Learning Objectives
- Explain how liquidity, order types, market design, and intraday trading regimes shape observed market data and execution quality.
- Distinguish among major market data products, including L1, L2, L3, TAQ, and enriched bar datasets, and choose data that matches a research or trading objective.
- Parse message-based exchange data and reconstruct a venue-local limit order book while enforcing core lifecycle and accounting invariants.
- Interpret key order-book measures and empirical microstructure patterns, while recognizing the limits of visible single-venue data.
- Build and compare time-, activity-, and information-driven bars, including when trade-direction classification and Lee-Ready alignment are required.
- Apply intraday data-quality and sessionization checks that prevent sequencing, timestamp, and calendar errors from contaminating downstream analysis.
How microstructure impacts price formation
Market data is the observable output of matching engine rules plus order flow, and understanding these mechanics is essential for realistic strategy design. The section decomposes liquidity into spread, depth, and resiliency, explains how adverse selection and inventory risk shape quote formation (grounded in Kyle 1985 and Glosten-Milgrom 1985), and catalogs order types from basic limit/market through hidden, pegged, and post-only variants. It also covers intraday seasonality regimes (opening, midday, closing) and why the same signal can mean different things at different times of day.
The anatomy of modern market data feeds
The data hierarchy runs from Level 1 top-of-book quotes through Level 2 price-level depth to Level 3 order-level messages — each feed reveals and conceals different aspects of market state. The section covers trades-and-quotes packages, price conventions (midprice vs last trade vs close), message protocols (FIX, binary multicast, crypto REST/WebSocket), and five sample datasets spanning the full hierarchy. Feed selection is a strategy design decision with direct implications for what can be observed, researched, and backtested.
From raw messages to the limit order book
LOB reconstruction is presented as a state-machine engineering problem, using NASDAQ TotalView-ITCH as a case study with 423 million messages per day. The section covers binary parsing at scale (26 minutes Python vs under 3 minutes Rust), the order registry and price-level view that must be maintained, and integrity invariants that must be enforced. Empirical findings reveal a 97.6% cancellation rate with 41% of cancellations within 500ms, and predictive patterns including top-of-book depth imbalance (mean correlation ~0.1 with returns) and intraday U-shaped volume.
The art of sampling
Bar sampling methods are compared for converting raw tick data into analytical time series: time bars, tick bars, volume bars, dollar bars, and information-driven bars (tick imbalance, volume imbalance, run bars). Lee-Ready trade classification achieves 96% accuracy versus 84% for the tick test alone. Dollar bars consistently achieve the best return normality (JB=84.7 vs 3,838 for time bars on NVDA) and are the recommended default for most ML workflows due to good statistical properties and natural price-level scaling.
Detecting price jumps in intraday returns
Microstructure errors are typically inconsistencies (out-of-sequence events, invalid book transitions, stale quotes) rather than extreme values, making invariant checks more appropriate than statistical outlier filters. The section prescribes layered validation: event sequencing first, then trade/quote sanity, then LOB reconstruction invariants, then cross-source reconciliation. Session handling is a correctness requirement where misclassifying weekends, holidays, or early closes creates artificial returns. A microstructure-aware backtest checklist covers executable prices, maker/taker fees, partial fills, auction handling, and venue scope.
Microstructure data quality and sessionization
Summary
01 Itch Parser
02 Itch Lob Reconstruction
03 Itch Lob Analysis
04 Itch Order Lifecycle Analysis
05 Itch Trading Activity
06 Itch Intraday Patterns
07 Itch Stylized Facts
08 Databento Lob Reconstruction
09 Databento Mbo Analysis
10 Iex Lob Reconstruction
11 Algoseek Taq Eda
12 Algoseek Taq Lob Reconstruction
13 Algoseek Minute Bars Eda
14 Itch Bar Sampling
15 Itch Lee Ready
16 Itch Information Bars
17 Databento Bar Sampling
18 Algoseek Jump Detection
3 primer topics providing foundational concepts for this chapter.
Limit Order Book Mechanics and Queue Priority
A limit order book is not just depth at prices. It is a queueing system whose execution outcomes depend on message order.
Microstructure Noise and Bid-Ask Bounce
Tick-level predictability often reflects how trades print through the spread, not an exploitable signal about the efficient price.
Trade Signing and the Lee-Ready Algorithm
Trade direction is usually missing from market data, but many microstructure features only make sense after you reconstruct it.
94 references cited in this chapter.
Lawrence R. Glosten and Paul R. Milgrom (1985) — Journal of Financial Economics · 6045 citations
This paper develops a theoretical model showing that a bid-ask spread arises purely from adverse selection in a competitive, risk-neutral specialist market, even with zero transaction costs, and explores its dynamic properties and implications for observed returns.
Albert S. Kyle (1985) — Econometrica · 9334 citations
The foundational paper establishing 'Kyle's Lambda' as a measure of market impact, proving that an informed monopolist trades gradually to disguise information, resulting in constant market depth and prices that follow Brownian motion.
Charles M. C. Lee and Mark J. Ready (1991) — The Journal of Finance · 3054 citations
This paper evaluates methods for classifying trades as buys or sells using intraday data, identifies problems with quote-based methods (quote timing, trades inside the spread), and proposes improvements using time-adjusted quotes and tick tests.
Ananth Madhavan et al. (1997) — The Review of Financial Studies · 957 citations
This paper introduces the 'MRR' structural model to decompose intraday price movements into information asymmetry (permanent) and transaction costs (temporary), explaining why spreads are U-shaped even as information uncertainty declines.
Ananth Madhavan (2002) — Financial Analysts Journal · 47 citations
A practitioner-focused synthesis of market microstructure showing how order flow, trading protocols, and transparency shape prices, liquidity, and transaction costs—and why these frictions matter for execution, factor models, and corporate/international finance.
Ioanid Rosu (2008)
A theoretical framework modeling the Limit Order Book (LOB) driven by waiting costs and competition rather than asymmetric information, explaining spread dynamics, price overshooting, and order clustering.
David Easley et al. (2010)
The 2010 Flash Crash was caused by toxic order flow driving market makers to withdraw liquidity; the VPIN metric successfully predicted this toxicity hours in advance, unlike the VIX.
Market Microstructure Theory
Maureen O'Hara (2011) — Wiley · 1750 citations
David Easley et al. (2012)
High Frequency Trading (HFT) is defined not by speed, but by operating in 'event-based time' (volume clocks), which allows for statistical normality and the exploitation of Low Frequency Traders (LFTs) who operate in chronological time.
Nikolaus Hautsch and Ruihong Huang (2012) — Journal of Economic Dynamics and Control · 168 citations
Limit orders exert a permanent price impact similar to trades, with the magnitude determined by order aggressiveness, size, and the existing state of the order book depth.
David Easley et al. (2012)
Introduces VPIN, a high-frequency flow toxicity metric updated in volume-time that predicts liquidity crashes and volatility without complex parameter estimation.
Martin D. Gould et al. (2013) — arXiv:1012.0349 [physics, q-fin] · 403 citations
A comprehensive survey of the mechanics, empirical statistical properties, and modelling approaches for Limit Order Books (LOBs), contrasting economic game theory with physics-based zero-intelligence models.
Rama Cont et al. (2014) — Journal of Financial Econometrics · 414 citations
Price changes are primarily driven by Order Flow Imbalance (OFI)—the net change in bid/ask queue sizes—rather than trade volume, following a linear relationship inversely scaled by market depth.
Craig W. Holden et al. (2014) — Foundations and Trends® in Finance · 121 citations
A comprehensive synthesis of empirical research on market liquidity, covering measurement techniques, the impact of market design and HFT, and the role of liquidity in corporate finance and asset pricing.
David Easley et al. (2015)
The paper demonstrates that while traditional Tick Rules correctly identify the aggressor side of trades, the Bulk Volume Classification (BVC) algorithm better identifies the presence of informed trading and future price movements in modern high-frequency markets.
Maureen O’Hara (2015) — Journal of Financial Economics · 502 citations
A seminal overview of how high-frequency trading has rendered traditional microstructure metrics (like trade direction and realized spreads) obsolete, requiring new approaches to data analysis and execution.
Kingsley Y L Fong et al. (2017) — Review of Finance
Using 19 years of global intraday trades/quotes as ground truth, the paper identifies which daily-data liquidity proxies best approximate spreads and price impact—finding Closing Percent Quoted Spread (CPQS) is best for percent-cost, and Amihud/impact-style proxies are best for Kyle’s lambda correlations but not its level.
Advances in Financial Machine Learning
Marcos Lopez de Prado (2018) — John Wiley & Sons · 106 citations
Zihao Zhang et al. (2019) — IEEE Transactions on Signal Processing · 256 citations
A hybrid deep learning architecture combining CNNs (for spatial LOB features) and LSTMs (for temporal dynamics) that achieves state-of-the-art price prediction and demonstrates transfer learning capabilities across different stocks.
SEC (2020)
A comprehensive regulatory report detailing the dominance of algorithmic trading in U.S. markets, concluding that while it improves liquidity and efficiency in normal conditions, it introduces operational risks and potential instability during stress events.
David Easley et al. (2021) — The Review of Financial Studies · 68 citations
Machine learning demonstrates that traditional microstructure measures (specifically VPIN and Roll) retain significant out-of-sample predictive power for volatility and liquidity in modern high-frequency markets, especially when using dollar-volume bars and cross-asset features.
Matteo Aquilina et al. (2021) · 122 citations
This paper quantifies the negative impact of high-frequency trading's "latency arbitrage" by analyzing failed trade and cancel messages, finding it's frequent, fast, accounts for ~20% of trading volume, and imposes a ~0.5 basis point tax on trading, totaling billions globally.
Simon Jurkatis (2022) — Journal of Financial Markets
A new 'Full-Information' trade classification algorithm uses order book volume changes to identify liquidity demanders, significantly outperforming Lee-Ready and interpolation methods in fast markets or low-precision data.
Martin Reck (2022) — The Journal of Portfolio Management
A detailed insider's look at how exchange operators engineer market microstructure (specifically Deutsche Börse's Xetra) to maximize liquidity through hybrid market models, volatility interruptions, and incentive programs.
Robert A. Schwartz et al. (2022) — The Journal of Portfolio Management · 1 citations
The paper argues that the decoupling of price discovery (on exchanges) and quantity discovery (in dark pools) leads to excessive intraday volatility and transaction costs, advocating for hybrid market models like call auctions to improve order integration.
Andy Novocin and Bruce Weber (2022) — The Journal of Portfolio Management · 2 citations
The paper argues that blockchain, Generative AI, and DAOs will disrupt financial markets by decoupling trading functions from centralized exchanges, similar to how Wikipedia disrupted Britannica.
Craig W. Holden et al. (2023) · 2 citations
This paper develops two new methods to accurately classify trade direction by accounting for message latency between exchanges and Securities Information Processors (SIPs), significantly improving accuracy and altering inferences for liquidity measures.
Yichi Zhang et al. (2025) · 1 citations
This paper introduces ClusterLOB, a method that clusters market events in limit order books (LOB) into directional, opportunistic, and market-making participants, enhancing trading strategy performance by leveraging order flow imbalances within these clusters.
Robert H. Battalio et al. (2025)
This paper reveals that the NYSE Trade and Quote (TAQ) data, widely used in finance research, suffers from a systematic look-ahead bias due to variable exchange-to-SIP latency and event clustering, leading to incorrect trade signing and understated spreads, and proposes a new methodology to correct these errors.
Fads, Martingales, and Market Efficiency
Bruce N. Lehmann (1990) — Quarterly Journal of Economics
This paper demonstrates that short-term stock returns exhibit strong mean reversion (reversal), allowing a strategy of buying last week's losers and shorting last week's winners to generate consistent arbitrage profits, even after accounting for bid-ask spreads and transaction costs.
Joel Hasbrouck (1991) — The Journal of Finance · 1673 citations
This seminal paper introduces the use of Vector Autoregression (VAR) to separate the permanent price impact of trades (information) from transient effects (liquidity/inventory), finding that price impact is concave and higher when spreads are wide.
Maureen O'Hara (1999) — Financial Management
This paper argues that market microstructure, despite extensive research, has not yet clearly established its broader economic impact on asset pricing, corporate finance, or overall welfare, and proposes a research agenda to bridge this gap.
Tarun Chordia et al. (2000) — Journal of Financial Economics · 1662 citations
Liquidity is not merely an idiosyncratic asset attribute but contains a significant systematic component that co-moves across the market, driven by inventory risks and asymmetric information.
Martin D. D. Evans and Richard K. Lyons (2002) — Journal of Political Economy
This seminal paper introduces order flow as the primary driver of exchange rates, demonstrating that aggregate interdealer trades explain ~60% of daily price changes compared to <10% for traditional macroeconomic models.
Yakov Amihud (2002) — Journal of Financial Markets · 5780 citations
Introduces the 'Amihud Ratio' (ILLIQ)—a measure of price impact using daily data—and proves that illiquidity is a priced risk factor that predicts stock returns both cross-sectionally and over time.
Yuriy Nevmyvaka et al. (2006) — Association for Computing Machinery · 291 citations
This paper shows—using 1.5 years of millisecond NASDAQ limit-order-book data—that reinforcement learning can learn execution policies that reduce implementation shortfall by up to ~50% versus optimized “submit-and-leave” baselines.
Jean-Philippe Bouchaud et al. (2008)
This paper reviews theoretical and empirical work on how markets slowly incorporate fluctuations in supply and demand, highlighting the role of long-memory order flow and its impact on price formation, liquidity, and market impact.
James Angel et al. (2010)
This paper analyzes the evolution of equity trading due to technological advancements and regulatory changes, highlighting improved market quality and identifying areas for further regulatory consideration.
David Easley et al. (2012)
A rebuttal clarifying that VPIN successfully predicts liquidity crashes like the 2010 Flash Crash only when implemented using volume clocks and Bulk Volume Classification rather than time bars.
Modeling high-frequency limit order book dynamics with support vector machines
Alec N Kercheval and Yuan Zhang (2013) · 18 citations
This paper demonstrates how to use Support Vector Machines (SVMs) with engineered Limit Order Book (LOB) features—specifically order flow intensity and derivatives—to predict short-term mid-price moves and spread crossings with >80% accuracy.
Michael R. King et al. (2013)
A comprehensive survey and empirical analysis demonstrating that order flow, driven by private information and liquidity constraints, is the primary driver of short-term exchange rates, superseding traditional macroeconomic models.
Damian Eduardo Taranto et al. (2016)
This paper introduces the Mixture Transition Distribution (MTD) model to capture the joint dynamics of order flow and price changes, modeling the sequence of signed events (buy/sell and price changing/non-changing) as a discrete Markov process. The authors demonstrate that the MTD model, particularly a weakly constrained version estimated via GMM, effectively captures the high-frequency dynamics of signed events and outperforms simpler models in out-of-sample prediction.
Bernt Arne Ødegaard (2017)
These lecture notes survey the main empirical liquidity (and transaction cost) estimators—spreads, price-impact measures, non-trading-based measures, and estimators from autocovariances and high/low ranges—highlighting what each captures and the data requirements.
Justin Sirignano and Rama Cont (2019) — Quantitative Finance
A deep learning model trained on pooled high-frequency data from hundreds of stocks outperforms asset-specific models in predicting price direction, demonstrating that price formation is governed by a universal, stationary mechanism.
Feature Engineering for Mid-Price Prediction With Deep Learning
Adamantios Ntakaris et al. (2019) — IEEE Access
This paper demonstrates that handcrafted econometric features (volatility, noise measures) significantly outperform automated feature extraction (LSTM Autoencoders) for high-frequency mid-price prediction using deep learning.
Ziwen Ye and Ionuţ Florescu (2019) — High Frequency
Introduces Activity-Weighted Spread (AWS) and Activity-Weighted Return (AWR), two Level I order book metrics that capture microstructure dynamics and successfully detect market anomalies like the 2010 Flash Crash minutes before they occur.
Justin A. Sirignano (2019) — Quantitative Finance
A novel 'spatial' neural network architecture exploits the local structure of limit order books to model joint bid-ask price distributions, significantly outperforming standard deep learning and logistic regression in accuracy and computational efficiency.
Darrell Duffie (2020)
The paper analyzes the March 2020 Treasury market liquidity collapse, attributing it to regulatory constraints on dealer balance sheets, and proposes mandatory central clearing to enable netting and restore market elasticity.
Bruce Knuteson (2020) — SSRN Electronic Journal
Knuteson argues that the long-run pattern of positive overnight and negative intraday equity index returns (internationally, and in some individual US stocks) is so extreme and persistent that it is more consistent with systematic manipulation than with benign explanations like earnings timing or market hours.
Petter N. Kolm et al. (2021)
Deep learning models trained on stationary Order Flow (OF) inputs significantly outperform those trained on raw Limit Order Book (LOB) states, with predictive power peaking at a horizon of two average price changes.
Guido Baltussen et al. (2021) — Journal of Financial Economics
Market makers hedging short gamma positions (options and leveraged ETFs) are forced to trade in the direction of the day's trend just before the close, creating a predictable 'intraday momentum' effect in the last 30 minutes.
Jonathan Brogaard et al. (2022) — The Review of Financial Studies
A structural VAR decomposition of daily stock returns reveals that 31% of variance is noise, while public firm-specific information has overtaken private trading information as the dominant driver of prices since the mid-1990s.
Joseph A. Cerniglia and Frank J. Fabozzi (2022) — The Journal of Portfolio Management
A practitioner-oriented overview of how modern market structure, transaction-cost measurement, and optimal execution models determine whether an investment strategy’s paper alpha survives real-world implementation.
Emilio Said (2022) · 2 citations
This paper proposes a coarse-grained theoretical model proving that the 'square-root law' of market impact arises from a specific supply-demand equilibrium where the ratio of average impact to peak impact stabilizes at 2/3.
Mikołaj Bińkowski and Charles-Albert Lehalle (2022) — The Journal of Portfolio Management
Using 5-minute data for 300 global stocks (2011–2016), the paper shows how much of intraday liquidity (spread, turnover, volatility, top-of-book size) is predictable from its own recent past, and how predictability/memory vary by tick size, market cap, and country—directly informing execution/TCA models.
Thomas Ernst and Chester S. Spatt (2022)
While PFOF in equities is benign due to tight spreads and price improvement, PFOF in options creates a major conflict of interest where brokers are incentivized to push clients toward high-fee, wide-spread option products where execution quality is demonstrably worse.
Ananth Madhavan et al. (2022) — The Journal of Portfolio Management
Using 18,871 Russell index reconstitution events (2016–2021), the paper argues that the true, actionable “liquidity provision cost” of index rebalancing—measured as post-effective-date return reversal—is small on average and can even be negative once you net flows across all indexes.
Jean-Philippe Bouchaud (2022) — Quantitative Finance · 16 citations
This paper reconciles the macro-level Inelastic Market Hypothesis (buying $1 raises market cap by $M) with micro-level Latent Liquidity Theory, proving that long-term price impact is linear and driven by order flow rather than fundamental news.
Ilia Zaznov et al. (2022) — Mathematics
A critical review and reproduction study demonstrating that while deep learning models achieve >80% accuracy on LOB benchmarks, they fail to generate profit when realistic bid-ask spreads and transaction costs are applied.
Brad M. Barber et al. (2023)
This paper evaluates and modifies the Boehmer et al. (2021) algorithm for identifying retail trades using subpenny price improvements, finding that signing trades using quote midpoints significantly improves accuracy.
Yuki Sato and Kiyoshi Kanazawa (2024) · 11 citations
Using a complete, account-level order-book dataset for the Tokyo Stock Exchange (2012–2019), the paper finds the market impact exponent is statistically indistinguishable from 1/2 for essentially all liquid stocks and for individual active traders, and shows key non-universal theories fail empirical tests.
Hunter Ng et al. (2025)
This paper provides the first empirical evidence of cross-market price discovery in modern prediction markets, finding significant arbitrage opportunities and demonstrating that Polymarket leads Kalshi in information aggregation, driven by liquidity and informed 'whale' trades.
Yadh Hafsi and Edoardo Vittori (2025) · 5 citations
The paper trains a high-frequency (1-second) reinforcement learning agent in the ABIDES multi-agent limit-order-book simulator to learn an optimal execution policy that beats standard schedules like TWAP while controlling market impact.
Limit Order Flow, Market Impact and Optimal Order Sizes: Evidence from NASDAQ TotalView-ITCH Data
Nikolaus Hautsch
This paper quantifies the permanent price impact of limit orders on NASDAQ, demonstrating that limit orders (not just market orders) move prices, and provides a method to calculate optimal order sizes to minimize this impact.
Understanding the Market for U.S. Equity Market Data
Charles M Jones
An economic analysis arguing that U.S. equity market data fees are stable, competitively priced, and constitute a negligible fraction of total industry costs compared to commissions and third-party vendor fees.
Ananth Madhavan — Journal of Financial Markets
A foundational survey establishing the 'canonical model' of market microstructure, distinguishing between inventory control and asymmetric information as drivers of price formation, liquidity, and trading costs.
Robert Almgren and Neil Chriss (2001) — The Journal of Risk · 564 citations
The paper derives closed-form optimal trade schedules that minimize a chosen tradeoff between market-impact costs and volatility risk, producing an “efficient frontier” of execution strategies and introducing liquidity-adjusted VaR (L‑VaR).
Marco Avellaneda and Sasha Stoikov (2008) — Quantitative Finance · 522 citations
A seminal mathematical framework for market making that derives optimal bid-ask quotes as a function of inventory levels and market volatility to maximize risk-adjusted returns.
Zoltan Eisler et al. (2010) · 198 citations
This paper extends price impact models to include limit orders and cancellations, demonstrating that while large-tick stocks follow a constant impact model, small-tick stocks require a history-dependent model due to fluctuating order book gaps.
Bence Toth et al. (2011) — Physical Review X · 243 citations
The paper provides a theoretical and empirical basis for the 'square-root law' of price impact, arguing that markets self-organize into a critical state where liquidity vanishes linearly near the current price.
Anna A. Obizhaeva and Jiang Wang (2013) — Journal of Financial Markets
This paper introduces a dynamic Limit Order Book (LOB) model where liquidity replenishes over time ('resilience'), proving that optimal execution requires a mix of large discrete trades at the start and end, bridged by continuous trading.
Nina Karnaukh et al. (2015) · 164 citations
Validates that daily low-frequency data can accurately proxy FX transaction costs and demonstrates that liquidity evaporates globally when funding constraints (TED spread) and volatility (VIX) rise.
Steve Y. Yang et al. (2015) — Quantitative Finance · 42 citations
This paper introduces a novel approach to identify algorithmic trading strategies by inferring traders' reward functions using inverse reinforcement learning (IRL) with Gaussian processes, demonstrating improved classification accuracy compared to traditional summary statistics methods.
John Lehoczky and Mark Schervish (2018) — Annual Review of Statistics and Its Application
A historically organized survey of how equity-market data, statistical models, and trading strategies evolved from pre-CRSP fundamentals/technical analysis to statistical arbitrage, machine learning, and today’s electronic limit-order-book/HFT markets.
Damian Eduardo Taranto et al. (2018) — Quantitative Finance · 21 citations
Standard market impact models fail for large-tick stocks; separating trades into 'price-changing' and 'non-price-changing' events corrects this by capturing the mean-reverting nature of liquidity provision.
Bruce Knuteson (2019)
This paper argues that a simple, repeatable open/close trading pattern can mechanically “push” equity index prices over long horizons via market impact, potentially explaining the decades-long pattern of positive overnight and negative intraday index returns.
Jacobo Roa-Vicens et al. (2019) — arXiv:1906.04813 [cs, q-fin, stat]
The paper tests whether inverse reinforcement learning can recover an expert trader’s (possibly non-linear) objective from limit order book demonstrations, finding that Bayesian neural network and Gaussian-process IRL succeed on non-linear rewards while linear MaxEnt IRL does not.
Markus Pelger (2019) — Journal of Econometrics
This paper develops a statistical theory and practical estimators for large-dimensional factor models using high-frequency financial data, enabling the separation of continuous and jump risk factors and providing a robust method for determining the number of factors.
Igor Makarov and Antoinette Schoar (2020) — Journal of Financial Economics
This paper documents massive, persistent arbitrage spreads in cryptocurrency markets driven by capital controls and identifies that a single common order flow factor explains 80% of Bitcoin returns.
David Byrd et al. (2020) — ACM · 83 citations
ABIDES is an open-source agent-based market simulation environment designed to facilitate research in market applications by simulating the interactions of thousands of trading agents with an exchange, incorporating realistic network latency and message protocols.
Xavier Gabaix and Ralph S. J. Koijen (2021) · 181 citations
The aggregate stock market is highly inelastic, meaning a $1 capital inflow increases market valuation by approximately $5, suggesting that flows rather than fundamental news drive the majority of market volatility.
Alfred Lehar and Christine A. Parlour (2021) · 83 citations
This paper analyzes Uniswap, a prominent decentralized exchange using automated market making (AMM), characterizing equilibrium in liquidity pools, providing empirical evidence of their stability, and comparing it to centralized exchanges like Binance.
Zihao Zhang and Stefan Zohren (2021)
This paper introduces Sequence-to-Sequence and Attention-based deep learning models for multi-step Limit Order Book forecasting, demonstrating superior long-horizon accuracy and achieving ~6.5x faster training speeds using Graphcore IPUs compared to GPUs.
Ryan Donnelly (2022) — Applied Mathematical Finance · 15 citations
A comprehensive review of optimal execution models, ranging from the foundational Almgren-Chriss framework to modern extensions involving stochastic volatility, transient price impact, and alternative objectives like VWAP targeting.
Christopher Schwarz et al. (2022) · 7 citations
This paper uses a controlled experiment to show that execution prices for retail equity trades vary significantly across brokers, even with zero commissions, due to systematic pricing differences by off-exchange wholesalers, and that payment for order flow (PFOF) does not explain this variation.
Lin William Cong et al. (2023) — Management Science
A systematic forensic analysis of cryptocurrency exchanges revealing that over 70% of reported volume on unregulated platforms is wash trading, detected via statistical anomalies in trade size distributions.
Arash Aloosh and Jiasun Li (2024) — Management Science
Using leaked Mt. Gox data with user IDs, this paper provides ground-truth validation for statistical methods used to detect wash trading, finding that Benford's Law and size clustering are effective while Power Law tail analysis is unreliable.
Antonio Briola et al. (2024)
This paper demonstrates that the profitability of Deep Learning in LOB forecasting is strictly dependent on stock tick-size regimes and introduces a 'transaction probability' metric to replace misleading accuracy scores.
Songrun He et al. (2024)
This paper derives the theoretical no-arbitrage price for cryptocurrency perpetual futures and demonstrates that a threshold-based trading strategy exploiting deviations from this price yields Sharpe ratios between 1.8 and 3.5.
Yang Li et al. (2025)
FlowHFT is a novel imitation learning framework using a flow matching policy that learns from multiple expert HFT models to adaptively trade across diverse market conditions, consistently outperforming individual expert strategies.
Paul Glasserman et al. (2025)
Using 2.4M Reuters articles (1996–2022), the paper shows that persistent firm-level news-topic exposures and opposite intraday vs overnight price responses to the same topics explain a large share of the long-run U.S. “overnight drift” (overnight gains vs flat/negative intraday returns) and help forecast which stocks will do best overnight and worst intraday.
James J. Choi et al. (2025) — The Journal of Finance and Data Science
This paper demonstrates that stocks with higher predicted information asymmetry, proxied by past institutional trading aggressiveness, exhibit significantly higher expected returns, suggesting a compensation for uninformed investors.
Alejandro Lopez-Lira (2025)
The paper builds an open-source, realistic limit-order-book stock-market simulator where LLMs trade as heterogeneous agents, showing they can follow instructed strategies and generate market phenomena like price discovery and bubbles—useful for testing financial theories and systemic-risk scenarios.
Machine Learning for Market Microstructure and High Frequency Trading
Michael Kearns and Yuriy Nevmyvaka · 64 citations
This paper demonstrates how to apply Machine Learning (RL and censored estimation) to three core HFT problems: optimized trade execution, directional price prediction, and dark pool order routing.