agentleFS
brainbytes-devbrainbytes-dev/everything-claude-trading/skills/derivatives/exotic-options/SKILL.md

exotic-options

Skill16 starsChanged 8 months agoMITView on GitHub

Barriers, Asians, lookbacks, digitals, cliquets, autocallables, and quanto options. Pricing, hedging, and risk management. Classification - Knock-out: option ceases to exist if barrier is hit (down-and-out, up-and-out) - Knock-in: option comes into existence only if barrier is hit (down-and-in, up-and-in) - In-out parity: knock-in + knock-out = vanilla (same strike and expiry) - Rebate: fixed payment made when the barrier is hit (for knock-outs) or at expiry if barrier is never hit (for knock-ins) Pricing - Continuous monitoring: closed-form solutions…

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If underlying > autocall barrier (e.g., 100% of initial) on observation date: note is called, investor receives principal + coupon

    line 110
  2. Distinctive:

    If not called and underlying > coupon barrier: investor receives coupon for that period

    line 111
  3. Distinctive:

    If equity and FX are negatively correlated (typical for developed markets: currency weakens when stocks fall), the quanto drift adjustment …

    line 134
  4. Distinctive:

    All exotic prices must be bounded by appropriate upper and lower limits (vanilla, intrinsic value)

    line 195

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Exotic Options

Barriers, Asians, lookbacks, digitals, cliquets, autocallables, and quanto options. Pricing, hedging, and risk management.

When to Activate

  • User pricing or hedging barrier options (knock-in, knock-out)
  • Pricing Asian options (arithmetic or geometric average)
  • Analyzing lookback, digital/binary, or cliquet options
  • Understanding autocallable structures and their risks
  • Quanto options and cross-currency derivative pricing
  • Hedging exotic payoffs and understanding Greeks behavior

Core Concepts

Barrier Options

Classification

  • Knock-out: option ceases to exist if barrier is hit (down-and-out, up-and-out)
  • Knock-in: option comes into existence only if barrier is hit (down-and-in, up-and-in)
  • In-out parity: knock-in + knock-out = vanilla (same strike and expiry)
  • Rebate: fixed payment made when the barrier is hit (for knock-outs) or at expiry if barrier is never hit (for knock-ins)

Pricing

  • Continuous monitoring: closed-form solutions exist under BSM (reflection principle)
  • Discrete monitoring (daily, weekly): no closed-form; use MC with Brownian bridge correction or lattice methods
  • Broadie-Glasserman-Kou continuity correction: shift barrier by beta * sigma * sqrt(dt) where beta = 0.5826
  • Near-barrier behavior: Greeks become extreme (delta can flip sign, gamma spikes)

Key Risks

  • Pin risk near the barrier: small moves determine whether the option exists or not
  • Barrier shift risk: if the barrier is monitored at fixing times, the effective barrier differs from contractual
  • Hedging difficulty: near the barrier, delta changes rapidly and gamma is very large
  • Gap risk: underlying can gap through the barrier (especially over weekends, overnight)

Asian Options

Types

  • Fixed-strike Asian call: max(avg(S) - K, 0) — compare average price to fixed strike
  • Floating-strike Asian call: max(S_T - avg(S), 0) — compare terminal price to average
  • Average can be arithmetic (standard) or geometric (tractable)
  • Averaging period: full life or partial (e.g., last 3 months of a 1-year option)

Pricing

  • Geometric average Asian: closed-form under GBM (geometric average of log-normals is log-normal)
    • Geometric Asian call: BSM formula with adjusted vol (sigma/sqrt(3)) and drift
  • Arithmetic average Asian: no closed-form; use MC with geometric Asian as control variate
  • Turnbull-Wakeman approximation: match first two moments of arithmetic average to log-normal
  • As averaging dates increase, Asian option value decreases (averaging reduces effective volatility)

Properties

  • Cheaper than vanilla options (averaging reduces payoff volatility)
  • Popular for commodity hedging (averaging reflects actual purchase/sale prices)
  • Less sensitive to manipulation near expiry (one closing price cannot dominate)
  • Vega is lower than vanilla; delta behavior depends on how much averaging has occurred

Lookback Options

Types

  • Fixed-strike lookback call: max(S_max - K, 0) — payoff based on maximum price observed
  • Floating-strike lookback call: S_T - S_min — buy at the minimum, sell at current
  • Floating-strike lookback put: S_max - S_T — sell at the maximum, buy at current
  • Partial lookback: lookback feature applies to only part of the option's life

Pricing

  • Closed-form under continuous monitoring (Goldman, Sosin, Gatto 1979)
  • Floating-strike lookback call: S_T * N(a1) - S_min * exp(-rT) * N(a2) - S_T * sigma^2/(2r) * [...]
  • Very expensive: the ability to buy at the low / sell at the high has significant value
  • Typically 2-3x the price of an ATM vanilla option
  • Discrete monitoring: use MC with Brownian bridge correction for continuous max/min estimation

Digital (Binary) Options

Types

  • Cash-or-nothing call: pays fixed amount Q if S_T > K, else 0
  • Asset-or-nothing call: pays S_T if S_T > K, else 0
  • Standard call = asset-or-nothing call - K * cash-or-nothing call (decomposition)

Pricing under BSM

  • Cash-or-nothing call: Q * exp(-rT) * N(d2)
  • Asset-or-nothing call: S * exp(-qT) * N(d1)
  • Delta of digital: Q * exp(-rT) * phi(d2) / (S * sigma * sqrt(T)) — can be very large near ATM at expiry

Hedging Challenges

  • Digital options have discontinuous payoff — delta approaches infinity near ATM at expiry
  • In practice: replicate with tight call spread (buy K call, sell K+epsilon call, scale by Q/epsilon)
  • The call spread replication has bounded delta and gamma
  • Skew sensitivity: digital price is very sensitive to the vol skew (slope of IV around the strike)
  • Overhedge: use call spread width that accounts for realistic hedging frequency

Cliquet (Ratchet) Options

Structure

  • Series of forward-starting options, each resetting at the end of the previous period
  • Each period: captures return = max(0, S_{t+1}/S_t - 1), typically with local cap and floor
  • Total payoff: sum of capped/floored periodic returns
  • Embedded in equity-linked insurance products and structured notes

Pricing Considerations

  • Cliquet value depends critically on forward volatility and forward skew
  • Cannot be hedged with vanilla options alone — requires forward-starting option hedges
  • Very sensitive to vol-of-vol and correlation structure
  • Stochastic volatility models (Heston, SABR) are essential for accurate pricing
  • Local vol models can misprice cliquets because they underestimate forward smile dynamics

Autocallable Notes

Structure

  • Periodic observation dates (e.g., quarterly, semi-annually)
  • If underlying > autocall barrier (e.g., 100% of initial) on observation date: note is called, investor receives principal + coupon
  • If not called and underlying > coupon barrier: investor receives coupon for that period
  • At maturity, if underlying < put strike (e.g., 60% of initial): investor bears downside loss
  • Most popular structured product globally (especially in Asia and Europe)

Risk Profile

  • Short a deep OTM put (downside exposure)
  • Short a series of digital options at the autocall barrier
  • Correlation with dividends and rates (forward price matters for autocall probability)
  • Path-dependent: early autocall eliminates remaining coupons and downside risk
  • Issuer is typically net long vol and long correlation for worst-of autocallables

Quanto Options

Definition

  • Option on a foreign asset with payoff converted at a fixed FX rate (not the prevailing rate)
  • Example: USD investor buys a call on Nikkei 225 with payoff in USD at a fixed JPY/USD rate
  • Eliminates FX risk from the investor's perspective

Pricing Adjustment

  • Under BSM, quanto adjustment modifies the drift of the foreign asset:
    • r_quanto = r_foreign - rho * sigma_S * sigma_FX
    • rho = correlation between foreign asset and FX rate
    • sigma_FX = volatility of the FX rate
  • If equity and FX are negatively correlated (typical for developed markets: currency weakens when stocks fall), the quanto drift adjustment is positive — quanto option is worth more
  • Volatility of the quanto option = sigma_S (no FX vol component)

Examples

Down-and-Out Call Pricing

S=100, K=100, Barrier B=85, r=5%, sigma=25%, T=0.5

Vanilla call (BSM): $7.62
Down-and-out call (closed-form): $5.84
Knock-out discount: $1.78 (23% cheaper due to barrier risk)

If S drops to 86 (1 point above barrier):
  Vanilla call value: $2.10
  D&O call value: $0.15 (nearly worthless — about to knock out)
  Delta of D&O: -2.5 (negative — further drop kills the option)
  Gamma: extremely large (discontinuity at barrier)

Arithmetic Asian Option

S=100, K=100, r=5%, sigma=30%, T=1.0, monthly averaging (12 dates)

Vanilla call: $14.23
Geometric Asian call (closed-form): $8.42
Arithmetic Asian call (MC with control variate): $8.91

Asian discount: 37% cheaper than vanilla.
Effective vol of average: approximately 30% / sqrt(3) = 17.3% (for geometric).

Worst-of Autocallable

Underlying: worst of SPX, EuroStoxx 50, Nikkei 225
Autocall barrier: 100%, observed quarterly
Coupon: 8% p.a. if worst-of > 70%
Put strike: 60% of initial (investor bears loss below 60%)
Maturity: 3 years

Pricing requires:
  - 3-asset correlated MC simulation
  - Calibrated local-stochastic vol model for each underlying
  - Quanto adjustment for cross-currency underlyings
  - Dividend assumptions for forward price computation

Key sensitivities:
  - Correlation: lower correlation = lower autocall probability = higher coupon value but more downside risk
  - Vol: higher vol = wider range of outcomes = more expensive embedded put
  - Dividends: higher dividends = lower forward = less likely to autocall

Quality Gate

  • Barrier options: verify with in-out parity (knock-in + knock-out = vanilla within tolerance)
  • Barrier monitoring: specify continuous vs. discrete and apply appropriate corrections
  • Asian options: arithmetic Asian price must be bounded by geometric Asian (lower) and vanilla (upper)
  • Digital options: replicate with call/put spread in practice; never hedge a pure digital
  • Cliquet pricing must use a stochastic vol model — local vol alone systematically misprices forward skew
  • Autocallables: Monte Carlo must have sufficient paths (>100,000) and correct early termination logic
  • Quanto: verify the correlation sign and magnitude between asset and FX are reasonable
  • All exotic prices must be bounded by appropriate upper and lower limits (vanilla, intrinsic value)
  • Greeks must be computed via bump-and-revalue with appropriately small bumps (especially near barriers)
  • Model risk assessment: compare prices from at least two models (BSM, local vol, stochastic vol)
  • Hedging strategy must be defined before trading: specify instruments, rebalancing frequency, Greeks targets

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.