Research, Tools, and Market Insights
Market Insights
A curated library of analytical resources for practitioners working at the edge of liquidity provision risk.
Explore programsResources hub
Everything our team has built to support deeper analytical thinking
Liquidity provision risk sits at the intersection of microstructure theory, stochastic modeling, regulatory frameworks, and empirical market analysis. The resources collected here — research commentary, a comprehensive terminology glossary, practical modeling tips, market news, and upcoming events — are designed to support practitioners who are already working at a serious analytical level and need reference material that matches that depth. Our team maintains these resources on a rolling basis, revising content when new empirical findings or regulatory developments make an update analytically necessary.
Contact usContent update cycle
Latest Research and Market Commentary from Our Team
Dealer Balance Sheet Constraints and Market-Making Viability Under Stress
The question of whether a systematic liquidity provider can maintain activity through a market stress event is not purely a microstructure problem — it is a balance sheet problem that interacts with the microstructure dynamics in ways that standard adverse selection models do not fully capture. This article examines how funding liquidity constraints, as formalised through the LCR and NSFR frameworks, interact with stochastic inventory dynamics under the kind of correlated volatility and order flow imbalance that characterises a genuine market stress episode. We draw on the March 2020 Treasury market dislocation as the primary case study, examining where dealer balance sheet constraints became binding before order flow toxicity reached levels that would have triggered a withdrawal under a pure microstructure model. The analysis has direct implications for how practitioners should think about position sizing, spread widening decisions, and the sequencing of risk reduction actions when funding and market liquidity pressures arrive simultaneously rather than in the orderly succession that most models implicitly assume.
Regime-Switching in Equity Microstructure: What the Standard Models Miss
Most practitioners working in equity liquidity provision are familiar with the observation that market conditions shift — sometimes abruptly — between regimes characterised by very different adverse selection intensity, order flow composition, and intraday liquidity cycles. What is less commonly discussed is the extent to which the standard microstructure models used to inform spread-setting and inventory management decisions are structurally ill-equipped to detect and respond to those regime transitions in real time. This article examines the empirical evidence on regime-switching in equity microstructure from the perspective of systematic liquidity providers, reviewing the academic literature on Markov-switching models for order flow dynamics and assessing the operational gap between what those models can deliver in a calibrated research context and what they can deliver on a desk where the calibration must be updated continuously against live data with significant noise and structural non-stationarity. We propose a practical framework for monitoring regime indicators that does not require full model re-estimation at each observation interval, and examine its performance against selected equity market episodes in the Australian and international context.
Reference glossary
Liquidity Risk Terminology for Serious Practitioners
Bid-Ask Spread Decomposition
The analytical separation of the observed bid-ask spread into its three component parts: the order processing cost, the inventory holding cost, and the adverse selection cost. Each component reflects a distinct economic mechanism, and understanding their relative magnitudes in a given market context is essential for setting spreads that are both competitively viable and analytically defensible from a risk management perspective.
Adverse Selection Cost
The expected loss incurred by a liquidity provider when trading with a counterparty who holds superior information about the future direction of the asset's price. Adverse selection is the central risk in systematic liquidity provision — it is the mechanism through which informed order flow extracts value from the market maker's book, and it increases materially during periods of elevated information asymmetry or ahead of significant announcements.
Amihud Illiquidity Ratio
A widely used empirical measure of market illiquidity defined as the average ratio of absolute daily return to daily trading volume over a specified period. The ratio captures the price impact of order flow and is commonly applied in research on liquidity risk premia, dealer inventory management, and the cross-sectional relationship between illiquidity and expected asset returns in equity markets.
Inventory Risk
The exposure faced by a systematic liquidity provider arising from accumulated net positions that result from imbalanced order flow across the bid and ask sides of the book. Inventory risk combines directional price risk — the possibility that the accumulated position moves against the dealer — with funding risk, since carrying a large inventory position consumes balance sheet capacity and may attract regulatory capital charges under applicable prudential frameworks.
Liquidity Coverage Ratio
A prudential liquidity requirement under the Basel III framework, requiring financial institutions to maintain a sufficient stock of high-quality liquid assets to survive a standardised thirty-day stress scenario. The LCR interacts directly with the economics of systematic liquidity provision because high-quality liquid asset buffers compete with market-making inventory for balance sheet capacity, creating a regulatory constraint on dealer activity that becomes binding precisely when market liquidity is most stressed.
Net Stable Funding Ratio
A structural liquidity requirement under the Basel III framework designed to ensure that financial institutions maintain stable funding over a one-year horizon relative to the liquidity profile of their assets and off-balance-sheet activities. The NSFR constrains the degree to which dealers can fund long-dated or illiquid market-making positions through short-term wholesale borrowing, with direct implications for the viability and cost structure of systematic liquidity provision in fixed income markets.
Order Flow Toxicity
A measure of the degree to which the order flow arriving at a market maker's book is dominated by informed traders rather than uninformed liquidity demanders. High order flow toxicity indicates elevated adverse selection risk and is associated with deteriorating market-making conditions. The volume-synchronised probability of informed trading measure, developed in the academic microstructure literature, is one of the most widely applied empirical proxies for order flow toxicity in equity markets.
Avellaneda-Stoikov Framework
A stochastic optimal control framework for market-making that models the dealer's problem as the continuous-time optimisation of expected utility over a finite horizon, accounting for inventory accumulation, price impact, and the dealer's risk aversion. The framework produces closed-form expressions for optimal bid and ask quotes as functions of current inventory, time to horizon, and model parameters — and is a central reference point in the quantitative literature on systematic liquidity provision, both for its analytical elegance and for the practical limitations that emerge when its assumptions are tested against real market data.
Market Depth
A measure of a market's capacity to absorb order flow without generating material price impact, typically quantified as the volume of buy or sell orders available at successive price levels in the limit order book. Market depth is a forward-looking indicator of liquidity conditions and is a key input to the inventory risk assessment of systematic liquidity providers, since shallow depth implies greater price impact for any given position reduction and therefore higher effective carrying costs for accumulated inventory.
Fundamental Review of the Trading Book
A Basel Committee reform package restructuring the regulatory capital requirements for market risk in bank trading books, with significant implications for the economics of dealer market-making activity. The FRTB framework introduces a more granular and sensitive approach to capturing liquidity risk in capital calculations, replacing the value-at-risk measure with expected shortfall and introducing liquidity horizons that vary by asset class and instrument type — creating direct connections to the microstructure and inventory risk analytics covered in our curriculum.
Jump Process
A class of stochastic process that incorporates sudden discrete movements in a modelled quantity — such as asset price or order arrival rate — in addition to the continuous diffusion component captured by standard Brownian motion models. In the context of liquidity provision risk modeling, jump processes are used to represent sudden liquidity withdrawal events, order flow imbalances, or price dislocations that occur with low frequency but high impact and that are systematically underweighted by models built on purely continuous diffusion assumptions.
Adverse Selection in Fixed Income
The manifestation of information asymmetry in government and corporate bond markets, where sophisticated counterparties exploit informational advantages about credit conditions, central bank activity, or funding stress to trade at advantageous prices with liquidity providers. Adverse selection in fixed income markets operates differently from equity markets due to the over-the-counter structure of most bond trading, the role of dealer-to-dealer and dealer-to-client segmentation, and the periodic but significant impact of central bank and sovereign issuance activity on order flow composition and information dynamics.
Practical Modeling Tips for Liquidity Risk Practitioners
Concrete guidance on data sourcing, model validation, stress scenario construction, and the interpretation of liquidity metrics when market conditions move away from the assumptions embedded in standard frameworks.
Source bid-ask data at the right granularity for your model
The granularity at which you source bid-ask and order flow data materially affects the conclusions you can draw from microstructure analysis. Daily closing spreads are adequate for some cross-sectional liquidity risk premia work but are systematically misleading for intraday adverse selection analysis or inventory risk assessment. Before calibrating any spread decomposition model, establish whether your data source captures trade-level, quote-level, or interval-aggregated observations — and document explicitly how that choice affects the model's sensitivity to intraday regime shifts and order flow composition changes that occur within the aggregation window.
Source bid-ask data at the right granularity for your model
The granularity at which you source bid-ask and order flow data materially affects the conclusions you can draw from microstructure analysis. Daily closing spreads are adequate for some cross-sectional liquidity risk premia work but are systematically misleading for intraday adverse selection analysis or inventory risk assessment. Before calibrating any spread decomposition model, establish whether your data source captures trade-level, quote-level, or interval-aggregated observations — and document explicitly how that choice affects the model's sensitivity to intraday regime shifts and order flow composition changes that occur within the aggregation window.
Validate your adverse selection measure against known episodes
Any empirical proxy for adverse selection — whether volume-synchronised probability of informed trading, order flow imbalance, or a regression-based decomposition — should be validated against historical episodes where adverse selection is known to have been elevated before it is used as a live risk monitoring signal. Select two or three episodes relevant to your asset class and examine whether your chosen measure moved in the direction and with the timing you would expect. If it did not, understand why before relying on it as an input to spread-setting or position-sizing decisions in live market conditions.
Validate your adverse selection measure against known episodes
Any empirical proxy for adverse selection — whether volume-synchronised probability of informed trading, order flow imbalance, or a regression-based decomposition — should be validated against historical episodes where adverse selection is known to have been elevated before it is used as a live risk monitoring signal. Select two or three episodes relevant to your asset class and examine whether your chosen measure moved in the direction and with the timing you would expect. If it did not, understand why before relying on it as an input to spread-setting or position-sizing decisions in live market conditions.
Build stress scenarios from the inside out, not the outside in
The most analytically robust stress scenarios for liquidity provision books are constructed by first identifying the specific model assumptions that would need to fail simultaneously to create a severe outcome — and then asking whether a plausible market narrative could cause those failures to coincide. This inside-out approach, sometimes called reverse stress testing, surfaces the genuine tail risks embedded in a liquidity provision book more reliably than the conventional approach of starting from a macroeconomic narrative and working forward to estimated position-level impacts through a chain of assumptions that typically understates correlation during stress.
Build stress scenarios from the inside out, not the outside in
The most analytically robust stress scenarios for liquidity provision books are constructed by first identifying the specific model assumptions that would need to fail simultaneously to create a severe outcome — and then asking whether a plausible market narrative could cause those failures to coincide. This inside-out approach, sometimes called reverse stress testing, surfaces the genuine tail risks embedded in a liquidity provision book more reliably than the conventional approach of starting from a macroeconomic narrative and working forward to estimated position-level impacts through a chain of assumptions that typically understates correlation during stress.
Interpret the Amihud ratio in the context of market structure changes
The Amihud illiquidity ratio is a robust and widely used measure, but its level is not stationary across structural changes in market microstructure — such as the introduction of new trading venues, changes in tick size regimes, or significant shifts in the composition of active participants. When using the Amihud ratio to compare liquidity conditions across time periods that span a structural change in market design, apply a break-point analysis before drawing conclusions about trend. A declining ratio may reflect genuine improvement in market liquidity or simply a change in the relationship between volume and price impact driven by venue fragmentation or participant composition shifts.
Interpret the Amihud ratio in the context of market structure changes
The Amihud illiquidity ratio is a robust and widely used measure, but its level is not stationary across structural changes in market microstructure — such as the introduction of new trading venues, changes in tick size regimes, or significant shifts in the composition of active participants. When using the Amihud ratio to compare liquidity conditions across time periods that span a structural change in market design, apply a break-point analysis before drawing conclusions about trend. A declining ratio may reflect genuine improvement in market liquidity or simply a change in the relationship between volume and price impact driven by venue fragmentation or participant composition shifts.
Calibrate jump process parameters with a held-out validation window
When calibrating Poisson or compound jump process parameters to model sudden liquidity withdrawal events, reserve a portion of your historical data as a held-out validation window rather than using the full sample for calibration. Jump events are rare enough that parameter estimates are sensitive to which specific episodes fall within the calibration window — and a model that fits the calibration period well may perform poorly when the next tail event arrives with a different arrival rate or magnitude distribution. The held-out window allows you to assess out-of-sample performance before committing the calibrated model to a live risk monitoring role.
Calibrate jump process parameters with a held-out validation window
When calibrating Poisson or compound jump process parameters to model sudden liquidity withdrawal events, reserve a portion of your historical data as a held-out validation window rather than using the full sample for calibration. Jump events are rare enough that parameter estimates are sensitive to which specific episodes fall within the calibration window — and a model that fits the calibration period well may perform poorly when the next tail event arrives with a different arrival rate or magnitude distribution. The held-out window allows you to assess out-of-sample performance before committing the calibrated model to a live risk monitoring role.
Connect regulatory liquidity metrics to microstructure dynamics explicitly
LCR and NSFR calculations are often managed as balance sheet compliance exercises with limited connection to the desk-level microstructure analytics used to monitor market-making risk. This separation creates a blind spot: the conditions under which regulatory liquidity buffers become binding — simultaneous funding stress and market stress — are precisely the conditions under which adverse selection intensity and inventory accumulation risk are most elevated. Building an explicit analytical bridge between your regulatory liquidity metrics and your microstructure risk monitoring allows you to identify the combined stress threshold earlier and with more precision than either framework achieves in isolation.
Connect regulatory liquidity metrics to microstructure dynamics explicitly
LCR and NSFR calculations are often managed as balance sheet compliance exercises with limited connection to the desk-level microstructure analytics used to monitor market-making risk. This separation creates a blind spot: the conditions under which regulatory liquidity buffers become binding — simultaneous funding stress and market stress — are precisely the conditions under which adverse selection intensity and inventory accumulation risk are most elevated. Building an explicit analytical bridge between your regulatory liquidity metrics and your microstructure risk monitoring allows you to identify the combined stress threshold earlier and with more precision than either framework achieves in isolation.
Upcoming Events and Practitioner Sessions
Stochastic Inventory Modeling in Fixed Income Markets
A live session examining the application of stochastic optimal control frameworks to fixed income market-making, with particular attention to how funding constraints alter the optimal quoting strategy derived under the standard Avellaneda-Stoikov assumptions. Case material drawn from sovereign bond market episodes.
Adverse Selection Metrics: What Works and What Misleads
A practical workshop examining the most commonly used empirical proxies for adverse selection intensity — including volume-synchronised probability of informed trading and order flow imbalance measures — assessing their reliability across market conditions and discussing the scenarios where each measure systematically misleads practitioners who rely on it without understanding its construction assumptions.
Regulatory Frameworks and Dealer Economics: LCR, NSFR, and FRTB
This session covers the interaction between Basel III prudential requirements and the economics of systematic liquidity provision, examining how LCR and NSFR constraints shape dealer behaviour during stress and how FRTB capital requirements change the cost structure of maintaining a market-making book across different asset classes and liquidity horizons.
Australian Financial Markets Conference — Liquidity Risk Panel
Members of our research team will participate in a panel discussion on evolving liquidity provision risk dynamics in Australian equity and fixed income markets, covering recent empirical findings on market depth, order flow toxicity, and the interaction between prudential regulation and market-making viability in the current macroeconomic environment.
Stress Testing Liquidity Books: Reverse Scenario Design Workshop
A hands-on workshop in the reverse stress testing methodology documented in our curriculum, guiding participants through the construction of inside-out stress scenarios for liquidity provision books. Participants will work through a structured exercise using historical episode data from equity and fixed income markets as the empirical foundation for scenario calibration.
