Pairs Trading: Riding High or Risking All?

Key Takeaways

  • Pairs trading seeks profit by exploiting price disparities between correlated assets.
  • Issues arise when prices diverge, correlations are misestimated, model assumptions don’t hold and shorting restrictions limit trading ability.
  • LTCM's downfall highlights the dangers of overleveraging and inadequate risk management in pairs trading.

Pairs trading stands as a favored strategy within financial circles, renowned for its potential for quick wins and large returns. However, the substantial risks of this strategy should not be overlooked or underestimated, as we may be surprised by major losses. This article dives deeper into what pairs trading entails, how it can be performed and what can go wrong.

The Strategy

Pairs trading, first established in the 1980s by a quant team at Morgan Stanley, is a strategy rooted in statistical arbitrage where one aims to generate riskless profits by capitalizing on asset mispricings and market inefficiencies. By exploiting these disparities, traders not only secure gains but also contribute to market efficiency by correcting price discrepancies. As an arbitrage strategy, pairs trading is done by simultaneously buying an underpriced and selling (shorting) an overpriced asset, where the two are highly correlated. The difference in prices leads to an arbitrage profit as they converge.

Pairs trading has a relatively simple basis - pairing two assets with a strong positive correlation and price discrepancy, where we expect their prices to converge. This straightforward approach attracts not only hedge funds but also retail investors. To implement the strategy, computations of historical correlations of the assets are required – the goal is to spot differences in current correlations compared to the expected correlations, as per the historical estimations.

Figure 1 illustrates the approach; historically the correlation of assets A and B is strongly positive. Then prices begin to diverge and their current correlation at time t = 8 (see the x axis) is much lower than the correlation we expect based on historical data. Thus, it is time to execute the long/short trades (long cheaper, short expensive) and wait for the anticipated price convergence. At time t = 12 the prices have converged – the historical correlation is restored – and we can close out both positions, leaving us with a realized profit in the size of their price discrepancy. As it's indifferent to the direction of the broad market, pairs trading is a market-neutral strategy, one of its main advantages.

Figure 1: Price convergence of correlated stocks

Source: Quantified Strategies

Price Disparity

Now you may wonder: why do similar securities have different prices? Since pairs trading is an arbitrage strategy, the reason is (temporary) market inefficiency. While academics generally view markets as efficient, opportunities for arbitrage still emerge. Investors with superior information or competitive advantages exploit them, enforcing efficiency. For instance, identical securities from the same issuer may vary in price across exchanges.

What if prices don’t converge? Variations in pricing could stem from greater information disparities in certain markets. Additionally, a security's limited trading volume on a smaller exchange may warrant a liquidity risk premium, resulting in a discounted price for this asset. The price discrepancy is not a mispricing but rather reflects the disadvantage of not being able to sell the asset easily. Furthermore, investor behavior isn't always rational. Personal preferences or documented behavioral biases may lead to favoring one security over its identical counterpart, perpetuating the price gap – again a convergence of prices cannot be expected. The simplicity of the pairs trading strategy only works if the large positive historical correlations become restored after a price divergence, but this is not guaranteed.

What if prices diverge further? In this scenario, the price of the purchased asset decreases while the price of the sold (shorted) asset increases. This creates potential for unlimited losses, as the price of the shorted asset may continue to rise. The longer the investor delays closing out the short position in anticipation of price convergence, the greater the damage.

Trading Troubles

Non-convergence is not the sole hurdle obstructing the strategy’s success. Identifying assets with sufficiently high correlations (ideally above 0.8) is already challenging, but computing historical correlations brings its own issues. Determining the historical time period for computing correlations introduces variability. Return correlations can fluctuate over time due to evolving businesses and market-wide distress periods, which can skew historical correlations. Even stocks in the same industry may diverge during economic shocks.

Bringing it back to the basics, relying on historical data to gauge current correlation presents challenges, as it assumes history will repeat itself, which may not always be the case. Moreover, using statistical models to assess asset relationships introduces additional assumptions, each with its own limitations that may not align with reality.

In finance we stand by Markowitz's diversification to optimize risk-adjusted returns. In pairs trading a long and short position is taken on two assets with similar betas (due to their high positive correlation) which leads to a hedged position as the short asset’s beta offsets a large proportion of the long asset’s beta, leading to only low levels of market risk in the portfolio. However, since only two assets are involved in this strategy, idiosyncratic risk, which investors are not compensated for, is not diversified away, resulting in a sub-optimal portfolio with respect to mean-variance optimization. Investors should engage in pairs trading while also holding a large diversified portfolio, or attempt to trade many pairs of assets in different industries and geographies to not retain such high levels of idiosyncratic risk.

Lastly, the problem of non-perfect markets arises due to costs and fees associated with trading which can drastically cut any profit from the assets’ price convergence. Furthermore, there are legal restrictions on short sales and regulation is subject to change over time.

A Cautionary Tale

Before embarking on your pairs trading journey, consider the case of Long-Term Capital Management (LTCM). Led by renowned bond trader John Meriwether and Nobel laureates such as Myron Scholes and Robert Merton who are known for their derivatives pricing model, LTCM famously engaged in pairs trading. LTCM’s strategy involved pairs trading US, Japanese and European sovereign bonds, expecting convergence of mispriced securities. The fund took advantage of price discrepancies across exchanges and even traded sovereign bonds of emerging markets, backed by dollars.

The pairs of assets that would eventually lead to their downfall were on-the-run (most recently issued) and off-the-run (older, issued some time ago) US treasury bonds with the same maturities. Since they had the same maturity, by the law of one price, they should have the same price. However, LTCM found the newly issued bonds to be more expensive because of the larger demand at their issuance, while the longer-dated bonds of the same maturity had already been previously traded and so, the demand for them was not as great as for the fresh new bonds. This price discrepancy lead LTCM to buy off-the-run and short on-the-run treasury bonds of the same maturity.

Many large (institutional) investors endowed significant capital to the fund, resulting in $1.3 billion assets under management at its inception in 1994, and yielded over 40% of returns during its first two years. By the end of 1997, the fund managed around $7 billion, but only returned around 27%, comparable to US equities that year. At this point, Meriwether returned over $2 billion of the fund's capital back to investors because "investment opportunities were not large and attractive enough". However, since LTCM was highly levered to maximize profits with $125 billion of debt against $5 billion of equity, a leverage ratio of ca 25 (see Figure 2), the total portfolio under the fund’s management amounted to $100 billion with a net asset value of $4 billion and interest rate swaps valued at $1.25 trillion which is 5% of the entire global market.

Figure 2: LTCM’s leverage ratio

Source: Marc Rubinstein, Net Interest

However, the Russian financial crisis in August 1998 triggered a flight to quality, meaning demand soared for newly issued treasuries as longer-dated bonds were seen as riskier even though they were still “risk-free” treasury bonds. The large demand pushed prices up but LTCM was shorting these bonds; off-the-run bonds the fund had long positions on dropped in price due to lack of demand. Instead of converging, prices departed further from each other, leading to large losses as shown by Figure 3. Due to these losses, LTCM’s equity took large hits, decreasing to a mere $600 million, compared to its previous $5 billion.  

Figure 3: LTCM’s daily and cumulative profit and loss 

Source: Marc Rubinstein, Net Interest

Naturally, the market reacted to the crisis and loss news immediately, resulting in an almost vertical decline in the fund’s market value of equity, see Figure 4. Ultimately, a consortium of banks injected $3.5 billion to manage LTCM, averting a systemic meltdown since many large banks were invested with LTCM. The aftermath saw widespread financial damage, prompting President Clinton to address systemic risks in financial markets in April 1999.

Figure 4: LTCM’s market value of equity

Source: Edelweiss Capital Research

Even though it was the Russian crisis, which led to the materialization of market risks that were out of LTCM’s control, that caused its downfall, the magnitude of losses could have been much less severe with sufficient capital holdings and lower leverage levels.

Trade with Grit, not just Wit

Pairs trading can be your ticket to profits or your plunge into peril, it all depends on how you choose your assets and manage risk. Triple check your choice of securities, estimates and assumptions; this is where sensitivity analyses come in handy. Persistently monitor your risks and do not get overconfident. Pairs trading works best and at lowest risk when you focus on small frequent profits over time instead of trying to get rich quick.

Link 1 Link 2 Link 3 Link 4 Link 5 Link 6 Link 7 Link 8 Link 9