Cliquet Options and Structured Products

A cliquet option aggregates periodically reset returns, typically applying a local cap and floor to each period and sometimes a global floor to the total. It turns a long horizon into a path-dependent series of short-dated options, so forward volatility, serial dependence, and the treatment of caps dominate its economics.

Payoff construction

For reset dates t0 ... tn, define period return Ri = S(ti)/S(t_{i-1}) - 1. A common cliquet coupon is

C = min(global_cap, max(global_floor,
    sum_i min(local_cap, max(local_floor, R_i))))

Many retail notes use a local cap with no local floor and a global guarantee; others include participation factors, averaging, memory coupons, or issuer-call rights. “Cliquet” therefore identifies a payoff family, not a complete term sheet. Model code should represent its payoff in a declarative schedule and preserve reset observations once fixed.

Element	Economic effect
local cap	sells upside of each period
local floor	limits each period’s loss
global floor	embeds long-horizon downside protection
reset frequency	sets forward-volatility exposure
participation	scales returns before aggregation

Replication intuition

Ignoring global constraints, a locally capped and floored return can be decomposed into a forward return plus short call and long put spreads around the reset:

min(c, max(f, R)) = R - max(R-c, 0) + max(f-R, 0)

Each reset therefore involves forward-starting option exposure. A local cap makes the investor short upside convexity over every period; the dealer issuing the note is correspondingly long it. The global floor couples periods and prevents a simple static sum of vanilla options, especially after partial returns have been fixed.

This decomposition explains why a cliquet is not priced correctly by a single terminal implied volatility. It needs the joint distribution of future returns at each reset, including the forward smile. A surface calibrated only to spot-starting options leaves that distribution weakly identified.

Simulation and state management

Monte Carlo is flexible for global floors, memory features, and callable terms. Simulate under the risk-neutral model, apply each reset payoff, and discount the final cash flow. Use antithetics and control variates, but do not smooth away the actual cap/floor discontinuities when computing risk.

import numpy as np

def cliquet_payoff(spots, local_floor, local_cap,
                   global_floor, global_cap):
    returns = spots[:, 1:] / spots[:, :-1] - 1.0
    clipped = np.clip(returns, local_floor, local_cap)
    total = clipped.sum(axis=1)
    return np.clip(total, global_floor, global_cap)

def pv_from_paths(spots, r, T, **terms):
    return np.exp(-r*T) * cliquet_payoff(spots, **terms).mean()

For a live trade, known resets should be inserted exactly and only future periods simulated. The remaining global floor can make the profile sharply nonlinear: a negative accumulated return can turn future periods into a digital-like race to recover. Report the conditional price by accumulated-return bucket, not just one aggregate PV.

Model selection and calibration

Local-volatility models reproduce today’s vanilla surface and often price cliquets higher or lower than stochastic-volatility models depending on the forward skew they imply. There is no model-free answer. Compare plausible calibrated models, examine forward-starting smile predictions, and reserve for model dispersion where no liquid hedge exists. Heston stochastic volatility provides one stochastic-volatility baseline.

Correlation is also relevant for basket cliquets. Dispersion across constituents affects the distribution of each reset and the probability that global bounds bind. Correlation skew, dividends, FX conversion, and index rebalancing must be modeled at the same level of fidelity as the advertised payoff.

Hedging structured notes

Delta evolves around each reset and can jump when a local cap or floor becomes likely to bind. The vega profile is distributed across forward periods; hedging only the note's final maturity leaves early-reset risk. Dealers often use strips of listed options, variance exposure, and dynamic delta hedges, but gaps, liquidity, and smile moves create residual P&L.

An issuer's note economics also include funding, credit spread, distribution fees, and hedging friction. The derivative PV is not the investor’s fair yield by itself. Governance should separate the model value, hedging reserve, issuer credit risk, and sales margin.

Key takeaways

• Cliquets aggregate capped and floored periodic returns, creating forward-starting option exposure.
• Local cap/floor decomposition gives intuition but global limits introduce path dependence.
• Known resets are valuation state and can radically change remaining risk.
• Forward smile and model choice are major uncertainties, not calibration details.
• Hedge risk across reset dates and account separately for funding and issuer economics.