TQBTHEQUANTBATEMAN
TQB/ learn/ equity/ sabrEN · DARK
volatility · front-office

SABR

Forward smile dynamics through backbone, correlation and vol-of-vol

BY THE END, YOU CAN

01Interpret α, β, ρ and ν economically.

02Read the Hagan approximation in forward-moneyness coordinates.

03Diagnose normal versus lognormal conventions.

04Calibrate with stable parameter transformations.

01
INTUITION

Build the state before the equation.

SABR couples a forward with a stochastic volatility factor. β controls the backbone, ρ the skew channel and ν the smile curvature; the widely quoted formula is an asymptotic approximation, not the SDE itself.

01

β is commonly fixed because all four parameters are not equally identifiable.

02

Shifted lognormal or normal variants are required near negative rates.

03

The approximation loses accuracy in extreme wings and long horizons.

02
WHY MARKETS CARE

The product exists before the model.

SABR remains a central quoting and interpolation model for caps, floors and swaptions and is useful for studying forward-smile dynamics.

INSTRUMENTS

caps and floors

swaptions

commodity options

FX variants

QUOTE CONVENTION

State normal, lognormal or shifted-lognormal volatility, forward, shift, delta/strike coordinate and annuity conventions.

03
MATHEMATICS

Notation, units and exact claims.

F_t: forwardα_t: stochastic scaleβ: backbone elasticityν: vol of volρ: correlation
QUANT NOTESYNTHETIC · EDUCATIONAL

SABR dynamics

A stochastic scale drives a CEV-style forward diffusion.

dFt=αtFtβdWt1,dαt=ναtdWt2dF_t=\alpha_tF_t^\beta dW_t^1,\qquad d\alpha_t=\nu\alpha_tdW_t^2
Read the equation together with its financial domain and convention.
QUANT NOTESYNTHETIC · EDUCATIONAL

Correlation

Correlation contributes the leading skew effect.

dW1,W2t=ρdtd\langle W^1,W^2\rangle_t=\rho\,dt
Read the equation together with its financial domain and convention.
QUANT NOTESYNTHETIC · EDUCATIONAL

ATM leading order

ATM volatility mixes scale, backbone, correlation and vol-of-vol corrections.

σATMαF1β[1+T((1β)2α224F22β+ρβνα4F1β+(23ρ2)ν224)]\sigma_{ATM}\approx\frac{\alpha}{F^{1-\beta}}\left[1+T\left(\frac{(1-\beta)^2\alpha^2}{24F^{2-2\beta}}+\frac{\rho\beta\nu\alpha}{4F^{1-\beta}}+\frac{(2-3\rho^2)\nu^2}{24}\right)\right]
Read the equation together with its financial domain and convention.
04
DERIVATION

Do not jump to the final expression.

DERIVATION

Reading the SABR asymptotic structure

The complete expansion is lengthy; the useful derivation is to identify leading scale and first correction channels.

  1. 01

    Freeze stochastic scale

    At leading order, α/F^(1-β) supplies the local ATM volatility.

  2. 02

    Introduce log-moneyness

    The expansion uses z proportional to ν/α times a backbone-adjusted forward-strike distance.

  3. 03

    Correct the strike geometry

    The z/x(z) ratio introduces asymmetric smile response through ρ.

  4. 04

    Add maturity corrections

    Backbone curvature, correlation and vol-of-vol contribute O(T) terms.

  5. 05

    Control the limit at ATM

    Use the analytical z→0 limit rather than direct division to avoid cancellation.

SABR is powerful because parameters map to smile features, but robust use requires convention control, asymptotic guards and calibration governance.

05
MODEL / PRICING

Fit, compute, then challenge the assumptions.

METHOD

Fix or regularize β, transform constrained parameters, price under the correct volatility convention and use stable ATM limits.

CALIBRATION

Fit α, ρ and ν to liquid smile quotes with bounded transforms and maturity-to-maturity regularization.

LIMITATIONS

Asymptotic error in wings and long maturities.

Parameters can be non-unique.

Lognormal form fails at non-positive forwards without a shift.

MODEL COMPARISON

Static fit is not dynamics.

QuestionBlack–ScholesLocal volatilityHeston
Volatility stateOne constant σσ(S,t) deterministicvₜ stochastic
Fits today’s surfaceNoExactly, in ideal theoryApproximately by calibration
Forward dynamicsFlat smileSpot-drivenVariance + correlation driven
Primary strengthTransparent baselineVanilla-consistent diffusionRicher smile dynamics
Primary failureNo smileOften unrealistic forward skewParameter and calibration instability
ComputeLowMedium: PDE/MCMedium–high: Fourier/PDE/MC
Hedge implicationGreeks at one σState-localized vol hedgeVariance and vol-of-vol risk
Implementation with current QuantLib

Current QuantLib separates market structures, processes, instruments, engines and calibration helpers. Use those abstractions only after the lesson’s conventions, domains and numerical checks are explicit.

API authority: upstream QuantLib reference pinned in the source registry.
06
PYTHON LAB

Theory → implementation → checks.

PYTHON 3 · NUMPY / SCIPY

SABR parameter-domain guard

Validate a calibration state and compute the leading ATM scale.

REUSABLE EXAMPLE
01from __future__ import annotations
02
03import math
04
05def sabr_atm_leading(forward: float, alpha: float, beta: float, rho: float, nu: float) -> float:
06 if forward <= 0 or alpha <= 0 or nu < 0 or not 0 <= beta <= 1 or not -1 < rho < 1:
07 raise ValueError("invalid SABR parameter domain")
08 result = alpha / forward ** (1.0 - beta)
09 if not math.isfinite(result) or result <= 0:
10 raise ArithmeticError("invalid ATM volatility")
11 return result
12
13sigma = sabr_atm_leading(0.035, 0.03, 0.5, -0.25, 0.4)
14assert abs(sigma - 0.1603567451) < 1e-9
15print(f"{sigma:.4%}")
EXPECTED OUTPUT16.0357% leading-order ATM volatility.
SANITY CHECKS

Forward and α are positive.

β and ρ respect domains.

Known analytical leading value is asserted.

07
INTERACTIVE LAB

Move the state. Challenge the equation.

BACKBONE, CORRELATION AND VOL-OF-VOL

SABR smile lab

Move α, β, ρ and ν; inspect backbone, smile and parameter sensitivities.

SYNTHETIC · CONTROLLED SCENARIOS
ATM19.00%
ρ proxy-0.65
ν proxy0.45
ACTIVE STATE

Negative rhoCorrelation steepens downside skew. Move the intensity control and inspect every series with pointer or touch.

08
FRONT OFFICE

Where the model meets the book.

ON THE DESK
If β moves every day, the optimizer may be explaining the coordinate system.
VISIBLE INPUTS

forward and annuity

smile quotes

vol convention

shift

β policy

CALIBRATION

Fit α, ρ and ν to liquid smile quotes with bounded transforms and maturity-to-maturity regularization.

RISK

alpha level

rho skew

nu curvature

backbone delta

DAILY WORKFLOW
  1. normalize convention
  2. fix β/shift
  3. calibrate
  4. check limits
  5. compare stability
Production failure modes
  • wrong normal/lognormal quote
  • ATM cancellation
  • unbounded rho
  • parameter jumps
09
MACRO CONNECTION

Map the transmission channel.

MACRO CONNECTION

Rate regime and the SABR backbone

The forward level and volatility convention change how an identical basis-point shock appears in percentage-volatility space.

Rate regime

moves forward and sign constraints

Backbone β

maps level to volatility

Smile parameters

fit skew and curvature

Desk quote

depends on convention

10
COMMON PITFALLS

Most failures begin outside the formula.

01

Using lognormal SABR at non-positive forwards.

02

Calibrating every parameter without identifiability checks.

03

Ignoring the ATM limit.

04

Presenting asymptotic output as exact model price.

11
SOURCES / FURTHER READING

Attribution with implementation authority.

researchBSD-3-Clause

Volatility, Monte Carlo and stochastic-volatility lectures

Research map for the mathematical progression and numerical experiments; prose, examples and code are original.

Source
Computational Finance Course
Author
L. A. Grzelak
Ref
main
OPEN ORIGINAL SOURCE ↗LICENSE ↗
implementation referenceQuantLib permissive license

Current volatility structures, processes, calibration helpers and tests

Implementation reference for production abstractions and validation patterns.

Source
QuantLib upstream
Author
QuantLib contributors
Ref
v1.42.1
OPEN ORIGINAL SOURCE ↗LICENSE ↗