Cliquets And Ratchet Options

A ratchet wrench turns a bolt in one direction and clicks in place rather than sliding back when you reverse the handle — every partial turn is locked in permanently. A cliquet option applies that same click-and-lock mechanism to a stock's returns. Instead of one option that pays off on the total move from start to expiry, a cliquet is a series of consecutive forward start options, each locking in its own period's gain (floored at zero) before the next period starts fresh, at-the-money, regardless of what happened before.

The payoff

A cliquet with $n$ periods, each with a local floor of zero and a local cap $c$, has total payoff:

$$\text{Payoff} = N \sum_{i=1}^{n} \min\!\Big(\max\big(R_i,\ 0\big),\ c\Big),$$

where $R_i$ is the percentage return of the underlying over period $i$ only, computed relative to that period's starting price. In words: $N$ is the notional; each period's return $R_i$ is floored at zero (a losing period contributes nothing, but subtracts nothing either) and capped at $c$ (a huge winning period only counts up to the cap); the payoff sums those floored-and-capped local returns across all $n$ periods. A losing period never erases a previous period's locked-in gain — that's the ratchet action — but it also can't be offset against a later winning period, since each is capped and floored independently.

Worked example 1 — three periods, no cap

A 3-year cliquet resets annually, notional $10,000, floor 0%, no cap. Annual returns: Year 1 = +12%, Year 2 = −20%, Year 3 = +8%. Floored contributions: $\max(0.12,0)=0.12$, $\max(-0.20,0)=0$, $\max(0.08,0)=0.08$. Sum $= 0.12+0+0.08=0.20$. Payoff $= 10{,}000 \times 0.20 =$ $2,000. Compare to a plain call on the three-year cumulative return: the underlying fell 20% in year 2 and only partially recovered, so its three-year cumulative return is $1.12 \times 0.80 \times 1.08 - 1 \approx -3.2\%$, meaning an ordinary three-year call would pay nothing. The cliquet paid $2,000 by locking in the two winning years separately and simply ignoring the losing one.

Worked example 2 — the cost of a cap

Same structure, but now capped at $c = 8\%$ per year, same three annual returns (+12%, −20%, +8%). Contributions: $\min(\max(0.12,0),0.08) = 0.08$, $\min(\max(-0.20,0),0.08)=0$, $\min(\max(0.08,0),0.08)=0.08$. Sum $=0.08+0+0.08=0.16$. Payoff $=10{,}000\times0.16=$ $1,600 — $400 less than the uncapped version, entirely because the cap clipped Year 1's 12% return down to 8%. That clipped $400 is the premium the issuer saves by capping each period, which is exactly why capping local returns makes the cliquet cheaper to sell (and cheaper to buy).

Each period of a capped-and-floored cliquet behaves like the collar payoff shown above, applied fresh every period — drag the cap and floor and see how narrowing the collar (tighter cap, higher floor) trades away tail upside for a cheaper, steadier structure, period after period.

What this means in practice

Cliquets are popular in retail-facing structured products (often called "ratchet notes" or "auto-ratchets") precisely because "your gains lock in, your losses are floored" sells well — the marketing writes itself. Under the hood, each period is a forward start option, so pricing depends heavily on forward implied volatility and, critically, on the implied correlation between how volatility behaves across periods (the "volatility of volatility"), not just today's spot vol. This makes cliquets one of the more model-sensitive exotics — a small change in assumed forward vol dynamics can move the price a lot.