Pairs Trading: Definition, Strategie und Beispiel

Pairs trading is a popular strategy that involves matching a long position with a short position on securities with a high correlation. Learn more about pairs trading and the risks involved.

What is a pairs trading strategy?

A pairs trading strategy is a means of taking a market-neutral position by finding two assets with a high correlation that are currently trading outside of their normal range.

To perform a pairs trade, you’d go short on the asset that’s trading above the historical range and long on the asset that’s trading below it.

It’s based on a similar concept to mean reversion trading – which assumes that assets trading above or below their average will always revert to the mean over time. The correction doesn’t happen immediately, which is why pairs trading (and mean reversion) can be both a long-term strategy and a short-term strategy depending on the conditions.

Pairs trading was first introduced in the 1980s by a group of analyst researchers at Morgan Stanley who used statistical and technical analysis to find market-neutral profits.

Does pairs trading work?

Pairs trading can work but you’ll need to have a lot of historical analysis backing any decisions up. The two assets need to have a high positive correlation that is confirmed over a long period of time.

Once this correlation falters, that’s when a pairs trade can be deployed, to take advantage of the move back to the historical pattern.

Pairs trading is a common hedging strategy because if both assets rise in value, one trade will make a profit while the other makes a loss – the same for if the assets fall – creating a market-neutral position. This means that the trade will neither make nor lose money.

In an ideal scenario, the overvalued asset should fall in price, while the undervalued asset would rise – and the trader would profit from both positions. But it’s more common for profit to be based on the relative returns: one of the positions profiting more than the other loses. 

How to identify pairs trades

Most pairs trade correlations are computed using software that identifies the ‘correlation coefficient’, a value that sits between -1.0 and +1.0. If the correlation is perfect, then it would have a value of +1.0, and if the assets move in completely different directions – meaning there is no correlation – they’d have a value of -1.0.

Usually, to qualify as a pairs trade, a trader would be looking for two assets to have a correlation of +0.80. But where to start?

To find a pairs trade, it’s a good idea to look at assets you think will be statistically correlated based on – essentially – an educated guess. Without narrowing down your search parameters using common sense, you’ll end up having to review every single possible combination of assets.

Typically, the assets involved in a pairs trade will be two securities with a clear economic link – for example, companies in the same sector, which manufacture the same products, such as Pepsi and Coca-Cola.

Example of pairs trading

Pepsi and Coca-Cola both manufacture beverages. The ties between the companies' income streams, business models and supply chains mean that the movement in the price of one company’s shares should be fairly like that of the other.

Naturally, there’s a normal distribution, meaning the share prices aren’t identical given other factors that can come into play. But there is a correlation. In fact, over the three months ending August 16^th, the two companies had a correlation of 0.92 – meaning that the price of Coca-Cola has a relative statistical influence on the price of Pepsi.

So, a pairs trader might look to diversify risk by combining positions on Coca-Cola and Pepsi. For example, if Coca-Cola was trading below its range and Pepsi above it, they might match a long position in Coca-Cola with a short position on PepsiCo.

Disadvantages of pairs trading

While pairs trading can result in reduced market risk and even profit, there are a few drawbacks to the strategy that it’s important to acknowledge.

Firstly, it relies on an extremely high correlation between the two assets in question. If there isn’t a concrete statistical relationship, the strategy will fail. Finding these can be challenging, as it requires extremely comprehensive and clean data sets.

Second, historical prices are no guarantee of future results. This means that even if a correlation has proven true over time, there’s no way to completely predict how the assets’ prices will move in the future.

What is machine-learning-based pairs trading?

Machine-learning-based pairs trading is the idea that technology that uses components of artificial intelligence will one day be able to identify opportunities for pairs traders.

Unlike the current computer programs used in algorithmic trading or automated trading, these machine-learning programs would be completely unsupervised and be able to adapt to changing market conditions.

Currently, the technology is theoretical but there’s a lot of interest in the future of AI trading.