Skip to content
Financial Context Wiki

Basket Options: Weights, Correlation, Rebalancing, and Multi-Asset Payoffs

For educational purposes only; not investment advice.

Direct answer

Section titled “Direct answer”

A basket option pays on the value or return of a specified weighted group of assets rather than on one security. A simple fixed-strike call has payoff max(B_T − K, 0) × multiplier, where the basket B_T is calculated exactly as the contract defines.

For normalized return weights, one common definition is B_t = B_0 × Σ[w_i × (S_i,t / S_i,0)], with weights summing to one. Other contracts use fixed share quantities, market-cap weights, price weights, currency-adjusted values, caps, floors, or scheduled rebalancing. “Basket” is not a complete payoff specification.

Why correlation matters

Section titled “Why correlation matters”

Basket variance depends on every component’s volatility and every pairwise covariance:

σ_B² ≈ Σ_i Σ_j w_i w_j σ_i σ_j ρ_ij

This return-based approximation assumes stable weights over the measurement interval. Lower or negative correlations can reduce basket volatility below constituent volatility; correlations rising toward one remove diversification. Since a call is convex, changes in basket variance affect option value even when each constituent’s individual volatility is unchanged.

For two equally weighted assets each with 30% volatility, estimated basket volatility is:

  • At correlation 0.2: √(0.25×0.30² + 0.25×0.30² + 2×0.25×0.30²×0.2) ≈ 23.24%.
  • At correlation 0.8: the same formula gives ≈ 28.46%.

The increase is pure correlation input. Real baskets add changing weights, volatility skew, jumps, dividends, borrow, forward curves, and cross-currency dependence. Implied correlation inferred from constituent and basket options is model dependent, not one observable number.

A basket option differs from a strip of single-name options: max(Σw_iS_i − K, 0) is not equal to Σw_i max(S_i − K_i, 0). Gains in one constituent can offset losses in another before the basket option payoff is applied. It also differs from a worst-of option, whose result is driven by the weakest constituent rather than the weighted aggregate.

Weighted-return basket example

Section titled “Weighted-return basket example”

Suppose a basket begins at 100 and contains three normalized assets with weights 50%, 30%, and 20%. At maturity their levels relative to initial values are 120%, 90%, and 80%:

B_T = 100 × (0.50×1.20 + 0.30×0.90 + 0.20×0.80) = 103

A basket call with strike 100 and multiplier $1,000 per basket point pays max(103 − 100, 0) × $1,000 = $3,000. The strongest constituent gained 20% and the weakest lost 20%, but the weighted basket gained only 3%.

A “worst-of” call or note on the same names would examine the 80% constituent, not the 103 basket level, and could pay nothing or suffer a loss depending on its terms. Buying separate calls on each name would also produce a different payoff because positive single-name convexity is not netted against the losing assets before exercise.

If the contract instead holds fixed share quantities, weights drift as prices move. If it resets to 50/30/20 monthly, it sells relative winners and buys relative losers at each rebalance. These two baskets can end at different values despite the same constituent endpoints because the rebalanced version is path dependent.

Contract and model checklist

Section titled “Contract and model checklist”
  • List every component, identifier, exchange, currency, initial level, weight or share quantity, and price source.
  • Specify whether weights are fixed, drifting, rebalanced, market-cap based, or subject to caps and replacement rules.
  • Record dividend, corporate-action, delisting, merger, disruption, holiday, stale-price, and missing-component treatment.
  • Confirm FX conversion rate, fixing time, quanto feature, settlement currency, multiplier, and rounding.
  • Reproduce the basket level independently for initial, observation, and final dates.
  • Build the complete volatility and correlation matrix; check that it is mathematically valid and economically plausible.
  • Calibrate constituent skews and basket quotes at consistent timestamps; stale single-name quotes can imply impossible correlation.
  • Stress all correlations toward one, sector-specific divergence, one component near zero, jumps, and trading halts.
  • For rebalanced baskets, simulate the exact rebalance calendar, transaction rules, and path-dependent weight changes.
  • Include counterparty credit, collateral, liquidity, dealer unwind cost, model reserves, tax, and calculation-agent discretion.

Hedging requires trading multiple assets, potentially across timezones and currencies. Component markets may close at different times or become illiquid together. A theoretically defined payoff can therefore have substantial basis and execution risk even when the basket formula is simple.

Common misconceptions

Section titled “Common misconceptions”
  • “A basket is automatically diversified.” Concentrated weights and rising correlations can erase diversification.
  • “Basket volatility is the weighted average of volatilities.” Covariances and squared weights are essential.
  • “A basket option equals several single-name options.” Convexity is applied after aggregation, not separately.
  • “Basket and worst-of mean the same multi-asset exposure.” One uses an aggregate; the other is driven by the weakest asset.
  • “Weights stay constant.” Fixed-share baskets drift, while rebalanced baskets become path dependent.
  • “Historical correlation is sufficient for pricing.” Pricing uses forward-looking joint-distribution assumptions and smile dynamics.
  • “A published basket level eliminates contract risk.” Component substitutions, disruptions, FX, and calculation-agent rules still matter.
Section titled “Related topics”

Authoritative sources

Section titled “Authoritative sources”