Order Flow Imbalance (OFI) & Price Impact Lab

Explore Rama Cont, Arseniy Kukanov, and Sasha Stoikov's (2014) celebrated framework for limit order book event decomposition, contemporaneous price impact, and high-frequency market depth estimation.

Price Impact β -- Cents per 1k shs
Explanatory Power R² -- Fit with ΔP
Market Depth (1/β) -- Shares per $1.00 move
Net OFI / ΔP -- --

Limit Order Book Parameters

1 tick
1,200 shs
200 shs
+5%
15%
200
Cont-Kukanov-Stoikov (2014)
Regression t-stat: --
Avg Quoted Spread: --
Avg Level 1 Depth: --

Theoretical Foundations & Mathematical Formulation

1. Order Flow Imbalance (OFI) Definition

In modern electronic limit order books, price changes do not occur in continuous time; rather, they arise from discrete events: limit order additions, market order executions, and order cancellations. Rama Cont, Arseniy Kukanov, and Sasha Stoikov (2014) defined the Level 1 Order Flow Imbalance between events \(t-1\) and \(t\) as:

\[ OFI_t = e_{b,t} - e_{a,t} \]

where the bid event flow \(e_{b,t}\) and ask event flow \(e_{a,t}\) are conditionally decomposed based on price changes:

\[ e_{b,t} = \begin{cases} q_{b,t}, & P_{b,t} > P_{b,t-1} \\ q_{b,t} - q_{b,t-1}, & P_{b,t} = P_{b,t-1} \\ -q_{b,t-1}, & P_{b,t} < P_{b,t-1} \end{cases} \]
\[ e_{a,t} = \begin{cases} -q_{a,t}, & P_{a,t} < P_{a,t-1} \\ q_{a,t-1} - q_{a,t}, & P_{a,t} = P_{a,t-1} \\ q_{a,t-1}, & P_{a,t} > P_{a,t-1} \end{cases} \]

2. Contemporaneous Price Impact & Market Depth

Cont, Kukanov, and Stoikov demonstrated that contemporaneous mid-price changes \(\Delta P_t = M_t - M_{t-1}\) exhibit a remarkably tight linear relationship with \(OFI_t\):

\[ \Delta P_t = \alpha + \beta \cdot OFI_t + \epsilon_t \]

where the price impact coefficient \(\beta = \frac{\text{Cov}(\Delta P, OFI)}{\text{Var}(OFI)}\) measures the price sensitivity per unit of volume imbalance. Inverting this coefficient yields the Market Depth / Liquidity Resilience:

\[ D = \frac{1}{\beta} \quad \left[\text{Shares required to move price by } \$1.00\right] \]

Unlike static book volume imbalance (VOI) which ignores the velocity and direction of incoming orders, OFI captures the exact transition dynamics, resulting in empirical \(R^2\) values exceeding 70% to 85% in major equity and futures markets.

Dual Institutional Sensitivity Analysis

Sensitivity 1: Spread Width vs Order Size on Price Impact β (¢ / 1k shs)

Evaluates how quoted bid-ask spread and average order sizes affect the marginal price impact per 1,000 shares of order flow imbalance.

Spread \ Order Size 100 shs 200 shs 300 shs 400 shs 500 shs

Sensitivity 2: Flow Bias (θ) vs Cancellation Rate (γ) on Regression R² (%)

Analyzes how directional order flow accumulation and quote cancellation noise alter the explanatory power \(R^2\) of the OFI price impact regression.

Flow Bias \ Cancel Rate 10% 20% 30% 40% 50%

Institutional Mastery Quiz: Order Flow Imbalance

1. Under the Cont-Kukanov-Stoikov (2014) definition, if the best bid price increases from \(P_{b,t-1}\) to \(P_{b,t}\) (\(P_{b,t} > P_{b,t-1}\)), what is the bid event flow \(e_{b,t}\)?

2. Why does Order Flow Imbalance (OFI) exhibit significantly higher empirical \(R^2\) than static Volume Imbalance (VOI)?

3. In market microstructure, how is market depth \(D\) formally related to the price impact slope \(\beta\)?

4. What is the primary econometric effect of very high order cancellation rates (γ > 50%) on the OFI regression?

Frequently Asked Questions

Order Flow Imbalance (OFI), formalized by Rama Cont, Arseniy Kukanov, and Sasha Stoikov (2014), is a quantitative metric measuring the net volume change at the best bid and ask quotes between consecutive limit order book events. It accounts for price-level shifts, new limit order arrivals, cancellations, and market order executions.

Static Volume Imbalance (VOI) merely measures the ratio or difference between resting bid depth and ask depth at a single point in time. In contrast, OFI tracks dynamic event transitions over time, separating aggressive price improvements, size additions, and queue cancellations to explain contemporaneous price changes.

Cont, Kukanov, and Stoikov demonstrated that contemporaneous mid-price changes are essentially linear in OFI, with empirical R^2 values frequently exceeding 65% to 85% over short intervals. This high explanatory power arises because price changes in limit order books are mechanically driven by the consumption and replenishment of queue liquidity.

Multi-level OFI extends the best bid/ask Level 1 formulation to deeper book levels (L2 through L5). Each depth level contributes to cumulative buying or selling pressure, typically weighted by an exponential decay kernel that assigns lower weight to quotes further away from the current mid-price.

Algorithmic execution desks use OFI to gauge adverse selection and calibrate trade urgency: when OFI surges in the direction of the trade, algorithms accelerate execution to beat price drift. High-frequency market makers monitor OFI to skew quotes asymmetrically, fading adverse flow and avoiding toxic fill runs.

Related Microstructure & Algorithmic Execution Simulators

Gatheral Transient Impact Lab

Model optimal execution under decaying transient impact, order book resilience kernels, and absence of price manipulation.

Avellaneda-Stoikov Market Making Lab

Model high-frequency optimal quote placement, reservation price indifference, and inventory control.

VPIN Order Flow Toxicity Lab

Calculate Volume-Synchronized Probability of Toxicity and forecast toxic order flow runs.

Bipower Variation & Jump Test Lab

Disentangle high-frequency continuous integrated volatility from discrete jumps using BNS tests.