Optimal Transaction Quotes

• Optimal transaction quotes are defined as the bid and ask prices set by algorithms to balance transaction costs, price and informational risks, and regulatory fees.
• They are derived using methods from optimal control, stochastic calculus, and asymptotic analysis, leading to insights such as the cube-root law and the formulation of no-trade bands.
• Practical implementations incorporate latency effects, market impact, adverse selection, and regulatory frameworks, enabling robust market making and multi-asset trading strategies.

Optimal Transaction Quotes represent the fundamental prices at which a market participant or algorithm chooses to buy (bid) or sell (ask) an asset, given specific objectives and constraints such as transaction costs, price risk, informational risk, execution latency, and regulatory fees. Determining such quotes is a central problem in modern financial engineering, blending optimal control, stochastic calculus, asymptotic analysis, and market microstructure. This article surveys the principal models, asymptotic laws, algorithmic recipes, and implementation guidelines for optimal transaction quoting, referencing recent research on execution, market making, inventory risk, informational asymmetries, latency, and liquidity.

1. Asymptotic Regimes for Transaction Costs and No-Trade Bands

In regimes with small proportional transaction costs (one-sided cost per unit traded $\epsilon \ll 1$), the optimal trading problem is characterized by "no-trade" bands—regions in which positions are not adjusted, as the cost of trading exceeds the marginal benefit of tracking the frictionless target. The universal law for the width of the optimal no-trade band in position space is

$$\delta\theta^* \simeq \left( \frac{3\epsilon G \widehat{\Gamma}_0^2}{2} \right)^{1/3}$$

where $G$ is a risk-budget scale, and $$\widehat{\Gamma}_0^2 = \frac{\sigma_{\hat{\theta}}^2}{\sigma_X^2}$$ quantifies the trading-speed relative to asset volatility (with $\sigma_{\hat{\theta}}$ the instantaneous volatility of the frictionless target and $\sigma_X$ that of the asset price) (Martin, 2016, Mayerhofer, 2024).

To convert these position-space bands into price-space quotes, when the frictionless target $\hat{\theta}$ is a smooth function of the mid-price $p$ (e.g., $\hat{\theta}(p)$), the half-spread in price is

$$\delta p^* = \frac{\delta\theta^*}{\hat{\theta}'(p)}$$

thus optimal bid and ask quotes are $p \pm \delta p^*$. For delta-hedging a European option, this yields

$$\delta S^* = \left( \frac{3\epsilon G}{2} \right)^{1/3} \Gamma^{-1/3}$$

with $\Gamma = \partial^2 C/\partial S^2$ the option gamma. The familiar cube-root dependence on transaction cost, risk aversion, and target speed emerges as a universal feature in small-cost asymptotics (Martin, 2016, Mayerhofer, 2024, Mayerhofer, 2024).

2. Control-Limit Strategies, Shadow Prices, and Higher-Order Corrections

The optimal policy with proportional transaction costs takes the form of a reflected diffusion for the risky fraction $\pi_t$ in the interval $[\pi_-, \pi_+]$ around the frictionless target $\pi^* = \mu/(\gamma \sigma^2)$ (for CRRA utility of risk aversion $\gamma$) (Mayerhofer, 2024, Mayerhofer, 2024). The shadow-price approach identifies a fictitious frictionless mid-price $S_t^\epsilon = g(\pi_t)S_t$ constrained between $[(1-\epsilon)S_t, S_t]$, with $g$ satisfying free-boundary and smooth-pasting conditions.

Band locations admit higher-order asymptotic expansions: $$\pi_\pm = \pi^* \pm \Delta - \frac{(1-\gamma)\pi^*}{3}\left(\frac{4\gamma}{\pi^*(1-\pi^*)}\right)^{1/3} \epsilon^{2/3} + O(\epsilon)$$ with $$\Delta = \left[ \frac{\pi^*(1-\pi^*)}{4\gamma} \right]^{1/3}\epsilon^{1/3}$$. The leading cost appears at $O(\epsilon^{2/3})$ in long-run growth or equivalent safe rate (ESR), with the $O(\epsilon)$ correction explicit (Mayerhofer, 2024). Fourth-order improvements for $\gamma \neq 1$ are implementable by perturbed bands, yielding ESR improvement at $O(\epsilon^{4/3})$. The shadow-price approach, however, becomes inapplicable in the risk-neutral ($\gamma=0$) case, where no frictionless price exists to reproduce the two-boundary rebalancing policy (Mayerhofer, 2024).

Discrete-trade implementations maintain $O(\epsilon)$-optimality by executing finite-size trades of size $\Delta\pi \sim \epsilon^{2/3}$ at hitting times of inner boundaries, preserving scaling properties of annualized transaction costs (Mayerhofer, 2024).

3. Latency and Execution: Brownian Hitting-Time Models

When quoting into limit order books, execution latency fundamentally alters optimal quoting. In models where price evolves as a Brownian motion and order placement has deterministic round-trip delay $\ell$ (e.g., network latency or exchange queues), fill probabilities for limit orders decompose into immediate market fills and genuine resting fills, both as explicit functions of the quoted distance $y$ and $\ell$: $$P_y(\text{market fill}) = \Phi\left(-\frac{y}{\sqrt{\ell}}\right), \quad P_y(\text{limit fill}) \approx 2\Phi(-y)(1-\Phi(-y/\sqrt{\ell})) + \frac{2}{\sqrt{2\pi\ell}}e^{-y^2/(2\ell)}$$ The optimal choice of $y_k^*$ for posting quotes at stage $k$ solves a Bellman recursion incorporating these probabilities and an exponential-utility (mean-variance) reward functional. Critical findings are that increasing latency $\ell$ requires posting deeper quotes (larger $y$), results in slower execution, reduces mean net profit, and increases variance, as observed in simulation and closed-form comparative statics (Ma et al., 1 Apr 2025).

4. Informational Risks: Adverse Selection and Price Reading

In market making for RFQ and electronic platforms, the optimal quoting problem must incorporate both adverse selection from informed traders and price reading from directional algorithms that infer inventory from quote asymmetry. The market-maker’s value function satisfies a nonlocal, nonlinear HJB equation, with optimal quotes at each tier and trade size given by

$$\widehat{\delta}^{n,k,b*}(q) = \tilde{\delta}^{n,k,b*}\left(D_+^k \widehat{V}(q)\right), \quad \widehat{\delta}^{n,k,a*}(q) = \tilde{\delta}^{n,k,a*}\left(D_-^k \widehat{V}(q)\right)$$

where $D_\pm^k$ are finite differences in the inventory variable, and $\tilde{\delta}$ is informed by arrival intensities, price impact of informed flow, and price substitution effects arising from inventory-dependent quote skew (Barzykin et al., 27 Aug 2025). In the exponential-intensity case, closed-form feedback rules are available.

Key comparative statics show that increased adverse selection may either flatten or steepen the value function curvature depending on the distribution of informed flow, causing either spread narrowing or widening. Price reading always widens the overall spread but flattens the skew presented to tier clients sensitive to inventory asymmetry.

Calibration and practical implementation depend on regressing RFQ arrival intensities versus quoted offsets, estimating adverse selection kernels and price-reading effects from post-trade slippage, and updating the approximation at inventory jumps. Hard inventory cutoffs, tier/size-specific risk limits, and real-time monitoring are recommended (Barzykin et al., 27 Aug 2025).

5. High-Frequency and Multi-Asset Quoting under Liquidity Constraints

In high-frequency environments and multi-asset spread trading, optimal quoting generalizes via joint modeling of temporal order-flow (using multivariate Hawkes processes) and instantaneous structural liquidity (via a Composite Liquidity Factor, "CLF"). The Hawkes process captures clustering and cross-excitation in order flow, delivering forward simulation intensities for estimating reference price stability, while the CLF quantifies the instantaneous slippage based on limit order book (LOB) depth and price discontinuities (Anantha et al., 8 Jul 2025).

A rolling voting framework fuses dynamic (Hawkes) and static (CLF) signals to adaptively select the quoting reference (e.g., which contract's side to peg to). Backtests indicate that integrated Hawkes–CLF voting reduces average per-trade slippage and aligns more closely with market-optimal quoting than either signal alone, with a reported slippage reduction from 2.3 INR (naïve) to 0.95 INR in exchange-traded derivatives on India's NSE. The architecture is robust to regime shifts in flow or liquidity but imposes computational costs for real-time kernel reestimation (Anantha et al., 8 Jul 2025).

6. Regulatory and Institutional Constraints: Principal-Agent Models and Make-Take Fees

In contexts where market making is incentivized via exchange-imposed make-take fees or contracts, the optimal quoting problem extends to principal-agent frameworks. The exchange offers a fee contract, and the market-maker's optimal quotes incorporate inventory risk, volatility, fee pass-through, and risk-sharing adjustments. The closed-form optimal quote (relative to the mid-price $S_t$ and inventory $q$) is

$$\delta^{i,*}(t,q) = \frac{\sigma}{k}\ln\frac{u(t,q)}{u(t,q-\epsilon_i)} + \frac{1}{\gamma}\ln\left(1+\frac{\sigma\gamma}{k}\right) - c - \frac{1}{\eta}\ln\left(1 - \frac{\sigma^2\gamma\eta}{(k+\sigma\gamma)(k+\sigma\eta)}\right)$$

where $c$ is the taker fee, $\gamma$ and $\eta$ the risk aversion of market-maker and exchange, and $u$ solves a linearized value function. Optimal contracting narrows spreads, increases liquidity, and adjusts quotes in response to shared risks (Euch et al., 2018).

7. Execution and Algorithmic Implementation: Market Impact, Risk, and Utility

Optimal execution of parent orders under market impact and transaction costs is typically approached via mean-variance or tail-risk utility functionals. Canonical models, such as Almgren–Chriss, prescribe the "sinh" law for continuous-time trading rates: $$X(t) = N\frac{\sinh[\kappa(T-t)]}{\sinh[\kappa T]}, \qquad v(t) = -X'(t)$$ with $\kappa = \sqrt{\lambda\sigma^2/\gamma}$, $\gamma$ the market-impact constant, $\lambda$ risk aversion, and $\sigma$ volatility. Nonlinear impact models, heavy-tailed returns, and alternate risk objectives (e.g., tail-drawdown metrics) require numerical optimization (Marcos, 2020).

Practical implementation involves calibration from historical transaction cost analysis (TCA), selection of impact model, and discretization into executable child orders. In high-frequency and microstructural contexts, adaptation of optimal quotes to microstructure, volatility, and real-time order book state is critical for minimizing realized execution costs (Marcos, 2020).

---

Summary Table: Key Optimal Quote Formulas under Selected Regimes

Setting	Optimal Quote Formula	Reference
Small proportional cost	$$\delta p^* = [(3\epsilon G(\sigma_{\hat{\theta}}^2/\sigma_X^2)/2]^{1/3}/\hat{\theta}'(p)$$	(Martin, 2016)
Control-limit bands	$$\pi_\pm = \pi^* \pm \Delta - \frac{(1-\gamma)\pi^*}{3}\cdots\epsilon^{2/3} + O(\epsilon)$$	(Mayerhofer, 2024)
Market making w/ latency	$y_k^*$ optimizes Bellman recursion with latency-dependent fill probabilities	(Ma et al., 1 Apr 2025)
Informational risk	$\widehat{\delta}^{n,k,b/a*}(q)$ via value-function finite differences and informational kernels	(Barzykin et al., 27 Aug 2025)
Make-take contract	$\delta^{i,*}(t,q)$ as explicit sum of inventory, risk-floor, fee, and risk-sharing adjustments	(Euch et al., 2018)
Execution (impact/risk)	AC: $X(t) = N\sinh[\kappa(T-t)]/\sinh[\kappa T]$, $v(t)=-X'(t)$	(Marcos, 2020)

---

• Universal trading under proportional transaction costs (Martin, 2016)
• The effect of latency on optimal order execution policy (Ma et al., 1 Apr 2025)
• High Frequency Quoting Under Liquidity Constraints (Anantha et al., 8 Jul 2025)
• Optimal Quoting under Adverse Selection and Price Reading (Barzykin et al., 27 Aug 2025)
• Asymptotic methods for transaction costs (Mayerhofer, 2024)
• Almost Perfect Shadow Prices (Mayerhofer, 2024)
• Transaction Costs in Execution Trading (Marcos, 2020)
• Optimal make-take fees for market making regulation (Euch et al., 2018)