Market Microstructure Diffusion Limit Analytical Volatility

Cont & de Larrard (2013) Order Book Diffusion Lab

Simulate Rama Cont and Adrien de Larrard’s (2013) heavy-traffic limit for high-frequency limit order books. Explore conformal harmonic exit probabilities $p_{\text{up}} = \frac{2}{\pi}\arctan(q_b / q_a)$, the analytical microstructure volatility theorem $\sigma^2 = \frac{\pi \mu \delta^2}{4 D^2}$, and $t^{-1/2}$ heavy-tailed price durations.

Model: 2D Brownian Motion in $\mathbb{R}_+^2$ Exit Probability: Conformal Angle $\frac{2}{\pi}\arctan(q_b / q_a)$ Volatility Theorem: $\sigma = \frac{\delta}{D}\sqrt{\frac{\pi \mu}{4}}$
Theoretical Innovations
  • • Rigorous bridge from order flow to macro asset volatility
  • • Exact conformal mapping for 2D first-passage absorption
  • • Explains volatility clustering via $t^{-1/2}$ duration Pareto tails
  • • Invariant to queue size scaling under heavy traffic
P(Price Increase)
--%
$\frac{2}{\pi}\arctan(q_b/q_a)$
P(Price Decrease)
--%
$1 - p_{\text{up}}$
Macro Volatility $\sigma_{\text{ann}}$
--%
$\frac{\delta}{S_0} \frac{\sqrt{\pi \mu T_{\text{yr}}}}{2D}$
Mean Duration $E[\tau]$
--s
Between price changes
Jumps per Hour
--
Expected price moves / hr
Order Book Parameters
Configure queue depths and order flow arrival rates
Select a preset to inspect distinct market liquidity profiles.
1500
50 5,000
1500
50 5,000
2000
100 (Thin Book) 6,000 (Thick Book)
40
5 / sec 150 / sec
40
0.10
Microstructure Diagnostics
Harmonic $p_{\text{up}}$ Probability --%
Linear OBI Probability --%
Order Book Imbalance (OBI) --
Diffusion Coefficient $\sigma$ --
Annualized Dollar Volatility --
Expected Queue Life $E[\tau]$ --
Chart Notes: Tab 1 contrasts Cont-de Larrard’s conformal harmonic exit probability $\frac{2}{\pi}\arctan(q_b / q_a)$ with the naive linear order book imbalance. Tab 2 evaluates macroscopic price volatility scaling across order arrival intensities and book depth levels. Tab 3 depicts the heavy-tailed power law $P(\tau > t) \sim t^{-1/2}$ of inter-arrival durations between price movements.
5×5 Microstructure Sensitivity Heatmaps
Evaluate directionality and macro volatility across market states
Harmonic $p_{\text{up}}(q_b, q_a)$ Probability (%)
Bid Depth $q_b$ (rows) vs Ask Depth $q_a$ (cols) $\to \frac{2}{\pi}\arctan(q_b / q_a)$
$q_b \backslash q_a$ 200 500 1,000 2,000 4,000
Green: High upward price probability ($> 65\%$). Red: Downward price pressure ($< 35\%$).
Annualized Macro Volatility $\sigma_{\text{ann}}$ (%)
Order Rate $\mu$ (rows) vs Mean Depth $D$ (cols) $\to \frac{\delta}{S_0}\frac{\sqrt{\pi\mu T_{\text{yr}}}}{2D}$
$\mu \backslash D$ 200 500 1,000 2,000 4,000
Green: Low/stable volatility ($\le 20\%$). Red: Elevated/turbulent volatility ($> 35\%$).

Mathematical Foundations: Cont & de Larrard (2013)

1. Heavy-Traffic Diffusion Limit

In high-frequency markets, the number of limit orders, cancellations, and market orders per second is large. Under heavy-traffic scaling ($\lambda \approx \mu$), the discrete bid and ask queue processes $(q_b(t), q_a(t))$ converge weakly to two-dimensional Brownian motion in the positive quadrant $\mathbb{R}_+^2$:

$$d X_t = \sigma_q d W_t^b, \quad d Y_t = \sigma_q d W_t^a, \quad (X_t, Y_t) \in [0, \infty) \times [0, \infty)$$

The process terminates at the first exit time $\tau = \inf\{t > 0 : X_t = 0 \text{ or } Y_t = 0\}$. If $Y$ hits zero first (ask queue depleted), the mid-price jumps upward by one tick ($+\delta$). If $X$ hits zero first (bid queue depleted), the price moves downward ($-\delta$).

2. Conformal Harmonic Exit Probability

The absorption probability $p_{\text{up}}(x, y)$ satisfies Laplace’s equation with Dirichlet boundary conditions:

$$\Delta p_{\text{up}} = 0 \text{ in } \mathbb{R}_+^2, \quad p_{\text{up}}(x, 0) = 1, \quad p_{\text{up}}(0, y) = 0$$

By conformal mapping $z \mapsto z^2$ from the quadrant to the upper half plane, the harmonic measure yields the exact closed-form solution:

$$p_{\text{up}}(q_b, q_a) = \frac{2}{\pi} \arctan\left(\frac{q_b}{q_a}\right)$$

3. Microstructure-to-Macro Volatility Theorem

By the Functional Central Limit Theorem (FCLT) for semi-Markov renewal processes, the macroscopic mid-price process $P_t$ converges on long horizons to a diffusion with analytical variance:

$$\sigma^2 = \frac{\pi \mu \delta^2}{4 D^2} \implies \sigma_{\text{ann}} = \frac{\delta}{S_0} \frac{\sqrt{\pi \mu T_{\text{yr}}}}{2 D}$$
  • Tick Size $\delta$: Volatility scales linearly with minimum price increment.
  • Order Rate $\mu$: Volatility scales with $\sqrt{\mu}$ (Poisson transaction intensity).
  • Book Depth $D$: Volatility is inversely proportional to average resting queue depth.

4. Heavy-Tailed Price Duration Distribution

The first exit time $\tau$ of planar Brownian motion from a cone of angle $\pi/2$ has heavy Pareto tails:

$$\mathbb{P}(\tau > t) \sim \frac{2}{\pi} \frac{D}{\sqrt{\pi \mu t}} \propto t^{-1/2} \quad (\text{as } t \to \infty)$$

This explains why high-frequency financial data exhibits endogenous volatility clustering and long memory without requiring exogenous regime shifts.

Quantitative Mastery Quiz: Order Book Diffusion

Test your understanding of the Cont-de Larrard (2013) heavy-traffic limit, conformal exit probabilities, and microstructure volatility.

Frequently Asked Questions: Cont-de Larrard (2013) Model

Published in the SIAM Journal on Financial Mathematics, Rama Cont and Adrien de Larrard (2013) established a heavy-traffic diffusion limit for high-frequency limit order books. They proved that as order flow arrival rates become large, the discrete queue dynamics converge to a two-dimensional Brownian motion in the positive quadrant, yielding exact closed-form expressions for price move probabilities, duration tails, and macroscopic price volatility.

In the diffusion limit, the probability $p_{\text{up}}(q_b, q_a)$ that the ask queue depletes before the bid queue satisfies the 2D Laplace equation $\Delta u = 0$ in the positive quadrant with boundary conditions $u(x, 0) = 1$ and $u(0, y) = 0$. Solving this via conformal mapping or polar coordinates yields the exact harmonic solution $p_{\text{up}}(q_b, q_a) = \frac{2}{\pi} \arctan(q_b / q_a)$, contrasting with naive linear order book imbalance $q_b / (q_b + q_a)$.

Under the Functional Central Limit Theorem for semi-Markov renewal processes, Cont and de Larrard proved that low-frequency mid-price volatility is directly governed by order book microstructure: $\sigma^2 = \frac{\pi \mu \delta^2}{4 D^2}$, where $\mu$ is the market order arrival rate, $\delta$ is the tick size, and $D$ is the average depth of resting orders at the best quotes. Volatility is proportional to tick size and $\sqrt{\mu}$, and inversely proportional to book depth $D$.

The time $\tau$ between consecutive price changes corresponds to the first exit time of 2D planar Brownian motion from the positive quadrant. The tail probability asymptotically obeys $\mathbb{P}(\tau > t) \sim c \cdot t^{-1/2}$, producing an infinite-mean, heavy-tailed Pareto distribution that naturally replicates empirical volatility clustering and trade duration bursts without exogenous regime shifts.

Cont, Stoikov, and Talreja (2010) formulated the limit order book as a discrete-state birth-death Markov chain solving queue depletion times numerically via tridiagonal systems. Cont and de Larrard (2013) scaled the system to its continuous diffusion limit, deriving analytical closed-form formulas for conformal price change probabilities and proving the fundamental bridge between high-frequency order flow and macroscopic asset price volatility.

Related Market Microstructure & Volatility Labs

Cont-Stoikov Order Book Lab

Model discrete birth-death queue transitions, cancellation scaling ($\theta q$), and tridiagonal depletion times.

Order Flow Imbalance (OFI) Lab

Cont-Kukanov-Stoikov (2014) multi-level order flow imbalance regressions and queue depth resilience.

Gatheral SVI Volatility Lab

Calibrate Gatheral's SVI volatility smile, analyze Roger Lee wing bounds, and prevent butterfly arbitrage.