The Heston model is a stochastic-volatility model in which the asset price and its instantaneous variance
satisfy
|
(1)
| |||
|
(2)
|
with correlated Wiener processes
|
(3)
|
Here stochastic volatility means that the volatility is itself a stochastic
process, rather than a fixed parameter. The variance
is mean-reverting:
is the mean-reversion rate,
is the long-run variance,
is the volatility of variance,
and
is the instantaneous correlation
between price and variance shocks. The condition
is the usual Feller
condition ensuring that zero is unattainable when
.
The logarithmic price
and
form an affine diffusion: their conditional
characteristic function has the exponential-affine
form
|
(4)
|
where
and
solve Riccati
differential equations. Here affine means that the logarithm of the transform
is an affine function of the current state variables
and
. This structure gives Fourier-based formulas for values of
European options while allowing volatility skew
and time-varying conditional variance. Unlike the constant-volatility
model of Black-Scholes theory, the Heston
model has a random variance factor.
Time-inhomogeneous extensions allow ,
,
,
,
, and
to depend on time. For coefficients that are piecewise
constant functions, the exponential-affine transform can be propagated recursively
through the time intervals by solutions of the Riccati
differential equations, preserving methods based on the characteristic
function while fitting a term structure of volatility skew (Andersen et al.
2026).