Spread & Pairs Trading

Trading the relationship, not the direction

Most strategies bet on direction — long or short. Spread and pairs trading is different: you bet on the relationship between two instruments returning to normal. If Stock A is usually 10% above Stock B and the gap has widened to 20%, you buy B and sell A, profiting when the spread normalizes — regardless of whether both stocks go up, down, or sideways.

Kaufman dedicates extensive coverage to this approach because it has a crucial advantage: reduced directional risk. When you're long one instrument and short a related one, broad market moves that affect both roughly cancel out. Your P&L depends on the spread, not the market.

Calendar spreads

The simplest spread trade: buy one futures contract month and sell a different month of the same product. For example, buy December wheat and sell March wheat. The trade profits when the price relationship between the two contract months changes.

Calendar spreads are popular in commodities because the spread between contract months reflects storage costs, interest rates, and seasonal supply/demand factors — all of which have fundamental logic and historical patterns. A grain trader knows that the harvest season compresses the spread between near and far contracts (supply surge), while a storage shortage can blow it out.

Advantages:

• Lower margin requirements (exchanges reduce margins for spread positions).
• Reduced volatility compared to outright futures.
• Fundamentally driven by carry-cost economics, not just speculation.

Disadvantage:

• Profit potential is smaller than outright positions.
• Spread dynamics can be disrupted by delivery constraints, squeezes, or regulatory changes.

Pairs trading: the equity version

Pairs trading applies the same concept to stocks. Find two stocks that normally move together — say, Coca-Cola and PepsiCo — and trade the deviation:

1. When the spread widens: go long the underperformer, short the outperformer.
2. When the spread normalizes: close both legs for a profit.

The key question is how to define "normally move together." Two common approaches:

Correlation-based pairs

Measure the historical correlation between two stocks. If the correlation is consistently high (above 0.7–0.8), the pair is a candidate. When their prices diverge more than usual, take the reversion trade.

Kaufman notes an important nuance: too-high correlation is actually undesirable.

Very high correlation (above 0.9) means the two stocks almost never diverge enough to create a tradeable spread. Very low correlation means they're not really related. The sweet spot for pairs trading is typically in the 0.4 to 0.7 range — related enough that divergences revert, but divergent enough that tradeable setups occur.

Cointegration-based pairs

The more rigorous approach. Two series are cointegrated if a linear combination of them is stationary (mean-reverting) even though each individual series is non-stationary (trending). Correlation measures whether two things move in the same direction; cointegration measures whether the distance between them reverts to a mean.

Example: two stocks can have low short-term correlation (they zig-zag differently) but high cointegration (the spread between them always reverts). For pairs trading, cointegration is more important than correlation.

The standard test is the Engle-Granger two-step or the Johansen test. These statistical tests tell you whether the spread between two instruments is genuinely mean-reverting or just happens to look that way in your sample.

Z-score entry rules

Once you've identified a pair with a mean-reverting spread, you need rules for when to enter and exit. The standard approach uses the z-score — the number of standard deviations the current spread is from its mean:

where $S_t$ is the current spread, $\bar{S}$ is the mean spread over your lookback window, and $\sigma_S$ is the standard deviation.

Common rules:

The ±2σ entry threshold comes from the same logic as Bollinger Bands: at 2σ from the mean, the spread is extreme enough that reversion is probable. But just like Bollinger Bands, the 2σ doesn't guarantee reversion — it can widen to 3σ or 4σ if a fundamental shift is occurring.

When spreads break

The biggest risk in spread trading: the relationship permanently changes. This happens when:

• One company in the pair faces a bankruptcy or fraud.
• A sector undergoes structural change (e.g., one company pivots to a different business).
• A merger, acquisition, or regulatory action changes the fundamental link.
• Commodity delivery logistics change (for calendar spreads).

When the spread widens past your stop-loss (e.g., z-score hits ±4), close the trade. Do not average down into a broken spread. Mean reversion requires a mean — if the mean has shifted, you're not reverting; you're catching a falling knife with both hands.

Risk management for spreads

Even though spreads are "hedged," they're not risk-free:

1. Leg risk. One leg of your pair can move violently while the other doesn't. You're still exposed to idiosyncratic risk in each name.
2. Liquidity risk. If one leg becomes illiquid (halted, thinly traded), you can't close the spread cleanly.
3. Margin risk. In extreme market stress, correlations spike to 1.0 (everything moves together) — but your specific pair might diverge even more as one name faces unique trouble.
4. Size asymmetry. If the two legs have different betas or dollar exposures, a "neutral" position might actually have directional exposure. Beta-weight or dollar-neutral the legs.

Quick check

What you now know

• Spread trading profits from the relationship between two instruments, not outright direction.
• Calendar spreads exploit contract-month price relationships in futures, with lower margin and reduced volatility.
• Pairs trading identifies stocks that move together and trades their divergences.
• Cointegration is more important than correlation — it tests whether the spread is genuinely mean-reverting.
• Z-score ±2σ is the standard entry trigger; z = 0 is the profit target; ±3–4σ is the stop-loss.
• When a spread breaks past the stop-loss, close it — do not average down into a potentially broken relationship.

Next: Cash Flow Shenanigans — Schilit's four techniques companies use to inflate operating cash flow, with real-world cases from Delphi, Enron, and Global Crossing.