Pairs and Ratio Trading
Learn how crypto pairs and ratio trades express relative views, estimate hedge ratios, distinguish correlation from cointegration and manage spread-break risk.
Reading progress — saved on this device
Pairs trading takes offsetting positions in related assets to express a view on their relative performance rather than the direction of the entire crypto market.
Learning objectives
- Define a spread or ratio and choose a defensible hedge ratio.
- Distinguish correlation from a stable mean-reverting relationship.
- Manage structural breaks, borrow/funding and asymmetric liquidity.
What it is
A simple pair trade might be long ETH and short BTC in a chosen ratio, expressing a view that ETH will outperform BTC. More formal statistical strategies model a spread using a hedge coefficient estimated from historical data.
A price ratio is easy to compute but assumes a particular hedge structure. Regression-based hedges can reduce common-factor exposure, but the coefficient changes over time.
How it works
High correlation does not imply the price difference or ratio mean reverts. Two assets can trend upward together while their ratio drifts for years. Cointegration is a stronger statistical concept, but even estimated relationships can break.
Hedge ratio should be linked to the objective: dollar neutrality, beta neutrality, volatility neutrality or statistical spread stationarity. These produce different quantities.
Short legs introduce borrow or perpetual funding costs and potentially squeezes. A theoretically neutral spread can accumulate large carry losses.
Liquidity asymmetry matters. If the smaller token gaps while the larger hedge remains liquid, closing both legs at the modelled spread may be impossible.
How to analyse and apply it
| Check | Why it matters | What to verify |
|---|---|---|
| Relationship thesis | Explains why relative pricing may be stable. | Use sector/economic linkage plus statistical evidence. |
| Hedge ratio | Controls common-factor exposure. | Specify dollar, beta, volatility or regression hedge. |
| Spread stability | Determines whether mean reversion is plausible. | Test out-of-sample and monitor structural change. |
| Carry/liquidity | Can dominate convergence. | Include funding, borrow, spread and close-out depth. |
A strategy is not complete until the signal, sizing, execution, invalidation and review process are explicit. Any discretionary override should be recorded so it can be separated from the tested rule set.
Worked example and thought exercise
Suppose a model estimates ETH beta to BTC at 0.70 over the selected window. A beta-neutral long £70,000 ETH / short £100,000 BTC may still have residual risk because beta is estimated and changes with regime.
If ETH underperforms for a fundamental reason—say a lasting fee-economics change—the spread may not revert to its historical mean. A stop based on spread statistics and thesis review is necessary.
Thought exercise: why is “these two coins are 0.9 correlated” insufficient evidence for a mean-reversion pair trade?
Common mistakes and practical workflow
- Treating correlation as proof of cointegration or mean reversion.
- Choosing hedge ratio solely to make a backtest look smooth.
- Ignoring funding/borrow on the short leg.
- Averaging into a spread after a genuine structural break.
Practical workflow
- State the economic relationship being traded.
- Choose and document hedge-ratio objective.
- Test spread behaviour out of sample with costs.
- Set entry, exit and structural-break limits.
- Re-estimate carefully on a schedule rather than changing β mid-trade to rescue the thesis.
✅ Knowledge checkpoint
- Why does high correlation not guarantee a stable price ratio?
- What are different objectives for choosing a hedge ratio?
- How can funding costs affect a pair trade?
- What is the key risk of assuming every extreme z-score will revert?
FAQs
❓ Is pairs trading market-neutral?
It can reduce common market exposure, but residual beta, basis, liquidity and model risk remain.
❓ Should I use ratios or regression spreads?
Both are possible. The choice must match the hedge objective and be tested consistently.
❓ What is cointegration?
A statistical property where a linear combination of non-stationary series may be stationary; it is stronger than simple correlation but not a guarantee of future stability.
❓ Can the relationship break permanently?
Yes. Fundamentals, token design and market structure can change, creating a structural break.
📋 Summary
Pairs trading replaces an outright directional thesis with a relative one, but it introduces model and relationship risk. Robust strategies define the hedge objective, verify spread stability, include carry/liquidity costs and accept that some historical relationships break permanently.
Want this in a personalised order?
Take the crypto assessment and get a custom path of 10 modules matched to what you already know. Free, no card required.
Build my path →