Tracking High-Frequency Trading Activity in China and Its Effects on Market Quality

Li Guo; Frank Weikai Li; Ding Qi; Lijian Sun

AI summary · arXiv: 2610.33400v1 · q-fin.TR

Tracking High-Frequency Trading Activity in China and Its Effects on Market Quality

Li Guo, Frank Weikai Li, Ding Qi, Lijian Sun

Original paper Submitted to arXiv on 24 Oct. 2026 License: NONEXCLUSIVE-DISTRIB 1.0 Also in: fr

How to read this summary? Illustrated guide

The essentials

Exact-match limit orders identify latency-sensitive trading in Shenzhen; faster information delivery increases their prevalence, short-run price variation and adverse selection, while modestly reducing pricing error.

Question

Can precise placement of limit orders at current best quotes identify latency-sensitive high-frequency trading without trader identifiers? How does improved information timeliness affect this activity and market quality in China?

Method

The authors reconstruct Shenzhen order books and analyze approximately 0.1% systematically sampled orders from July 2023 through October 2025, retaining 42,506,216 limit orders across 2,760 stocks and 566 trading days after excluding the smallest 30% of stocks. They validate active exact matching against response speed, cancellation behavior, market conditions and direction-adjusted future midpoint returns. A difference-in-differences design uses 24 trading days of opening congestion-driven channel latency variation, comparing fast and slow groups around the fast group’s crossing below three seconds and censoring observations at the slow group’s crossing.

Findings

Result Active exact matching is associated with a 19.2-percentage-point higher probability of responding within 100 milliseconds and a 5.4-percentage-point higher probability of cancellation within 100 milliseconds among canceled orders.
See the passage in the paper
In the fully controlled specification of Table 2, active and passive quote matching have the two largest positive coefficients among the included order-characteristic indicators. They are associated with 19.2 and 17.7 percentage-point higher probabilities of responding within 100 milliseconds, respectively.
Result Active exact-match orders predict direction-adjusted midpoint returns at every tested horizon from 100 milliseconds to ten minutes.
See the passage in the paper
Figure 3 and Table 6 show that the active exact-match coefficient is positive at all nine horizons: 1.671 bp at 100ms, 2.268 bp at 10s and 3.747 bp at 600s.
Result Following the latency improvement, active exact-match share increases by 0.505 percentage points, providing the clearest activity response.
See the passage in the paper
Table 9 reports the static activity estimates: active matching increases by 0.447 orders per stock-second ($t=6.95$), active share by 0.505 percentage points ($t=3.03$), and passive share by 0.515 percentage points ($t=3.07$), with two-way clustering.
Result The latency improvement increases intrasecond midpoint range by 1.029 basis points and wide-spread incidence by 2.964 percentage points.
See the passage in the paper
Range increases by 1.029 basis points ($t=10.36$), with a pre-event joint $p=0.082$. WideSpread increases by 2.964 percentage points ($p<0.001$), with a pre-event joint $p=0.720$.
Result One-second adverse-selection exposure rises and realized spreads fall, while effective spreads show no statistically significant change.
See the passage in the paper
ES changes by $-0.049$ bp ($p=0.599$), AS1s increases by 0.648 bp, and RS1s decreases by 1.345 bp (both $p<0.001$).

Limitations

Causal attribution to HFT requires parallel trends, no concurrent channel-specific shocks and an exclusion restriction ruling out other effects of latency on market quality.
See the passage in the paper
A causal interpretation requires parallel trends and the absence of concurrent channel-specific shocks. Interpreting the market-quality responses as effects of HFT activity additionally requires an exclusion restriction: within the narrow censored event window, cross-channel latency affects market quality through the induced change in latency-sensitive trading, with no separate effect through other channels.
Pre-event diagnostics weaken the evidence for active counts and passive matching relative to active shares.
See the passage in the paper
All three static activity estimates are significantly positive, but their pre-event diagnostics differ. We place the greatest weight on active share as evidence that the latency shock increases proxy-measured HFT activity.
The pricing-error result is complementary evidence because its model-based construction restricts coverage to eligible transaction-seconds.
See the passage in the paper
Because this model-based measure is available only for transaction-seconds satisfying the estimation and warm-up requirements, we report it as complementary evidence.

Why it matters

The authors offer precise contemporaneous price placement as a behavioral HFT proxy that complements speed- and message-based measures without requiring trader identifiers. They argue that China’s trading constraints help channel speed advantages toward directional liquidity taking, showing why market design matters for the market-quality effects of faster information.

Written by AI from the paper. Passages are checked word for word against the text we analysed; the rest may contain errors.

AI-generated keywords: High-frequency trading; Order-book reconstruction; Information latency; Market quality; Adverse selection

Read further

Comprehensive Summary
Layman's Summary
Blog article
Original abstract

This paper examines how to identify high-frequency trading (HFT) without proprietary trader labels and how faster access to market information affects market quality in China. Its central premise is that precise order pricing reveals timely information processing. An “active exact-match” order is a limit order priced at the current best opposite-side quote: a buy at the best ask or a sell at the best bid. Because quotes often change within fractions of a second, consistently targeting these prices should be easier for low-latency traders. The authors therefore use this behavior as a proxy for latency-sensitive trading, not as definitive evidence identifying individual HFT firms. The authors reconstruct Shenzhen Stock Exchange order books using detailed records of order submissions, executions, and cancellations. The baseline period spans July 2023 through October 2025 and excludes the smallest 30% of stocks by previous-day market capitalization. Reconstruction uses the complete message stream, while the order-level analysis draws on an approximately 0.1% systematic sample comprising 42.5 million limit orders across 2,760 stocks and 566 trading days. Passive exact matching—posting at the best same-side quote—serves as an auxiliary measure. Several tests support the active exact-match proxy. After controls, these orders are 19.2 percentage points more likely to arrive within 100 milliseconds of the latest quote update. Among canceled orders, they are 5.4 percentage points more likely to be canceled within 100 milliseconds. They also cluster when displayed depth favors their direction and spreads are narrow. Active exact matching predicts subsequent direction-adjusted midpoint returns beyond the effects of order direction and aggressiveness. Its fully controlled ten-second coefficient is 2.157 basis points. In a common sample across horizons, coefficients range from 1.671 basis points at 100 milliseconds to 3.747 basis points at ten minutes. Order size matters substantially: ten-second coefficients are 2.711 basis points for orders smaller than opposite-side displayed depth, 1.818 basis points for orders equal to that depth, and −4.408 basis points for larger orders. To investigate market-quality effects, the paper exploits congestion-driven differences in opening message-delivery latency across exchange trading channels. A separate colocation feed supports a difference-in-differences analysis covering 24 trading days in October–November 2025. The authors compare fast-channel and slow-channel stocks around the fast group’s first crossing below three seconds of latency, ending the observation window before the slow group crosses that threshold. Stock-by-date and date-by-second fixed effects absorb stock-day differences and common second-level shocks. Following the latency improvement, the active exact-match share increases by 0.505 percentage points. The intrasecond midpoint range rises by 1.029 basis points, and the probability that a second contains a spread wider than one tick increases by 2.964 percentage points. Average quoted and effective spreads show no statistically significant change, while displayed depth declines weakly. One-second adverse-selection exposure rises by 0.648 basis points, realized spreads fall by 1.345 basis points, and model-based temporary pricing error declines modestly, by 0.0449 basis points. The authors conclude that faster information access in this setting primarily supports directional liquidity taking, increasing short-horizon risks for liquidity providers while modestly improving pricing accuracy. China’s T+1 resale restriction, absence of maker-taker rebates, and constraints on rapid order revision may help explain these findings. Important limitations remain. Exact matching is an imperfect behavioral proxy, and predictive returns do not represent net trading profits. Causal interpretation depends on parallel trends, the absence of concurrent channel-specific shocks, and an exclusion restriction linking latency effects to induced trading. Some outcomes fail pre-event diagnostics. Moreover, the experiment covers only brief opening windows, and pricing-error estimates have selective coverage. Their reported standard errors also omit uncertainty from historical parameter estimation.

Tracking High-Frequency Trading in China: What Faster Information Changes

High-frequency trading is difficult to study when researchers cannot identify the firms behind individual orders. How can they distinguish speed-sensitive activity from ordinary trading? And when information arrives faster, does the market become better for everyone—or simply more challenging for traders supplying liquidity?

In “Tracking High-Frequency Trading Activity in China and Its Effects on Market Quality,” Li Guo, Frank Weikai Li, Ding Qi, and Lijian Sun tackle both questions. Their approach starts with a simple idea: consistently pricing orders against rapidly changing quotes can reveal timely information processing. Their findings suggest that faster access can modestly improve pricing accuracy while increasing short-term risks for liquidity providers.

A Behavioral Fingerprint of Speed

The paper’s central measure is an “active exact-match” order: a limit order priced precisely at the best available quote on the opposite side of the market. A buyer targets the current best ask; a seller targets the current best bid.

That sounds straightforward, but quotes can change within fractions of a second. Repeatedly hitting these exact prices requires up-to-date information and timely reactions. The authors therefore treat active exact matching as a proxy for latency-sensitive trading—not as definitive identification of individual HFT firms. Posting at the best same-side quote, called passive exact matching, provides an auxiliary measure.

Building and Testing the Measure

The researchers reconstruct Shenzhen Stock Exchange order books from the complete stream of submissions, executions, and cancellations. Their baseline period runs from July 2023 through October 2025, excluding the smallest 30% of stocks by previous-day market capitalization. The order-level analysis uses an approximately 0.1% systematic sample: 42.5 million limit orders across 2,760 stocks and 566 trading days.

Several results support the proxy. After controls, active exact-match orders are 19.2 percentage points more likely to arrive within 100 milliseconds of the latest quote update. Among canceled orders, they are 5.4 percentage points more likely to be canceled within 100 milliseconds. They also cluster when displayed depth favors their direction and spreads are narrow.

Precise Pricing Also Predicts Price Direction

Active exact matching predicts subsequent midpoint movements in the order’s direction, beyond the effects of direction and aggressiveness alone. The fully controlled ten-second coefficient is 2.157 basis points. In a common sample across horizons, coefficients range from 1.671 basis points at 100 milliseconds to 3.747 basis points at ten minutes.

Order size changes the picture substantially. At ten seconds, the coefficients are:

2.711 basis points for orders smaller than opposite-side displayed depth.
1.818 basis points for orders equal to that depth.
−4.408 basis points for orders larger than that depth.

This contrast cautions against treating all exact-match orders alike. It also does not establish profitability: predictive midpoint returns are not net trading profits and do not account for every cost of implementing a strategy.

What Happens When Information Arrives Faster?

To investigate market quality, the authors exploit congestion-driven differences in opening message-delivery latency across exchange trading channels. A separate colocation feed supports a difference-in-differences analysis over 24 trading days in October–November 2025.

The comparison centers on the fast-channel group’s first crossing below three seconds of latency. The observation window ends before the slow-channel group crosses the same threshold. Stock-by-date and date-by-second fixed effects absorb stock-day differences and shocks shared within each second.

A Mixed Market-Quality Scorecard

Following the latency improvement, the active exact-match share rises by 0.505 percentage points. Other outcomes show why faster information is not an unqualified improvement:

The intrasecond midpoint range increases by 1.029 basis points.
The probability that a second contains a spread wider than one tick rises by 2.964 percentage points.
Average quoted and effective spreads show no statistically significant change, while displayed depth declines weakly.
One-second adverse-selection exposure rises by 0.648 basis points, and realized spreads fall by 1.345 basis points.
Model-based temporary pricing error declines modestly, by 0.0449 basis points.

Together, these findings suggest that faster access primarily supports directional liquidity taking in this setting. Prices become slightly more accurate, but liquidity providers face greater short-horizon exposure to trading against better-informed counterparties.

Why Context and Caveats Matter

China’s market structure may help explain this balance. Its T+1 resale restriction, absence of maker-taker rebates, and constraints on rapid order revision can shape how speed is used. Results from these opening windows should not automatically be generalized to other markets or the full trading day.

The causal interpretation also requires caution. It depends on parallel trends, no concurrent channel-specific shocks, and an exclusion restriction linking latency effects to induced trading. Some outcomes fail pre-event diagnostics. Pricing-error estimates have selective coverage, and their reported standard errors omit uncertainty from historical parameter estimation.

The paper’s contribution is therefore both methodological and practical: precise order pricing offers a useful, imperfect window into latency-sensitive activity. Faster information can improve price discovery without delivering uniformly better liquidity—and understanding that trade-off requires looking beyond average spreads.

Created on 02 Nov. 2026

Was this synthesis useful?

Score: 0

Ask questions about this paper to our AI assistant

How to chat with this paper? Illustrated guide

Compare with the PDF

You can also chat with multiple papers at once here.

AI assistant instructions?

Tools for this paper

Explain it differently 3 free rewrites left today

Save to a collection Group papers in collections to compare them, chat with them and export them. Create a free account Log in

Export BibTeX · RIS (Zotero, Mendeley, EndNote) · Obsidian (Markdown) · Notion (Markdown)

Advanced tools Premium

For researchers who follow a field closely: citation alerts, journal-club slides, version comparison and Zotero export. Discover Premium

Zotero

Citation alerts Citation alerts guide

Who works on it? The people and institutions who publish on its topic

Weekly letter Presented in "Economics and finance this week" of 2 November 2026. Economics and finance this week: When the measuring tool changes the answer

Journal club Premium+ Slides to present it Propose for the journal club

Versions What changed between versions

Talk about it in public Premium+ Press kit: release, questions, hype check Press kit guide

Papers around this one

See them on a map

Ranked by how close their title and abstract are to this paper’s. Select some to read them together.

Closest papers

Close Same question
The Importance of Low Latency to Order Book Imbalance Trading Strategies
David Byrd, Sruthi Palaparthi, Maria Hybinette et al. · 2020 · Trading and Market Microstructure · preprint, not peer reviewed · cited 4 times · AI summary available

The study tests how low-latency order-book access affects order-imbalance strategy profitability, while this paper examines latency’s effects on HFT activity and market quality.

Abstract
There is a pervasive assumption that low latency access to an exchange is a key factor in the profitability of many high-frequency trading strategies. This belief is evidenced by the "arms race" undertaken by certain financial firms to co-locate with exchange servers. To the best of our knowledge, our study is the first to validate and quantify this assumption in a continuous double auction market with a single exchange similar to the New York Stock Exchange. It is not feasible to conduct this exploration with historical data in which trader identity and location are not reported. Accordingly, we investigate the relationship between latency of access to order book information and profitabili…
Close Broader context
Analysis of trade packages in Chinese stock market
Fei Ren, Wei-Xing Zhou · 2011 · Quantitative Finance 13 (7), 1071-1089 (2013) · cited 6 times · AI summary available

The study analyzes Chinese stock-market trade packages and differences between institutional and individual investors, rather than latency-sensitive order placement.

Abstract
This paper conducts an empirically study on the trade package composed of a sequence of consecutive purchases or sales of 23 stocks in Chinese stock market. We investigate the probability distributions of the execution time, the number of trades and the total trading volume of trade packages, and analyze the possible scaling relations between them. Quantitative differences are observed between the institutional and individual investors. The trading profile of trade packages is investigated to reveal the preference of large trades on trading volumes and transaction time of the day, and the different profiles of two types of investors imply that institutions may be more informed than individua…
Close Broader context
Enhanced fill probability estimates in institutional algorithmic bond trading using statistical learning algorithms with quantum computers
Axel Ciceri, Austin Cottrell, Joshua Freeland et al. · 2025 · Quantum Physics · cited 8 times · Not summarized yet

The study applies statistical learning to estimate fill probabilities for institutional algorithmic orders in corporate bonds, not HFT in Chinese equities.

Abstract
The estimation of fill probabilities for trade orders represents a key ingredient in the optimization of algorithmic trading strategies. It is bound by the complex dynamics of financial markets with inherent uncertainties, and the limitations of models aiming to learn from multivariate financial time series that often exhibit stochastic properties with hidden temporal patterns. In this paper, we focus on algorithmic responses to trade inquiries in the corporate bond market and investigate fill probability estimation errors of common machine learning models when given real production-scale intraday trade event data, transformed by a quantum algorithm running on IBM Heron processors, as well a…
Close
Portfolio liquidation under transient price impact -- theoretical solution and implementation with 100 NASDAQ stocks
Ying Chen, Ulrich Horst, Hoang Hai Tran · 2019 · Mathematical Finance · cited 12 times · AI summary available

Abstract
We derive an explicit solution for deterministic market impact parameters in the Graewe and Horst (2017) portfolio liquidation model. The model allows to combine various forms of market impact, namely instantaneous, permanent and temporary. We show that the solutions to the two benchmark models of Almgren and Chris (2001) and of Obizhaeva and Wang (2013) are obtained as special cases. We relate the different forms of market impact to the microstructure of limit order book markets and show how the impact parameters can be estimated from public market data. We investigate the numerical performance of the derived optimal trading strategy based on high frequency limit order books of 100 NASDAQ s…
Close
Market Making under a Weakly Consistent Limit Order Book Model
Baron Law, Frederi Viens · 2019 · Trading and Market Microstructure · cited 14 times · AI summary available

Abstract
We develop a new market-making model, from the ground up, which is tailored towards high-frequency trading under a limit order book (LOB), based on the well-known classification of order types in market microstructure. Our flexible framework allows arbitrary order volume, price jump, and bid-ask spread distributions as well as the use of market orders. It also honors the consistency of price movements upon arrivals of different order types. For example, it is apparent that prices should never go down on buy market orders. In addition, it respects the price-time priority of LOB. In contrast to the approach of regular control on diffusion as in the classical Avellaneda and Stoikov [1] market-m…
Related
Learning the Spoofability of Limit Order Books With Interpretable Probabilistic Neural Networks
Timothée Fabre, Damien Challet · 2025 · Trading and Market Microstructure · cited 3 times · AI summary available

Abstract
This paper investigates real-time detection of spoofing activity in limit order books, focusing on cryptocurrency centralized exchanges. We first introduce novel order flow variables based on multi-scale Hawkes processes that account both for the size and placement distance from current best prices of new limit orders. Using a Level-3 data set, we train a neural network model to predict the conditional probability distribution of mid price movements based on these features. Our empirical analysis highlights the critical role of the posting distance of limit orders in the price formation process, showing that spoofing detection models that do not take the posting distance into account are ina…
Related
Exploring the Advantages of Transformers for High-Frequency Trading
Fazl Barez, Paul Bilokon, Arthur Gervais et al. · 2023 · Statistical Finance · cited 8 times · AI summary available

Abstract
This paper explores the novel deep learning Transformers architectures for high-frequency Bitcoin-USDT log-return forecasting and compares them to the traditional Long Short-Term Memory models. A hybrid Transformer model, called \textbf{HFformer}, is then introduced for time series forecasting which incorporates a Transformer encoder, linear decoder, spiking activations, and quantile loss function, and does not use position encoding. Furthermore, possible high-frequency trading strategies for use with the HFformer model are discussed, including trade sizing, trading signal aggregation, and minimal trading threshold. Ultimately, the performance of the HFformer and Long Short-Term Memory model…
Related
Profitability of simple stationary technical trading rules with high-frequency data of Chinese Index Futures
Jing-Chao Chen, Yu Zhou, Xi Wang · 2017 · Statistical Finance · cited 14 times · AI summary available

Abstract
Technical trading rules have been widely used by practitioners in financial markets for a long time. The profitability remains controversial and few consider the stationarity of technical indicators used in trading rules. We convert MA, KDJ and Bollinger bands into stationary processes and investigate the profitability of these trading rules by using 3 high-frequency data(15s,30s and 60s) of CSI300 Stock Index Futures from January 4th 2012 to December 31st 2016. Several performance and risk measures are adopted to assess the practical value of all trading rules directly while ADF-test is used to verify the stationarity and SPA test to check whether trading rules perform well due to intrinsic…

Go further

How are these papers found?

We compare the title and abstract of this paper with those of every paper summarized on this site, and with the papers Semantic Scholar recommends, those it cites and those citing it (for a much-cited paper, its most recent citations and, from OpenAlex, the most cited ones). Among the prior work and the follow-ups, the papers that use each other in depth, a method it uses and the much-cited follow-ups come a little higher. The levels (very close, close, related) come from the similarity of the texts, not from a judgment of quality; * marks a paper without an abstract, compared by its title. The sentence under a paper is written by an AI from the two abstracts: check it in the papers. A paper marked not relevant by three readers leaves the list.

Updated on 2 November 2026.

Disclaimer: The AI-based summarization tool and virtual assistant provided on this website may not always provide accurate and complete summaries or responses. We encourage you to carefully review and evaluate the generated content to ensure its quality and relevance to your needs.

Tracking High-Frequency Trading Activity in China and Its Effects on Market Quality

The essentials

Question

Method

Findings

Limitations

Why it matters

Read further

Definitions

Tracking High-Frequency Trading in China: What Faster Information Changes

A Behavioral Fingerprint of Speed

Building and Testing the Measure

Precise Pricing Also Predicts Price Direction

What Happens When Information Arrives Faster?

A Mixed Market-Quality Scorecard

Why Context and Caveats Matter

Ask questions about this paper to our AI assistant

Tools for this paper

Advanced tools Premium

Papers around this one

Closest papers

Go further

Cite this paper (BibTeX)