Interpolation
Interpolation is a mathematical technique for estimating unknown values within the range of a set of known data points, widely applied in finance to construct continuous yield curves and volatility surfaces from discrete market observations, to value instruments at non-standard maturities, and to fill gaps in time-series data. Linear interpolation assumes a constant rate of change between known points, while more sophisticated methods (cubic spline, log-linear, Nelson-Siegel) impose smoothness and economic constraints.
Key takeaways
- Linear interpolation is the simplest approach, computing intermediate values as a weighted average of adjacent known data points proportional to distance.
- Log-linear interpolation (interpolating in log-price space) is preferred for discount factors and zero-coupon bond prices, as it preserves positive prices and reflects compound interest mathematics more accurately.
- Cubic spline interpolation fits piecewise cubic polynomials through data points, ensuring first and second derivative continuity (smooth curves), making it suitable for yield curve construction.
- The Nelson-Siegel and Svensson models fit parametric functional forms to the yield curve, providing smooth, arbitrage-consistent curves suitable for central bank publication and risk management.
- In options markets, interpolation across strikes and expirations is used to build the implied volatility surface, with strict no-arbitrage conditions (calendar spread and butterfly constraints) required to ensure validity.
Explanation
Interpolation is an indispensable mathematical tool throughout quantitative finance, arising wherever continuous functions must be estimated from discrete observations. Interest rate markets present the clearest example: government bonds trade at specific maturities (3-month, 2-year, 5-year, 10-year, 30-year), but pricing, risk management, and hedging applications require continuously defined yield curves spanning every maturity from overnight to 50 years. Interpolation bridges the gaps between observed points to construct this continuous curve.
The choice of interpolation method has material consequences for practical applications. Simple linear interpolation between adjacent zero rates or par yields produces a kinked, piecewise-linear curve that creates artificial discontinuities in forward rates—the rate implied for a specific future period. In financial economics, forward rates represent the market's expectation of future short-term rates and should evolve smoothly. Log-linear interpolation of discount factors (P(0,T) = e^{-r(T)·T}) produces smoother forward rates and preserves the no-arbitrage condition that the forward rate must be non-negative. Cubic spline interpolation—fitting cubic polynomials on each interval such that the interpolating function is twice continuously differentiable—eliminates kinks and produces smooth forward rate curves, at the cost of potential oscillation in data-sparse regions.
The Bootstrap method is a widely used approach for constructing zero-coupon (spot rate) curves from coupon bond prices. Starting with the shortest-maturity instrument (e.g., a 3-month T-bill providing a direct 3-month spot rate), the method sequentially strips coupon bonds to extract zero-coupon rates at each successive maturity. Where no bond matures exactly at a desired interpolation point, the bootstrapped zero rates at adjacent maturities are interpolated to fill the gap. The resulting spot rate curve is internally consistent with all observed bond prices and can be used to value any fixed cash flow stream.
In options markets, interpolation of implied volatilities requires particular care because the resulting surface must satisfy no-arbitrage conditions that naive interpolation can violate. Interpolating IV linearly across strikes can produce negative probability density functions (butterfly spread arbitrage) or negative calendar spreads (forward-time arbitrage). Practitioners use several approaches to maintain arbitrage-free surfaces: parametric models (SVI—Stochastic Volatility Inspired—provides a tractable, arbitrage-free parameterization of the smile at each maturity), stochastic volatility models (Heston, SABR) calibrated to market prices, and regularization-based interpolation that enforces no-arbitrage constraints explicitly.
Beyond yield curves and volatility surfaces, interpolation appears throughout quantitative finance: in Monte Carlo simulation, where stock price paths may be needed at arbitrary time points between simulated dates; in scenario analysis, where risk factor moves at granular horizons are interpolated between stress-test scenarios; in historical data analysis, where missing prices must be filled for portfolio return reconstruction; and in derivatives valuation engines, where parameters must be interpolated from discrete calibration grids. Understanding the mathematical properties and limitations of each interpolation method—particularly the trade-offs between local accuracy, global smoothness, and no-arbitrage compliance—is a core competency for quantitative analysts in financial markets.
Formula
Linear Interpolation: y = y₁ + (x - x₁)/(x₂ - x₁) × (y₂ - y₁); Log-Linear (Discount Factors): P(T) = P(T₁)^[(T₂-T)/(T₂-T₁)] × P(T₂)^[(T-T₁)/(T₂-T₁)]
Example
A derivatives desk needs to price an interest rate swap with a 7-year maturity, but the observable market benchmark rates are for 5-year and 10-year swaps at 3.80% and 4.20% respectively. Using linear interpolation, the 7-year par swap rate is estimated as: 3.80% + (7-5)/(10-5) × (4.20%-3.80%) = 3.80% + 0.40 × 0.40% = 3.96%. Using log-linear interpolation of discount factors would give a slightly different result reflecting the compounding effect. For a more accurate result, the desk bootstraps the full zero curve from all liquid benchmark instruments (3M, 6M, 1Y, 2Y, 3Y, 5Y, 7Y, 10Y, 15Y, 20Y, 30Y swap rates) using cubic spline interpolation, obtaining a smooth zero curve from which the 7-year zero rate—and hence the swap's fair value—can be computed precisely.
Related terms
Arbitrage Bond Butterfly Spread Compound Interest Convexity Adjustment Copula Gaussian Copula Hedging Interest Rate Interest Rate Swap Monte Carlo Simulation Perpetuity