Skip to content

Scaling Laws for Greeks

This section provides a unified framework for understanding how Greeks scale with time-to-maturity \(\tau\) and moneyness.


Near-the-money ATM scaling table

For at-the-money options (\(S \approx K\)) in Black–Scholes:

Greek ATM Formula Scaling in \(\tau\) Behavior as \(\tau \to 0\)
\(\Delta\) \(N(d_1) \approx \frac{1}{2}\) \(\mathcal{O}(1)\) \(\to \frac{1}{2}\)
\(\Gamma\) \(\frac{1}{S\sigma\sqrt{2\pi\tau}}\) \(\tau^{-1/2}\) \(\to \infty\)
\(\Theta\) \(-\frac{S\sigma}{2\sqrt{2\pi\tau}}\) \(\tau^{-1/2}\) \(\to -\infty\)
\(\nu\) (Vega) \(\frac{S\sqrt{\tau}}{\sqrt{2\pi}}\) \(\tau^{1/2}\) \(\to 0\)
\(\rho\) \(\frac{K\tau}{2}e^{-r\tau}\) \(\tau\) \(\to 0\)

Theta-Gamma relationship

A fundamental identity links theta and gamma for delta-hedged positions:

\[ \boxed{\Theta + \frac{1}{2}\sigma^2 S^2 \Gamma = rV - rS\Delta} \]

For ATM options where \(\Delta \approx \frac{1}{2}\) and \(V \approx \frac{S\sigma\sqrt{\tau}}{\sqrt{2\pi}}\):

\[ \Theta \approx -\frac{1}{2}\sigma^2 S^2 \Gamma + \mathcal{O}(\sqrt{\tau}) \]

This shows that time decay and gamma are two sides of the same coin: options with high gamma (convexity benefit) must pay for it through time decay.


Dimensional analysis

Greeks can be understood through dimensional analysis. Let \([X]\) denote the dimension of \(X\):

Quantity Dimension
\(V\) (price) \(\$\)
\(S\) (spot) \(\$\)
\(\tau\) (time) \(T\)
\(\sigma\) (volatility) \(T^{-1/2}\)

Then: - \(\Delta = \partial V/\partial S\) is dimensionless - \(\Gamma = \partial^2 V/\partial S^2\) has dimension \(\$^{-1}\) - \(\Theta = \partial V/\partial t\) has dimension \(\$/T\) - \(\nu = \partial V/\partial \sigma\) has dimension \(\$ \cdot T^{1/2}\)

The ATM scalings follow from \(V_{\text{ATM}} \sim S\sigma\sqrt{\tau}\):

\[ \Gamma_{\text{ATM}} \sim \frac{V}{S^2} \cdot \frac{1}{\sigma\sqrt{\tau}} \sim \frac{1}{S\sigma\sqrt{\tau}} \]

Far from the money: exponential decay

For OTM options with log-moneyness \(x = \ln(K/S)\):

Greek OTM decay rate
\(\Delta\) \(\exp(-x^2/(2\sigma^2\tau))\)
\(\Gamma\) \(\frac{1}{S}\exp(-x^2/(2\sigma^2\tau))\)
\(\nu\) \(S\sqrt{\tau}\exp(-x^2/(2\sigma^2\tau))\)

The characteristic scale separating ATM from OTM is \(|x| \sim \sigma\sqrt{\tau}\).


Unified asymptotic expansion

For options near ATM as \(\tau \to 0\), all Greeks can be expanded in powers of \(\sqrt{\tau}\):

Price: [ V = V_0 + V_1 \sqrt{\tau} + V_2 \tau + \mathcal{O}(\tau^{3/2}) ]

where \(V_0 = (S-K)^+\) (intrinsic), \(V_1 = \frac{S\sigma}{\sqrt{2\pi}}\mathbf{1}_{\text{ATM}}\) (ATM time value).

Greeks: [ \Gamma = \frac{1}{S\sigma\sqrt{2\pi\tau}} + \mathcal{O}(1) ] [ \Theta = -\frac{S\sigma}{2\sqrt{2\pi\tau}} + \mathcal{O}(1) ] [ \nu = \frac{S\sqrt{\tau}}{\sqrt{2\pi}} + \mathcal{O}(\tau) ]


Medium maturity: vega dominance

For moderate maturities (\(\tau \sim 1\) year), vega often dominates portfolio risk:

  • Vega is maximized around \(\tau \sim 1/(2r)\) for typical interest rates
  • Gamma remains bounded: \(\Gamma \sim \frac{1}{S\sigma}\)
  • Theta-vega ratio: \(\Theta/\nu \sim -\sigma/(2\tau)\)

Long maturity asymptotics

For \(\tau \to \infty\):

Call price: \(C \to S\) (assuming no dividends)

Greeks: - \(\Delta \to 1\) - \(\Gamma \to 0\) as \(\tau^{-1/2}\) - \(\nu \to S\sqrt{\tau}N'(d_1) \sim S\sqrt{\tau}/\sqrt{2\pi}\) grows without bound - \(\rho \to K\tau e^{-r\tau}N(d_2)\) grows then decays


Practical implications

  1. Short-dated options: Large gamma, unstable delta hedging, high theta decay
  2. Medium-dated options: Vega dominance, sensitivity to implied vol changes
  3. Long-dated options: Low gamma, stable delta, model risk concerns

Greeks scaling summary

\[ \boxed{ \begin{aligned} &\text{ATM, short maturity:} & \Gamma &\sim \tau^{-1/2}, & \nu &\sim \tau^{1/2}, & \Theta &\sim -\tau^{-1/2} \\ &\text{OTM, any maturity:} & \text{All} &\sim \exp(-x^2/(2\sigma^2\tau)) \end{aligned} } \]

What to remember

  • Short-dated ATM options: large gamma (\(\tau^{-1/2}\)), unstable higher Greeks
  • Vega scales as \(\sqrt{\tau}\), peaking at moderate maturities
  • Theta and gamma are linked: \(\Theta \approx -\frac{1}{2}\sigma^2 S^2 \Gamma\) for ATM
  • Extreme strikes: tail-dominated sensitivities with exponential decay
  • Dimensional analysis provides quick scaling estimates

Exercises

Exercise 1. Using the dimensional analysis framework, verify that \(\Gamma_{\text{ATM}} \sim 1/(S\sigma\sqrt{\tau})\) has dimension \(\$^{-1}\) and that \(\Theta_{\text{ATM}} \sim -S\sigma/(2\sqrt{\tau})\) has dimension \(\$/T\). Check that the theta-gamma identity \(\Theta + \frac{1}{2}\sigma^2 S^2 \Gamma = r(V - S\Delta)\) is dimensionally consistent.

Solution to Exercise 1

Checking \(\Gamma_{\text{ATM}}\): The dimension of \(\Gamma = \partial^2 V / \partial S^2\) is \([\$] / [\$]^2 = \$^{-1}\).

For \(\Gamma_{\text{ATM}} \sim 1/(S\sigma\sqrt{\tau})\): the dimension of the denominator is \([\$] \cdot [T^{-1/2}] \cdot [T^{1/2}] = [\$]\), so \(\Gamma_{\text{ATM}}\) has dimension \(\$^{-1}\). Confirmed.

Checking \(\Theta_{\text{ATM}}\): The dimension of \(\Theta = \partial V / \partial t\) is \([\$]/[T]\).

For \(\Theta_{\text{ATM}} \sim -S\sigma/(2\sqrt{\tau})\): the dimension is \([\$] \cdot [T^{-1/2}] / [T^{1/2}] = [\$] / [T]\). Confirmed.

Checking the theta-gamma identity \(\Theta + \frac{1}{2}\sigma^2 S^2 \Gamma = r(V - S\Delta)\):

  • Left side, first term: \([\Theta] = \$/T\)
  • Left side, second term: \([\sigma^2 S^2 \Gamma] = [T^{-1}][\$^2][\$^{-1}] = \$/T\). Consistent.
  • Right side: \([r(V - S\Delta)] = [T^{-1}][\$] = \$/T\). Consistent.

All three terms have dimension \(\$/T\), so the identity is dimensionally consistent.


Exercise 2. For an OTM call with log-moneyness \(x = \ln(K/S) = 0.10\) (roughly 10% OTM), \(\sigma = 0.20\), and \(\tau = 0.25\) years, compute the exponential decay factor \(\exp(-x^2/(2\sigma^2\tau))\). How does this compare to the ATM vega at the same maturity?

Solution to Exercise 2

With \(x = 0.10\), \(\sigma = 0.20\), and \(\tau = 0.25\):

\[ \sigma^2\tau = 0.04 \times 0.25 = 0.01 \]
\[ \frac{x^2}{2\sigma^2\tau} = \frac{0.01}{0.02} = 0.5 \]
\[ \exp\!\left(-\frac{x^2}{2\sigma^2\tau}\right) = e^{-0.5} \approx 0.6065 \]

The ATM vega at the same maturity is \(\nu_{\text{ATM}} \approx S\sqrt{\tau}/\sqrt{2\pi} = 100 \times 0.5 / 2.507 \approx 19.95\).

The OTM vega is approximately \(\nu_{\text{OTM}} \approx 19.95 \times 0.6065 \approx 12.10\).

So the 10% OTM option retains about \(61\%\) of the ATM vega. This is because \(|x|/(\sigma\sqrt{\tau}) = 0.10/0.10 = 1\), meaning the option is only one characteristic width away from ATM. At this moderate distance, vega is reduced but far from negligible.


Exercise 3. The theta-gamma relationship states that for ATM options, \(\Theta \approx -\frac{1}{2}\sigma^2 S^2 \Gamma\). Verify this numerically for \(S = K = 100\), \(\sigma = 0.20\), \(r = 0.05\), \(\tau = 0.5\) by computing both sides using the closed-form Black--Scholes formulas.

Solution to Exercise 3

Using the Black-Scholes formulas with \(S = K = 100\), \(\sigma = 0.20\), \(r = 0.05\), \(\tau = 0.5\):

\[ d_1 = \frac{\ln(1) + (0.05 + 0.02) \times 0.5}{0.20\sqrt{0.5}} = \frac{0.035}{0.1414} = 0.2475 \]
\[ d_2 = d_1 - 0.20\sqrt{0.5} = 0.2475 - 0.1414 = 0.1061 \]
\[ N'(d_1) = \frac{1}{\sqrt{2\pi}}e^{-d_1^2/2} = \frac{1}{2.5066}e^{-0.0306} = 0.3988 \times 0.9698 = 0.3868 \]

Computing \(\Gamma\):

\[ \Gamma = \frac{N'(d_1)}{S\sigma\sqrt{\tau}} = \frac{0.3868}{100 \times 0.20 \times 0.7071} = \frac{0.3868}{14.14} = 0.02736 \]

Computing \(\Theta\): The exact Black-Scholes theta for a call is

\[ \Theta = -\frac{S\sigma N'(d_1)}{2\sqrt{\tau}} - rKe^{-r\tau}N(d_2) \]
\[ = -\frac{100 \times 0.20 \times 0.3868}{2 \times 0.7071} - 0.05 \times 100 \times e^{-0.025} \times N(0.1061) \]
\[ = -\frac{7.736}{1.4142} - 5 \times 0.9753 \times 0.5422 = -5.472 - 2.644 = -8.116 \]

Left side of identity: \(-\frac{1}{2}\sigma^2 S^2 \Gamma = -\frac{1}{2}(0.04)(10000)(0.02736) = -5.472\).

Right side: \(r(V - S\Delta)\). We need \(V\) and \(\Delta\):

\[ \Delta = N(d_1) = N(0.2475) \approx 0.5977 \]
\[ V = SN(d_1) - Ke^{-r\tau}N(d_2) = 100(0.5977) - 100(0.9753)(0.5422) = 59.77 - 52.89 = 6.88 \]
\[ r(V - S\Delta) = 0.05(6.88 - 59.77) = 0.05(-52.89) = -2.645 \]

Checking: \(\Theta + \frac{1}{2}\sigma^2 S^2\Gamma = -8.116 + 5.472 = -2.644 \approx -2.645 = r(V - S\Delta)\).

The identity is verified numerically, confirming \(\Theta \approx -\frac{1}{2}\sigma^2 S^2\Gamma\) up to the correction term \(r(V - S\Delta)\).


Exercise 4. A portfolio manager holds ATM options at three maturities: \(\tau = 0.08\) (1 month), \(\tau = 0.5\) (6 months), and \(\tau = 2\) (2 years). Using the scaling laws, rank these positions by: (a) gamma exposure, (b) vega exposure, (c) daily theta bleed. Which maturity dominates each risk?

Solution to Exercise 4

Using the ATM scaling formulas with \(S = K = 100\), \(\sigma = 0.20\):

(a) Gamma exposure (\(\Gamma \sim 1/(S\sigma\sqrt{2\pi\tau})\)):

  • \(\tau = 0.08\): \(\Gamma \approx 1/(100 \times 0.20 \times \sqrt{2\pi \times 0.08}) = 1/(20 \times 0.7090) = 0.0706\)
  • \(\tau = 0.5\): \(\Gamma \approx 1/(20 \times 1.7725) = 0.0283\)
  • \(\tau = 2\): \(\Gamma \approx 1/(20 \times 3.5449) = 0.0141\)

Ranking (highest to lowest gamma): 1-month \(>\) 6-month \(>\) 2-year.

(b) Vega exposure (\(\nu \sim S\sqrt{\tau}/\sqrt{2\pi}\)):

  • \(\tau = 0.08\): \(\nu \approx 100 \times 0.2828 / 2.5066 = 11.28\)
  • \(\tau = 0.5\): \(\nu \approx 100 \times 0.7071 / 2.5066 = 28.21\)
  • \(\tau = 2\): \(\nu \approx 100 \times 1.4142 / 2.5066 = 56.42\)

Ranking (highest to lowest vega): 2-year \(>\) 6-month \(>\) 1-month.

(c) Daily theta bleed (\(\Theta \sim -S\sigma/(2\sqrt{2\pi\tau})\), converted to daily):

  • \(\tau = 0.08\): \(\Theta/365 \approx -(100 \times 0.20)/(2 \times 0.7090 \times 365) = -20/517.6 = -0.0387\) per day
  • \(\tau = 0.5\): \(\Theta/365 \approx -20/(2 \times 1.7725 \times 365) = -20/1293.9 = -0.0155\) per day
  • \(\tau = 2\): \(\Theta/365 \approx -20/(2 \times 3.5449 \times 365) = -20/2587.8 = -0.00773\) per day

Ranking (highest daily theta bleed): 1-month \(>\) 6-month \(>\) 2-year.

Summary: Short-dated options dominate gamma and theta risk, while long-dated options dominate vega risk. This reflects the fundamental scaling: gamma and theta scale as \(\tau^{-1/2}\) (growing as maturity shrinks) while vega scales as \(\tau^{1/2}\) (growing with maturity).


Exercise 5. The characteristic scale separating ATM from OTM is \(|x| \sim \sigma\sqrt{\tau}\). For \(\sigma = 0.30\) and \(\tau = 1\) month, compute this scale in both log-moneyness units and as a percentage of spot price. An option with \(K = 105\) when \(S = 100\) --- is it effectively ATM or OTM at this maturity?

Solution to Exercise 5

With \(\sigma = 0.30\) and \(\tau = 1/12\) (1 month):

\[ \sigma\sqrt{\tau} = 0.30 \times \sqrt{1/12} = 0.30 \times 0.2887 = 0.08660 \]

This is the scale in log-moneyness units (dimensionless).

As a percentage of spot price: since \(|x| = |\ln(S/K)| \approx |S - K|/S\) for small deviations, the characteristic scale is about \(8.66\%\) of spot.

For \(K = 105\) and \(S = 100\):

\[ |x| = |\ln(105/100)| = \ln(1.05) \approx 0.04879 \]

The ratio of this log-moneyness to the characteristic scale is:

\[ \frac{|x|}{\sigma\sqrt{\tau}} = \frac{0.04879}{0.08660} \approx 0.564 \]

Since this ratio is less than 1, the \(K = 105\) option is well within one characteristic width of ATM. At this maturity, the option is effectively ATM --- its Greeks (delta, gamma, vega) will be only modestly reduced from their peak ATM values. The exponential decay factor is \(\exp(-0.564^2/2) = e^{-0.159} \approx 0.853\), so Greeks are at about \(85\%\) of their ATM levels.


Exercise 6. In the long-maturity regime (\(\tau \to \infty\)), vega grows as \(S\sqrt{\tau}/\sqrt{2\pi}\) while gamma decays as \(1/(S\sigma\sqrt{2\pi\tau})\). Show that the ratio \(\nu/\Gamma = \sigma S^2 \tau\) grows linearly with \(\tau\). What does this imply about the relative importance of vega versus gamma hedging for long-dated options?

Solution to Exercise 6

Using the ATM scaling formulas:

\[ \nu_{\text{ATM}} \approx \frac{S\sqrt{\tau}}{\sqrt{2\pi}}, \qquad \Gamma_{\text{ATM}} \approx \frac{1}{S\sigma\sqrt{2\pi\tau}} \]

Taking the ratio:

\[ \frac{\nu}{\Gamma} = \frac{S\sqrt{\tau}/\sqrt{2\pi}}{1/(S\sigma\sqrt{2\pi\tau})} = \frac{S\sqrt{\tau}}{\sqrt{2\pi}} \times S\sigma\sqrt{2\pi\tau} = \sigma S^2 \tau \]

This grows linearly with \(\tau\), confirming the claim.

Implications for hedging long-dated options: As \(\tau\) increases, vega grows much faster relative to gamma. For a portfolio of long-dated options:

  • A \(1\%\) change in implied volatility (\(\Delta\sigma = 0.01\)) causes P&L of \(\nu \times 0.01 \sim S\sqrt{\tau}/\sqrt{2\pi} \times 0.01\), which grows with \(\sqrt{\tau}\).
  • A \(1\%\) move in spot (\(\Delta S = 0.01 S\)) causes gamma P&L of \(\frac{1}{2}\Gamma(\Delta S)^2 \sim S/(2\sigma\sqrt{2\pi\tau}) \times (0.01)^2\), which decays with \(\sqrt{\tau}\).

For long-dated options (\(\tau \gg 1\)), vega dominates portfolio risk by a factor proportional to \(\tau\). Therefore, vega hedging is far more important than gamma hedging for long-dated positions. This is why traders managing long-dated books focus primarily on volatility surface risk rather than delta-gamma risk.