Valuation guides

Payoff scripting explained

What payoff scripting is, why banks use payoff languages, an example and how scripted payoffs are valued by Monte Carlo.

Definition

Payoff scripting describes a product's terms in a small programming language: observation dates, barriers, coupons with memory, autocalls, issuer calls and holder exercise. The same engine then prices any product written in the language, rather than needing a new pricer for each structure.

History

Banks built payoff languages in the late 1990s and 2000s to handle the growing variety of structured products. Peyton Jones, Eber and Seward (2000) described composing contracts from building blocks. Savine's 'Modern Computational Finance' (2018) describes scripting with automatic differentiation for risk.

Examples

A memory coupon

A script can say: on each date, if the worst share is above 70%, pay 2% plus any coupons missed earlier; if all shares are above 100%, pay back the notional and stop. The same engine then values it on correlated paths.

Variations

Scripted payoff
Any payoff in the language, on one to five assets.

Valuation methods

The script is run on all Monte Carlo paths at once for one to five correlated assets, under Black-Scholes, Dupire local volatility, Heston, jump or Lévy models, with discrete cash dividends. Calls and holder exercise are valued by Longstaff-Schwartz regression.

Models on RiskAtoms

InstrumentPricing models
Scripted payoff
Payoff scripting language on Monte Carlo
Monte Carlo (correlated lognormal)
Local volatility Monte Carlo
Heston Monte Carlo
Longstaff-Schwartz regression (early exercise)
Merton jump-diffusion
Bates (Heston with jumps)
Variance Gamma
Normal Inverse Gaussian (Lévy)
Kou double-exponential jump diffusion
CGMY tempered stable (Lévy)
Rough Bergomi (rough volatility)
Heston stochastic local vol, Monte Carlo
Discrete cash dividends on paths