Pair Trading & StatArb

Price-wise assets can be related to each other. This applies to financial assets, such as corporate shares or bonds, but also to physical assets, like commodities. Agricultural commodities are, for instance, related to energy for a variety of reasons. One reason concerns biofuels being created from agricultural products, leading to the substitution effect. After all, the price of corn may go up because the price of crude oil has gone up, such that it becomes interesting to replace gasoline created from petroleum by bio-ethanol created from corn. Obviously, this leads to a price effect for corn (due to substitution of crude for corn). This relationship creates trading opportunities, by means of taking a position in a cross-commodity spread. Investors and speculators analyse price relationships to setup positions in order to profit from expected changes. Pair trading and statistical arbitrage are strategies applied by investors and speculators.

Pair trading

A product and its related ‘alternative’ is called a ‘pair’. ‘Pair trading’ is a form of spread trading, as one product is bought, whereas the other product is sold short simultaneously. With a pair, the second leg can be seen as hedge on the first leg, or vice versa. After all, no outright position, but a combination is set up. No exposure to the markets as a whole is taken, but to the differential. This implies there is market risk, but relatively little.

Example:

An example of pair trading is the combination WTI crude oil futures contract and Brent crude oil futures contract as both concern crudes, both of them are of ‘about’ the same quality, and both are to be processed in the same types of refineries. Therefore, they can be considered substitutes of each other and, as a result, they can be seen as proxies for each other. All-in-all, this leads to the situation that prices of both physical products (and the prices of the related futures contracts) tend to move into the same direction, and to the same extend, at the same moment. This is not always the case; there are exceptions. After all, the products are not identical. Hence, the price correlation is not +1.00, but might be closed to it.

The price differential between two commodity futures contracts which form a pair could be rather constant over time, or may have some sort of ‘mean reverting’ level. If fundamentally nothing has changed, a market participant can expect the price of one of these products to rise if the other has already gone up significantly (‘lagging effect’). This party can try to profit from this expected scenario. After all, the purpose of spread trading is to profit from mispricing, or price discrepancy. Alternatively, two prices may no longer comply with their historically-based relationship, due to a change of the underlying fundamentals, leading to a new price relationship. In such a case, there may be left no opportunity to jump on. If, however, market fundamentals have not changed and the prices of two strongly correlated assets do not move in line, then, a market participant may profit from the situation (of expected temporarily mispricing) by setting up a product spread.

Statistical arbitrage

In case of ‘pair trading’, investors and speculators try to profit from price differentials that do not meet the statistical relationship. They do so by buying the under-valued product and selling the over-priced product. This is also the fundament of so-called ‘statistical arbitrage’, because a market participant tries to profit from a temporarily situation that does not hook up to the long-term statistical relationship between the prices of the two (or more) assets. The ‘mean-reversion’ paradigm is typically associated with market over-reaction; assets are temporarily under- or over-priced with respect to one or several reference market-prices.
Please note, however, that although it may be called statistical arbitrage it is not risk-free like the pure form of arbitrage (deterministic arbitrage). Moreover, with statistical arbitrage one is exposed to market risk. After all, it is not guaranteed that the spread indeed is going to move the way the market participant foresees. Moreover, it may turn out to move adversely.
Nevertheless, if the spread does move in the preferred direction, the market participant can liquidate his position at a profit. The process of setting up a spread and liquidating it might be done in a split second, but sometimes market participants (need to) wait a relatively long time; possibly days, weeks, or even longer. This length of time depends on various, including:

  • Spread volatility
  • Market liquidity
  • The market participant’s objective
  • The market participant’s expectations
  • The market participant’s mandates (limits)

To conclude, it can be said that pair trading is widely assumed to be the ancestor of statistical arbitrage. It is a statistical arbitrage strategy that allows executing transactions to capture anomalies, relative strength or even fundamental differences on two assets while maintaining a market neutral position.