Linear Price Impact Model Overview

• The Linear Price Impact Model is a parsimonious framework that relates mid-price changes directly to order flow imbalances, with a coefficient inversely proportional to market depth.
• Empirical studies demonstrate a robust linear relation with an R² of ~65% across stocks and reveal dynamic liquidity patterns, notably a U-shaped intraday depth variation.
• The model underpins modern optimal execution theory by explaining how linear microstructural impacts scale to a square-root law over longer intervals and guide risk management.

A linear price impact model posits that price changes are proportional to an observable imbalance in order flow, typically measured at the best bid and ask quotes, with a proportionality coefficient inversely related to the available liquidity at those price levels. This framework provides a parsimonious, empirically robust description of high-frequency market microstructure, underlies much of modern optimal execution theory, and links to both microstructural and macroeconomic perspectives on price dynamics.

1. Core Definition and Mathematical Structure

The canonical linear price impact model is defined by the relationship between price changes and the order flow imbalance (OFI) over a time interval Δt. Let $P^{A}$ and $P^{B}$ denote the best ask and best bid prices, respectively, with $P_{\mathrm{mid}}=(P^{A}+P^{B})/2$ as the midpoint. Over interval Δt, the mid-price change (in ticks) is

$$\Delta P(\Delta t) = \frac{ P_{\mathrm{mid}}(t+\Delta t) - P_{\mathrm{mid}}(t) }{ \delta }$$

where $\delta$ is the tick size.

The order flow imbalance is defined as

$$\mathrm{OFI}(\Delta t) = \sum_{n=1}^{N(\Delta t)} e_n$$

where $N(\Delta t)$ is the number of quote updates/events in Δt, and $e_n$ is the signed change in queue sizes at best quotes due to event $n$ (positive if a limit order increases bid or decreases ask, negative otherwise). The instantaneous market depth at the best quotes, $AD$, is the average available size.

Empirically, the model establishes the linear relation

$$\Delta P_k = \beta_i \cdot \mathrm{OFI}_k + \epsilon_k$$

for interval $k$ in half-hour block $i$, with $\beta_i$—the price impact coefficient—scaling inversely with depth: $\beta_i \approx \frac{c}{AD_i^\lambda}$ where $c\approx 0.4$–$0.5$, $\lambda\approx 1$ (Cont et al., 2010). Doubling the market depth halves the slope of price impact, confirming order-book theoretical reasoning.

2. Empirical Performance and Parameterization

The robustness of the linear price impact relation is evidenced by:

• Statistical fit: Across 50 US stocks, regression $R^2$ for $\Delta P$ vs. OFI is typically $\approx65\%$, substantially outperforming regressions on trade imbalance alone ($R^2\approx32\%$).
• Parameter values: Average $\beta_i\sim0.04$ ticks per share of OFI (10-second interval), with 95% confidence range $[0.02,0.06]$ ticks/share. Typical depths $AD_i$ of 200–500 shares yield $c\approx0.4$–$0.6$.
• Intraday dynamics: The depth follows a U-shaped pattern, lowest at the open/close. Consequently, price impact coefficients $\beta_i$ are maximal at open ($\approx2\times$ daily mean), lowest at close ($\approx0.5\times$ mean). The variance in $\Delta P$ aligns with $\beta_i^2\cdot\mathrm{Var}(\mathrm{OFI}_k)$, explaining classic volatility seasonality.

The linear model remains stable from sub-second intervals to several-minute sampling windows, showing only modest increases in $R^2$ with interval length (Cont et al., 2010).

3. Microstructural Basis and Limitations

Linear price impact is conceptually rooted in the supply-demand imbalance at the best quotes, integrating market, limit, and cancellation order effects. This mechanistic insight stands in contrast to models considering only executed trades, which fail to capture the observed explanatory power.

Limitations include:

• Level-I data restriction: Only best-quote depth is observed; hidden depth generates noise.
• Absence of permanent nonlinear effects: The model does not accommodate latency or nonlinear price shifts lasting beyond minutes.
• Low-liquidity breakdown: In illiquid assets or during rapid market events, core i.i.d. and limit theorems assumptions may fail.

Volume-only regression models yield much lower exponents than the square-root law for impact (estimated $H\sim0.2$–$0.4$ versus $1/2$), and the explanatory power collapses when OFI is included jointly, indicating that concavity in volume impact is a secondary, indirect effect induced by OFI–volume correlation (Cont et al., 2010).

4. From Linear Impact to Square-Root Law

Despite strict linearity at the event-by-event level, a scaling argument reveals that over longer intervals the linear OFI-law produces a square-root law for impact versus volume: $$|\Delta P_k| \approx \Theta_k \cdot (\mathrm{VOL}_k)^{1/2}$$ where $\Theta_k$ fluctuates randomly about its mean (Cont et al., 2010). This matches empirical observations and theoretical explanations in models with persistent order flow and random execution sizes, e.g., nearly-critical Hawkes processes (Jaisson, 2014), and reconciles microstructural linear impact with the ubiquitous concave square-root law prevalent in empirical market impact studies.

5. Theoretical Justifications and Extensions

The linear price impact paradigm is theoretically linked to:

• No-manipulation constraints: Permanent linear impact is essentially the unique law that precludes price-manipulation strategies in round-trip trades (0903.2428).
• Microstructural auction and game theory: Continuous-time market-making models yield the same linear impact law for both temporary and permanent components, derived as Nash equilibrium responses to order flow and dealer risk aversion (Singh, 2021).
• Optimal execution theory: Linear impact models underpin dynamic programming and utility-optimization frameworks for optimal trading, where impact costs enter as quadratic terms, and asymptotic analyses for small impact costs provide explicit formulae for trading policies and welfare losses (Moreau et al., 2014, Dolinskyi et al., 2023, Dolinsky, 29 Sep 2025).

Extensions include explicit modeling of transient versus permanent impact and context-dependent adaptation of impact coefficients via state variables such as market depth, volatility, and risk aversion.

6. Model Limitations and Transition to Nonlinear Regimes

At the finest time scales (tick-by-tick), empirical work demonstrates significant deviations from pure linearity:

• Concavity and nonlinearity: At single-trade or high-frequency resolutions, price impact is typically sub-linear, scaling as $|v|^\Phi$ with $\Phi\sim0.1$–$0.3$, and impact per unit volume declines with larger trade sizes (0903.2428, Patzelt et al., 2017).
• Meta-order execution: Even models with a strictly linear local impact rule yield globally concave expected impact for large meta-orders due to endogenous order book refill dynamics (Nadtochiy, 2020).
• Aggregate impact master curves: Universal empirical “master curves” for aggregated volume/sign impact are strongly nonlinear, with saturation or reversion at extremes, and decay of linear slope with aggregation timescale (Patzelt et al., 2017, Patzelt et al., 2017).
• Auction effects: Auction markets can exhibit regimes of zero impact, linear impact, and super-linear impact depending on the order size relative to the visible liquidity, a structure not captured by continuous-trading linear models (Salek et al., 2023).

The "square-root law" emerges as a macroscopic consequence of persistent order-flow memory, endogenous liquidity adaptation, and averaging effects present at coarser time scales.

7. Practical Implications for Trading and Risk Management

For execution algorithms and transaction cost budgeting:

• Execution scheduling: Execution strategies must account for time-varying depth, lowest at market open/close, and dynamically adapt to the U-shaped intraday liquidity profile for minimizing impact cost (Cont et al., 2010).
• Front-loading and risk-adjustment: The concave impact cost for meta-orders suggests front-loading schedules and risk-adjusted dynamic execution policies (Nadtochiy, 2020, Moreau et al., 2014).
• Real-time monitoring: Order-book imbalance and its drift serve as signals for execution acceleration or slowdown.
• Calibration: Robust estimation of price impact coefficients requires systematic regression against OFI using high-frequency order-book data and adjusting for market depth variability.

While linear price impact forms the backbone of optimal execution theory and market microstructure modeling, practical applications in real-world high-frequency trading require explicit recognition of nonlinear effects, transient resiliency, meta-order-induced concavity, and tick-size discreteness.

---

Key References:

"Price Impact of Order Book Events" (Cont et al., 2010); "Price Impact" (0903.2428); "Universal Scaling and Nonlinearity of Aggregate Price Impact" (Patzelt et al., 2017); "A Simple Microstructural Explanation of the Concavity of Price Impact" (Nadtochiy, 2020); "Market Impact as Anticipation of the Order Flow Imbalance" (Jaisson, 2014); "A Model of Market Making and Price Impact" (Singh, 2021); "Trading with Small Price Impact" (Moreau et al., 2014); "Optimal Liquidation with High Risk Aversion and Small Linear Price Impact" (Dolinskyi et al., 2023); "Price impact in equity auctions: zero, then linear" (Salek et al., 2023).