In this story I want to go over what is, how to calculate and how to use correlations in asset’s valuations to your advantage. I will also mention how to use this knowledge to hedge your positions and manage risk.
What is a correlation
Correlation is a statistical term to compare the similarities in direction and strength of volatility between two data sets. Positive correlation means that two data sets compared to each other follow each other’s trends. Negative correlation is the opposite — data sets go against each other’s trends.
To give an example (not related to trading yet) if we take set A={1,2,3,4,5} and set B={2,4,6,8,10}, we can easily tell these sets are linearly correlated. Volatility throughout next elements of both sets is exactly the same. We can even plot the set B as a linear function of set A like this:
We can see this easily because the set is small and the correlation if perfect; every element of set B is twice the element in set A. An example of a perfect negative correlation would be:
Correlations don’t have to be perfect — and in the real world they very rarely are. In the markets some assets are correlated to each other but it never is a perfect correlation. For example we can notice that in general, Ethereum and Bitcoin go up and down in similar times. Rarely one rallies and the other does nothing or goes down. There is some correlation between the two… but how do we measure it and why would anyone care?
Pearson Correlation Coefficient
This is the name of the most popular statistical method to measure the correlation between two data sets. It’s an equation that has two, equal in size data sets as an input and it outputs a number between -1 and 1 that is our measure of correlation.
But we can break this down a bit to make it simpler and easily understood.
In the numerator we have covariance. It’s a measure of joint volatility of X and Y data sets. If both sets have similar direction and strength of the volatility (correlation); the sum will get larger as the sums of every iteration (taking on next elements of the set) won’t be canceling each other out.
In the denominator we just multiply the standard deviations (measure of volatility) of both sets. We divide the covariance by standard deviations to normalize the results to a universal correlation scale of -1 to 1.
We interpret the results as follows: Positive number means a positive correlation; and negative means a negative correlation. The closer the number is to 1 or -1 means the correlation is stronger and more meaningful. If the correlation is exactly 1 or -1 we can talk about a perfect linear correlation. If it’s close to 0 we talk about a statistically insignificant correlation.
Correlations in the financial markets
Financial markets often have correlations between each other. The “data sets” we analyse are the prices of asset’s over a period of time. Markets have a tendency to be correlated because of the fact that all markets are often traded based on macroeconomic data, models and predictions, that are universal to many markets at once. Niche markets like crypto are correlated not only to other markets like stocks and indices but also correlated internally. When the sentiment changes about cryptocurrency market as a whole; they all move together.
Let’s give an example and analyse some cryptocurrencies’s correlations. Image below is a “correlation table” that is used to compare correlations of different assets at once.
Here we have Bitcoin, Ethereum and Chainlink correlation table. When analyzing PCC we care about everything below the main diagonal of this table, which are the three squares in green. We can see all of these asset’s price action is correlated pretty strongly as a 0.8 PCC is considered “strong”. BTC and ETH are the most correlated out of all of these three pairs with Chainlink being less correlated to the two (but still quite strongly).
If this were to continue we can expect BTC and ETH to have similar moves in the same direction over a period of time.
Hedging your positions — way to utilize market correlations in the markets
Ok, but how to use allthis in trading?
Correlations become useful when creating hedged portfolios. Hedged, meaning you simultaneously run positions that are inversely correlated to each other. This is because you want to limit overall market volatility effect on your positions. If you trade inversely correlated pairs, then losses from one position will be covered by the gains from the other position.
You want to look for strong correlations, and differences in possible trade outcomes. Then you pick your positions to maximize your risk adjusted returns on both positions that are correlated inversely. So if you are trading positively correlated assets, go short on one and long on the other (depending on what your analysis tells you would be more profitable) and if you are trading inversely correlated pairs you can go long or short on both.
This method allows you to limit your exposure to overall market volatility and trade based on analysis like: “asset X will outperform asset Y”. This way you don’t have to care if a market overall goes up or down; if your analysis about the performance of assets against each other was right you will make money as most market moves will hedged against each other, canceling out.
Hedged positions
To create a hedged position in the market you need to do a few things.
- Have a trade idea. This is the basics of all trading, you need something that could make you money in the market. You have to get an idea about a stock or a coin being under or overvalued and positioning yourself to benefit from the market correcting to your “fair value”.
- Find a hedge. You need to find another idea, this time it’s not unlimited. You have to pick a trade that is inversely correlated to your initial idea. Think what could under-perform your trade and is correlated with it substantially. In my opinion it should have a PCC > 0.75. You don’t want it too high either (like greater than 0.95) because it is impossible to generate any profit from a perfectly-inversely correlated price action as all your profits will be taken by losses from the hedge.
- Calculate the exposures. If you want to minimize the overall volatility of the market the exposures of your positions should be 1:1. It’s pretty simple you have to allocate the same amount of capital to both positions. If you are more sure of one trade in a pair than the other, you can increase it’s exposure but that will increase exposure to the overall market volatility as well.
- Calculate the volatility to adjust overall trade risk. This one is a bit more tricky. You need a volatility metric to determine the average volatility of both market pairs you are trading. I like to use a standard deviation divided by the average valuation over the same periods of time. Then i multiply the inverse of the absolute value of this pair’s PCC by the average volatility of both. This gives a rough estimate of the overall pair volatility over a period of time. Notice, that the closer the correlation is to 1 (or -1) the lower the overall volatility will get. You can use leverage on the whole position to artificially increase the volatility if you think it’s necessary.
After volatility and exposures are calculated and setup to your liking you can enter the trade and see how it plays out. Hedging trades like this are often good for higher time frames like 2–6 months (depending on the market), to let the analysis play out.
So… this is it for today’s story. If you have any questions about this topic feel free to ask on my twitter or in the comments.
Twitter: https://twitter.com/basedcrypto77
Free correlation tables for crypto: Cryptowatch
