> ## Documentation Index
> Fetch the complete documentation index at: https://raymondchang.me/llms.txt
> Use this file to discover all available pages before exploring further.

# Greeks and hedging error

> Gamma-theta P&L, discrete-hedging error, and transaction-cost-aware hedging bands.

> [!NOTE]
>
> **Status of this page:** P0 draft. Worked numbers were checked by calculation; model results (Leland, Whalley-Wilmott) are asymptotic and should be tested on the desk's own data.

## 1. P\&L of a delta-hedged option

Over a short interval, the delta-hedged P\&L is approximately

`P&L = 0.5 x Gamma_$ x (r^2 - sigma_imp^2 x dt) + vega x d(sigma_imp) + theta residual`

where Gamma\_\$ = Gamma x S^2 is dollar gamma, r is the realised return in the interval, and dt is the interval in years. The first term is the realised-versus-implied variance bet.

**Worked example (verified).** Gamma\_\$ = 5,000,000, implied vol 25%, daily interval.

* Breakeven daily move = 25% / sqrt(252) = 1.575%.
* A 2% move: 0.5 x 5,000,000 x (0.02^2 - 0.01575^2) = about +380.
* A 1% move: about -370.

Hedge frequency changes the realised path of this P\&L but not its expectation under the model.

## 2. Discrete hedging error

If the true process is Black-Scholes and the hedge is rebalanced N times to expiry, the hedging error has standard deviation of order 1/sqrt(N) times a term proportional to vega-like exposure (Boyle and Emanuel, 1980). Practical drivers beyond N:

* Gamma near the strike close to expiry.
* Jumps and gaps, which a continuous model cannot hedge.
* Misspecified vol: the hedge delta uses sigma\_hedge, and the P\&L depends on realised minus hedge-vol, not implied.

## 3. Hedging with transaction costs

**Leland (1985)** proposes pricing with an adjusted vol: sigma\_adj^2 = sigma^2 x (1 + Le), where Le = sqrt(2/pi) x k / (sigma x sqrt(dt)) and k is the proportional cost per trade.

Verified examples with k = 0.1% and sigma = 25%: daily rebalancing gives Le = 0.051 and sigma\_adj of about 25.6%; hourly rebalancing (8 per day) gives Le = 0.143 and sigma\_adj of about 26.7%.

Caveat: Zhao and Ziemba (2003) argue that Leland's main convergence theorem is flawed, and that exact replication is not achieved in the limit. Treat the formula as a heuristic for cost-adjusted vol, not a replication proof.

**No-trade band (Whalley and Wilmott).** The optimal policy under proportional costs is to hedge only when delta leaves a band around the Black-Scholes delta. The half-width scales with lambda^(1/3), where lambda is the proportional cost, and also depends on gamma and risk aversion. Doubling the cost widens the band by about 26%, not 100%.

## 4. Desk-level implications

* Hedge on delta bands tied to gamma and spread, not only to a clock.
* Aggregate across the book: netting reduces cost more than per-option bands do.
* Hedge instrument choice (futures vs cash vs ETF) changes both cost and basis risk. See [Delta-1 hedging](https://raymondchang.me/equities-derivatives/global/delta-1-hedging.md).
* Separate the P\&L explain into gamma-theta, vega, hedge slippage, and carry.

## Points to verify

* Exact band formula and constants in Whalley and Wilmott (1997) before implementing.
* HK stamp duty and fee exemptions for market-maker hedges, which materially change k.

## Sources

* Leland (1985), Option pricing and replication with transactions costs, Journal of Finance.
* Zhao and Ziemba (2003), On Leland's Option Hedging Strategy with Transaction Costs.
* Toft (1996), On the Mean-Variance Tradeoff in Option Replication with Transactions Costs, JFQA 31(2).
* Whalley and Wilmott (1997), An asymptotic analysis of an optimal hedging model for option pricing with transaction costs, Mathematical Finance.
* Boyle and Emanuel (1980), Discretely adjusted option hedges, Journal of Financial Economics.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.