Call and Put Delta

Delta as Sensitivity to the Stock Price

For a European option, delta measures how its current value changes when the stock price changes while time, strike, interest rate, and volatility are held fixed. In the Black–Scholes model, this derivative has a direct hedging interpretation: it is the number of shares in the replicating portfolio. Establishing the formula carefully matters, especially for puts, where differentiating the negative spot term is a common source of sign errors.

Work on a filtered probability space \((\Omega,\mathcal{F},(\mathcal{F}_t)_{t\ge0},\mathbb{P})\) carrying a Brownian motion \(B_t\). Assume a non-dividend-paying stock, constant interest rate \(r\), and constant volatility \(\sigma>0\). Under the physical measure \(\mathbb{P}\), the stock may have its physical drift; for risk-neutral pricing, use an equivalent risk-neutral measure \(\mathbb{Q}\), under which the stock satisfies \(dS_t=rS_t\,dt+\sigma S_t\,dB_t^{\mathbb{Q}}\). Here \(B_t^{\mathbb{Q}}\) is Brownian motion under \(\mathbb{Q}\). The stochastic integral in this equation is an Ito integral with progressively measurable integrand \(\sigma S_t\); on every finite time interval it is square-integrable, since \(\mathbb{E}^{\mathbb{Q}}[S_t^2]\) is finite and hence \(\mathbb{E}^{\mathbb{Q}}\int_0^T\sigma^2S_t^2\,dt<\infty\).

Let \(t<T\), \(\tau=T-t>0\), current spot \(S=S_t>0\), strike \(K>0\), and define

Write \(\Phi\) for the standard normal cumulative distribution function and \(\phi(x)=\frac{1}{\sqrt{2\pi}}e^{-x^2/2}\) for its density. The Black–Scholes call and put prices, with no continuous dividend yield, are

Differentiate the Two Prices

The key cancellation in both derivatives follows from a relation between the two normal densities. Differentiating \(d_1\) and \(d_2\) with respect to spot gives \(\frac{\partial d_1}{\partial S}=\frac{\partial d_2}{\partial S}=\frac{1}{S\sigma\sqrt{\tau}}\). Also, since \(d_1-d_2=\sigma\sqrt{\tau}\) and \(d_1+d_2=2(\ln(S/K)+r\tau)/(\sigma\sqrt{\tau})\),

Proof. For the call, use \(\frac{d}{dx}\Phi(x)=\phi(x)\), the spot derivatives above, and the product rule on \(S\Phi(d_1)\). This gives

where the last equality uses \(S\phi(d_1)=Ke^{-r\tau}\phi(d_2)\). For the put, differentiating \(\Phi(-d_i)\) gives \(-\phi(d_i)\frac{\partial d_i}{\partial S}\). In particular, differentiating its negative spot term contributes both a negative CDF term and a positive density term:

Again, the density terms cancel by the same identity. The final equality uses \(\Phi(x)+\Phi(-x)=1\). This proves both formulas, including the put’s negative sign. \(\square\)

Why Delta Is the Replicating Stock Holding

Delta is not merely a derivative used for reporting price sensitivity. It identifies the stock position needed to match the random part of the option’s value change. To see this, let \(V(t,S)\) be a smooth Black–Scholes price. Ito’s formula for the function \(f(t,s)=V(t,s)\), with \(f_t=V_t\), \(f_s=V_S\), and \(f_{ss}=V_{SS}\), gives

The Black–Scholes PDE is \(V_t+rSV_S+\tfrac12\sigma^2S^2V_{SS}=rV\). Thus the option’s dynamics are \(dV=rV\,dt+\sigma S V_S\,dB_t^{\mathbb{Q}}\). A self-financing portfolio holding \(\Delta_t\) shares and \(V-\Delta_t S_t\) in the bank account has dynamics \(dV=\Delta_t\,dS_t+r(V-\Delta_t S_t)\,dt\). Substituting the stock dynamics gives its diffusion term \(\Delta_t\sigma S_t\,dB_t^{\mathbb{Q}}\). Matching this with the option’s diffusion term requires \(\Delta_t=V_S(t,S_t)\), since \(\sigma S_t>0\). With this choice the drift is \(rV\,dt\), as required. For a call or put, the replicating stock holding is therefore its delta.

These dynamics and the pricing calculation are stated under \(\mathbb{Q}\), the risk-neutral measure, not under the physical measure \(\mathbb{P}\). The stock’s drift under \(\mathbb{P}\) need not equal \(r\). The delta formulas depend on \(S,K,r,\sigma,\tau\), not on that physical drift. On any interval ending before maturity, the Black–Scholes deltas are bounded in absolute value by \(1\); the associated stock integrand is square-integrable over finite horizons because \(\mathbb{E}^{\mathbb{Q}}\int_0^T \sigma^2S_t^2\,dt<\infty\). At maturity, the payoff has a kink at \(S=K\), so a classical derivative there need not exist; the formulas describe the hedge before maturity.

Turning Delta into a Hedge Position

A delta is a number of shares per option when the position is one option. If a contract represents \(m\) options and there are \(n\) contracts, the signed stock holding that offsets the position’s local stock-price exposure is \(-nm\Delta\). A positive resulting holding means buy shares; a negative one means sell shares. For a short option position, \(n\) is negative, so the sign reverses. This is a local hedge: delta changes as spot and time change, so a hedge generally must be rebalanced.

Common Sign and Interpretation Pitfalls

Pitfall: differentiating the put’s spot term with the wrong sign. A mistaken calculation may write \(\frac{\partial}{\partial S}[-S\Phi(-d_1)]=-\Phi(-d_1)-S\phi(d_1)\frac{\partial d_1}{\partial S}\). The second term is wrong: the inner derivative of \(\Phi(-d_1)\) is \(-\phi(d_1)\frac{\partial d_1}{\partial S}\), and the minus sign in front of \(S\) changes that contribution to positive. The correct derivative is \(-\Phi(-d_1)+S\phi(d_1)\frac{\partial d_1}{\partial S}\). Keeping this sign gives \(\Delta_P=-\Phi(-d_1)\), not a positive put delta.

Delta is not automatically an exercise probability. The call delta is \(\Phi(d_1)\), whereas the risk-neutral probability that \(S_T>K\) in the no-dividend Black–Scholes model is \(\Phi(d_2)\). These use different arguments because \(d_1=d_2+\sigma\sqrt{\tau}\). Delta is a price derivative and a hedge ratio; it should not be replaced by a probability. Also, a delta hedge offsets first-order stock sensitivity at the current state. It does not remove all risk over a finite interval, because the sensitivity changes as market conditions change.