and commodity markets, but I want to make small add-on. Still, there is a relation inside forex market itself – of one pair to another. That’s the topic that I intend to show you for further discussion. We call it as Intramarket correlation, while the previous one was Intermarket correlations and relations…

Pipruit: Hm, sounds interesting. I also suspect that some pairs have a solid relation of motion to each other, but not always in the same direction. Some pairs move simultaneously in the same direction, while others move in opposite directions, although they form almost mirror charts.

Commander in Pips: That’s right. By that observation we could say that there is some relation between currencies that exists. Even more, since currencies trade in pairs – if we speak about Forex and not about, say, the Dollar Index, then they just can’t be totally isolated from each other. For measurement of the strength of relation there are different statistical coefficients that exist and one of the most popular is correlation. This coefficient will show the strength of relation between different pairs.

Pipruit: And why we need to know it?

Commander in Pips: Mostly because this is quite important if you trade multiple pairs on your account. If you focus on some particular pair – this might be not so significant to you. But if you trade many pairs and do not know how they relate to each other – it’s a wash out. If some pairs have a solid relation and move together – this can significantly increases your risk if you trade both of them. So, the major application of correlation coefficient is risk management sphere.

Pipruit: I see. And how it works, your correlation stat?​

Commander in Pips: Well this is very simple. We will use an Excel work sheet for that purpose. First, we need quotes for some period of time on two different pairs. Let’s say we want to estimate the correlation between daily charts for EUR/USD and GBP/USD pairs from November 1st. First, we have to extract from our software quotes data of both pairs. Quotes should be on the same time frame – daily EUR/USD – daily GPB/USD, weekly - weekly and so on. Then you have to use build-in function CORREL in Excel – and tada! – we see correlation for the period November 1st till December 14th of 0.93 or 92.86%. This is a very significant correlation. The most important thing is that this correlation will not remain static – it will change over time. Even more, the fact that EUR and GBP correlation on the daily chart is 0.92 for our sample does not mean that this correlation will be the same for weekly, monthly or, say hourly time frames. It will be different. As well, correlation will change as you will use different data array. We use 1.5 months of data (from 1st of November till 14th of December). If you will use, say a 3 month array of data – then the correlation will be different.

Very often correlation is shown by correlation tables. They can be two kinds – for two pairs but for different time frames, or for single time frame but for multiple pairs:

Here is first example – when correlation table (or matrix) was created for the same period of analysis (say, 3 months), quotes of all pairs were daily. So, on crossing of any column and row we see corresponding Correlation. For example on crossing EUR/USD and GBP/USD we see our number of 0.9286.

Pipruit: And why EUR/USD to EUR/USD correlation is 1? And for other repeated pairs? Why EUR-USD correlation equals to GBP-USD correlation?
Commander in Pips: Hm. Since EUR/USD is repeated itself, that’s why it totally related with itself, hence correlation equals 1. Now, since EUR/USD to GBP/USD shows the same relation as GBP/USD to EUR/USD, then their correlation is the same. This is obvious once you understand how it works.

Pipruit: Right. And what is the second type table?
Commander in Pips: The second kind of table creates for single pair but in different time frames. It usually looks like this:

In fact, you may use any terms for which you would like to calculate correlation, depending on your needs. If you are short-term trader, probably you will be interested in 1 week correlation, while if you hold position long-term, then you probably will need 6 moths or even a 1-2 year correlation.

Pipruit:
Ok, I see. The major thing that I have to note is that high numbers of correlation tells that pairs mostly move in the same direction while negative number close to -1 tells that they move equal but in counter directions. If correlation is close to zero – then there is an absence of any relation and pairs move independently.
Commander in Pips: Yes, you’re absolutely right. But what will be application in trading?

Pipruit:
Hm… Really. How we can use it?

Commander in Pips: Ok, let’s turn to some examples:

Chart #1 | EUR/USD and GBP/USD 60-min
Just to confirm our calculation, although it was made on daily quotes – here you can see 60-min charts of EUR/USD and GBP/USD. They move quite simultaneously and are very harmonic. Let’s suppose that you think – “I will open one position with EUR/USD and another one with GBP/USD – this will be diversification and risk reducing issue.” In this case you will be wrong, since high correlation tells us that adding of GBP/USD will not give you any risk reduction – instead, it's vice versa. It will increase your risk, because the EUR and GBP very often move in the same direction. I tell you what, I give you a bit naïve understanding of correlation, but it might be useful and let you better understand this kitchen. Since we have it at 0.93 – treat it as replacement of 1 lot of GBP/USD with 0.93 lot of EUR/USD. Now imagine that if you will open 1 lot in EUR/USD and 1 lot in GBP/USD – this will be equal to 1.93 lot of EUR/USD. So, you almost double your risk, not reduce it!

Pipruit: Well, and what if I open 1 Long lot EUR/USD and 1 short lot of GBP/USD?

Commander in Pips: Well, with our simplified understanding this will give you just 0.07 lot in EUR/USD, or we can round it to zero – but this position has no sense. But, you should not invent the bicycle here – just enter long with EUR/GBP, if you want. Second, since correlation of these pairs is significant, their volatility might be different – and you probably will end your trade with some small loss or profit. Still this trade looks doubtful. You will not be able to predict corresponding moves of these pairs. One way how correlation has to be used here – is for understanding that your risk will double and adjust your trading lot accordingly if you still want to enter the same direction with EUR as with GBP – to match your risk management system.

Pipruit:
Probably you’re right, I understand. Enter with 2 lots of EUR/USD – will be almost the same… or with 2 lots of GBP/USD.
Commander in Pips: Absolutely.

Pipruit:
Wait a minute, let me guess. If we will take highly negatively correlated pairs – for example EUR/USD and USD/CHF and take opposite positions with them – look chart below… This will be the same as taking one direction positions on positively highly correlated pairs, right? So enter long with EUR/USD and simultaneously short USD/CHF will give us the same as enter long EUR/USD and GBP/USD or enter with twice the size long on EUR/USD?

Chart #2 | EUR/USD and USD/CHF Weekly
Commander in Pips: You understand everything correctly. If we speak about it in terms of risk and correlation – that’s it. The major thought that we are discussing now is that we have to understand and recognize in time when we increase our risk exposure. This could seem that we reduce risk when open position with 2 different pairs – in reality we should look at this through correlation. From the other point of view – open opposite positions in highly positively correlated pairs or same position in highly negatively correlated pairs have no special sense – because you will just cancel your initial position by this action. You may finish with some loss or even profit, but it will be small and hardly could be predicted. You may even finish with a loss on both positions.