Competition Law and Competition Concerns in Market Sensing Systems

1. Introduction

Market sensing systems are technologies used by firms to continuously collect, process and interpret information about market conditions. They may monitor:

• competitors' prices and discounts;
• consumer demand and purchasing patterns;
• inventory and capacity;
• product launches;
• advertising activity;
• supply-chain conditions;
• customer switching;
• competitor promotions;
• market sentiment and search behaviour;
• platform activity and transaction data; and
• real-time changes detected through AI or machine-learning systems.

Market sensing is not inherently anti-competitive. In fact, collecting publicly available market information can improve forecasting, reduce transaction costs and promote innovation. The competition-law concern arises when the system becomes a mechanism for exchanging commercially sensitive information, coordinating competitive conduct, facilitating exclusion, discriminatory pricing, or stabilising a cartel.

Modern competition authorities increasingly distinguish between an algorithm that merely observes the market and one that turns market observation into coordinated conduct. The European Commission, for example, expressly recognises that algorithms may legitimately monitor competitors' prices but can also make collusion easier by detecting deviations and enabling rapid punishment.

2. Meaning of Market Sensing Systems

A market sensing system can be understood as a technological system having four principal functions:

A. Market observation

The system collects information concerning:

• prices;
• output;
• demand;
• competitors;
• customers;
• inventory;
• advertising;
• market shares; and
• product characteristics.

B. Data aggregation

Information from multiple sources is consolidated into a common database or analytical environment.

C. Market prediction

AI or statistical models predict:

• future demand;
• competitor reactions;
• price movements;
• consumer switching;
• supply shortages; and
• likely market responses.

D. Competitive action

The system may then recommend or automatically implement:

• prices;
• discounts;
• quantities;
• advertising;
• inventory allocation;
• bidding strategies; or
• customer targeting.

The fourth stage creates the greatest competition-law risk because market intelligence becomes connected to actual competitive behaviour.

3. Competition-Law Framework

Market sensing systems can implicate several areas of competition law.

A. Anti-cartel provisions

The most serious concern is coordination between competitors.

For example:

Competitor A and Competitor B independently feed their confidential pricing information into the same market-sensing system, which recommends prices designed to avoid undercutting each other.

The technological system may become the mechanism through which competition is reduced.

Under EU law, algorithmic coordination can constitute a restriction of competition where algorithms are used to facilitate coordination on important competitive parameters.

B. Information exchange

Competition law is particularly sensitive to exchanges of commercially sensitive information.

Potentially problematic information includes:

• future prices;
• future production;
• individual customer information;
• margins;
• strategic discounts;
• capacity;
• future business plans;
• non-public demand forecasts.

The danger increases where information is:

1. current or forward-looking;
2. commercially sensitive;
3. individualised;
4. exchanged frequently; and
5. supplied to competitors.

A market-sensing platform that converts individual competitor information into a common competitive database may therefore create substantial risk.

4. Market Sensing and Algorithmic Collusion

There are several possible models.

Model 1 — Independent sensing

Each firm independently collects publicly available information.

Competition risk: comparatively limited.

Model 2 — Common information provider

Competitors purchase market intelligence from a common provider.

Risk: depends on the nature and sensitivity of the information supplied.

Model 3 — Competitor-data aggregation

Competitors provide non-public information to a common system.

Risk: substantially higher.

Model 4 — Algorithmic coordination

The system recommends prices or strategies based on competitors' confidential information.

Risk: potentially very high.

Model 5 — Autonomous coordination

AI systems independently learn that maintaining particular prices maximises profits.

This raises a difficult question:

Can competition law address coordinated outcomes when there is no conventional human agreement?

The answer differs among jurisdictions and depends heavily on evidence of communication, knowledge, design, participation and concerted conduct.

5. Six Major Case Laws

1. United States v. Topkins

Jurisdiction: United States
Law: Sherman Act §1

This is one of the foundational algorithmic-pricing cases.

Online sellers of posters allegedly agreed to coordinate prices and used pricing software to implement the arrangement. The important point is that the algorithm did not immunise the underlying cartel.

Competition-law principle

Technology used to implement an existing price-fixing agreement remains subject to ordinary antitrust law.

The case therefore illustrates the "messenger" model:

Human agreement → algorithmic implementation → coordinated prices.

The algorithm is the instrument rather than the legal source of the agreement.

2. Eturas UAB v Lietuvos Respublikos konkurencijos taryba

Court: Court of Justice of the European Union
Case: C-74/14

Eturas concerned an online travel-booking platform whose system implemented a restriction on discounts available to participating travel agencies.

The case is particularly important for market-sensing and platform systems because it considered whether knowledge of an automated platform restriction, together with conduct suggesting acceptance, could establish participation in concerted conduct.

Principle

An automated system message does not automatically establish that every recipient participated in a cartel.

Evidence concerning:

• receipt of the communication;
• knowledge;
• participation;
• distancing;
• subsequent conduct; and
• circumstances surrounding the system

remains important.

Relevance

Market-sensing platforms must therefore maintain clear separation between:

legitimate market intelligence → prohibited coordination.

3. United States v. RealPage, Inc.

Jurisdiction: United States

RealPage represents a modern application of antitrust principles to algorithmic market sensing.

The U.S. Department of Justice alleged that landlords supplied non-public information, including rental and occupancy information, to a pricing system that generated recommendations using aggregated data.

The important competition concern was not simply that landlords used software. It was the alleged combination of:

competitors' confidential information + common algorithm + pricing recommendations.

The DOJ's theory illustrates why a market-sensing system can become problematic when it transforms competitors' proprietary information into coordinated pricing recommendations.

Principle

The legal analysis may focus on the information architecture and competitive relationship, not merely on whether the final prices are identical.

4. Samir Agarwal v. Competition Commission of India

Court: Supreme Court of India
Context: Ola and Uber

This case concerned allegations relating to algorithmic pricing in ride-hailing.

The allegation was that algorithmic pricing reduced the ability of individual drivers to compete independently because the platform determined fares through its algorithm.

The case is significant because it illustrates the distinction between:

• independent pricing decisions by drivers;
• platform-mediated pricing;
• a hub-and-spoke arrangement; and
• an actual agreement to fix prices.

Principle

The existence of a common algorithm or common pricing mechanism does not by itself establish a cartel. Evidence of the necessary agreement or concerted arrangement remains critical.

This is particularly important for market-sensing systems because a common technological infrastructure can produce similar market outcomes without necessarily establishing unlawful coordination.

5. Ohio v. American Express Co.

Court: U.S. Supreme Court
Citation: 585 U.S. 529 (2018)

This case did not concern a market-sensing system specifically, but it is highly relevant to digital and data-driven multi-sided markets.

The Supreme Court treated the credit-card system as a two-sided transaction platform and required the competitive effects of the challenged conduct to be considered in the relevant market structure.

Relevance to market sensing

Market-sensing systems frequently operate in:

• platforms;
• marketplaces;
• advertising networks;
• payment systems;
• app ecosystems.

Their effects may therefore occur simultaneously on several sides of a market.

A competition analysis cannot necessarily examine only the side on which the algorithm directly changes prices or conduct.

6. United States v. Airline Tariff Publishing Co.

Court: U.S. courts / Department of Justice enforcement
Context: Airline pricing information

Airline reservation and tariff systems allowed carriers to communicate and observe fare changes through sophisticated information systems.

The case is historically important because it demonstrates how rapid dissemination of pricing information can facilitate coordination even when the technology appears to be merely an information system.

Principle

A technologically sophisticated information system can become competitively problematic where it:

• increases market transparency among competitors;
• communicates future pricing intentions;
• enables rapid retaliation;
• reduces uncertainty about competitors' behaviour.

This is directly analogous to modern AI market-sensing systems.

6. Additional Important Authority: Online Poster Sellers in the UK

UK competition enforcement involving online sellers of posters and frames provides another useful illustration.

Competing sellers used automated repricing software after agreeing to restrict price competition. The software automatically implemented the agreed pricing strategy.

Principle

Automated repricing does not change the underlying legal character of an agreement.

The important question remains:

What did the firms agree to do, and how was the technology used to implement that agreement?

7. Market Sensing as a Hub-and-Spoke Problem

One of the most important risks is the hub-and-spoke model.

Imagine:

Competitor A
↓
Market-Sensing Platform
↑
Competitor B

If the platform receives confidential information from A and communicates competitively significant information to B, the platform may become a hub connecting otherwise separate competitors.

The risk increases if competitors know that:

• rivals are participating;
• rivals are submitting confidential data;
• the same algorithm processes their information; and
• the system generates recommendations affecting competitive behaviour.

The Indian Ola/Uber litigation and EU Eturas jurisprudence demonstrate why the existence of a technological intermediary must be analysed together with evidence of communication, knowledge and agreement.

8. Market Sensing and Tacit Collusion

A particularly difficult issue is tacit coordination.

Suppose competing AI systems independently observe:

• competitor prices;
• demand;
• inventory; and
• customer responses.

Each algorithm learns that aggressive price cuts are unprofitable and begins maintaining higher prices.

There may be:

• no telephone call;
• no meeting;
• no written cartel agreement;
• no direct communication.

Nevertheless, prices may converge.

Competition-law difficulty

Traditional cartel law generally requires some form of agreement or concerted action.

Therefore:

parallel algorithmic behaviour ≠ automatically unlawful cartel.

But the risk becomes greater where there is evidence that firms:

• deliberately designed algorithms to coordinate;
• exchanged confidential information;
• used a common pricing mechanism;
• agreed to follow algorithmic recommendations;
• communicated future pricing intentions; or
• used the system to punish deviations.

The European Commission expressly recognises this distinction between legitimate independent algorithmic monitoring and algorithms used to facilitate collusion.

9. Market Sensing and Dominance

Market sensing systems can also create abuse-of-dominance concerns.

A dominant digital platform may have access to vastly more market information than competitors because it controls:

• transactions;
• search data;
• customer behaviour;
• seller information;
• advertising data;
• logistics information.

The dominant platform can potentially use this information to:

A. Identify emerging competitors

The system may detect successful competitors before they become significant.

B. Copy successful products

The platform can observe demand patterns and introduce competing products.

C. Discriminate against rivals

The system may identify merchants selling competing services and alter:

• ranking;
• visibility;
• commissions;
• access;
• recommendations.

D. Self-preference

The platform may sense market demand and use that information to promote its own products.

E. Foreclose competitors

Information advantages can potentially be converted into exclusionary strategies.

10. China-Specific Competition Concerns

China's competition-law framework is particularly relevant because digital platforms and algorithmic conduct have received increasing regulatory attention.

The Anti-Monopoly Law, platform-economy rules and related regulatory measures can address conduct involving:

• algorithmic pricing;
• discriminatory pricing;
• platform rules;
• data advantages;
• restrictions imposed on merchants;
• exclusionary platform practices.

China's 2026 regulatory measures specifically address algorithm-driven discriminatory pricing and "big data kill-the-familiar" practices involving different treatment of consumers based on platform data.

A particularly relevant 2026 development is Huolala.

SAMR ordered Huolala to stop using algorithms in ways that unreasonably suppressed freight rates and required changes to its dynamic-pricing practices. The regulator also required greater disclosure of pricing rules and explanations for price increases or reductions.

This is important because it demonstrates that competition concerns surrounding algorithms are not limited to classic competitor cartels. They can also concern how a dominant or powerful platform's sensing and pricing architecture affects participants on the platform.

At the same time, publicly available Chinese enforcement decisions specifically establishing an algorithmic-pricing cartel remain more limited than the regulatory attention devoted to the issue.

11. Data Accumulation as a Competition Concern

Market sensing depends heavily on data.

A large platform may have access to:

• millions of transactions;
• customer histories;
• real-time prices;
• competitor inventory;
• search histories;
• seller conversion rates;
• location information;
• demand elasticity.

This can create a data feedback loop:

More users → more data → better sensing → better predictions → better service → more users → still more data

This can produce significant economies of scale and potentially increase barriers to entry.

Competition authorities therefore need to distinguish between:

Legitimate data advantage

A firm becomes more efficient because it develops superior analytical capabilities.

Potentially exclusionary data advantage

A dominant firm uses its control over essential or strategically important data to disadvantage competitors.

12. Market Sensing and Predatory or Discriminatory Pricing

An advanced sensing system can estimate individual or group-level willingness to pay.

This creates potential risks involving:

• personalised pricing;
• discriminatory pricing;
• targeted discounts;
• loyalty-based price differentiation;
• exclusionary discounts.

The concern is especially significant where the platform possesses information unavailable to competitors.

China's recent platform regulation specifically addresses the use of big data for discriminatory pricing, demonstrating the increasing regulatory significance of algorithmic consumer segmentation.

13. Market Sensing and Merger Control

Market sensing systems also matter in merger analysis.

Two firms may each possess:

• customer databases;
• competitor intelligence;
• pricing data;
• demand forecasts;
• AI models.

A merger could combine these datasets.

The competition question becomes:

Does the transaction create a strategically important information advantage that competitors cannot realistically reproduce?

Relevant concerns include:

1. data concentration;
2. increased entry barriers;
3. reduced privacy or quality competition;
4. enhanced price discrimination;
5. increased ability to monitor competitors;
6. strengthening of ecosystem effects.

Thus, even where traditional market-share measures appear moderate, data and sensing capabilities may be relevant to competitive assessment.

14. Market Sensing and Essential Facilities

In some digital markets, market data can become an important input.

Potential examples include:

• financial transaction data;
• mobility data;
• electricity-grid data;
• search data;
• advertising data;
• healthcare data;
• logistics information.

If a dominant undertaking controls uniquely important data and refuses access to competitors, competition-law questions may arise concerning:

• indispensability;
• refusal to supply;
• discriminatory access;
• interoperability;
• objective justification;
• foreclosure.

However, not every commercially valuable dataset is an essential facility. The legal thresholds must be satisfied before compulsory access is required.

15. Competition Risks by System Design

16. Key Legal Tests

When examining a market-sensing system, competition authorities are likely to ask:

Question 1

What information is collected?

Public information presents a different problem from confidential competitor information.

Question 2

Who provides the information?

Information voluntarily supplied by competitors requires closer scrutiny.

Question 3

Who can access the information?

Access limited to one firm differs from a common competitor-facing system.

Question 4

How frequently is information updated?

Real-time information can substantially increase the ability to monitor deviations.

Question 5

Is information aggregated or identifiable?

Aggregated historical information may present less risk than individual forward-looking data.

Question 6

What does the algorithm do with the information?

Simple reporting differs from automatic pricing recommendations.

Question 7

Can the algorithm punish deviations?

This is particularly relevant to cartel stability.

Question 8

Is there evidence of agreement or concerted action?

Technological similarity alone should not automatically be equated with a cartel.

17. Compliance Framework for Market Sensing Systems

Businesses using market-sensing technology should consider the following safeguards.

A. Data classification

Classify information as:

• public;
• aggregated;
• historical;
• commercially sensitive;
• confidential;
• forward-looking.

B. Competitor-data firewall

Do not allow confidential competitor information to flow freely into pricing or strategy systems.

C. Independent pricing authority

Where possible, ensure that competitive decisions remain independently determined.

D. Algorithm governance

Maintain:

• model documentation;
• input logs;
• version histories;
• approval records;
• audit trails.

E. Third-party platform controls

Contracts with data and algorithm providers should identify:

• permissible data sources;
• data-sharing restrictions;
• competitor-data restrictions;
• audit rights;
• compliance obligations.

F. Human oversight

High-impact pricing or strategic recommendations should receive appropriate human review.

G. Competition-law audits

Algorithms should periodically be examined for:

• coordinated outcomes;
• discriminatory treatment;
• competitor-data leakage;
• self-preferencing;
• exclusionary effects.

18. Important Distinction: Sensing vs Coordination

The central competition-law distinction can be represented as follows:

Market observation

↓

Data collection

↓

Data analysis

↓

Prediction

↓

Independent business decision

→ generally legitimate competitive intelligence

Whereas:

Competitor information

↓

Common platform

↓

Shared algorithm

↓

Coordinated recommendation

↓

Parallel implementation

→ potentially serious competition-law concern.

The difference is not the mere presence of AI. It is the relationship between information, competitors, algorithmic architecture and competitive decision-making.

19. Six-Case Comparative Summary

20. Conclusion

Market sensing systems occupy a legally important boundary between legitimate competitive intelligence and technologically facilitated anti-competitive conduct.

The technology itself is generally not the problem. The central questions are:

1. what information is collected;
2. whether competitors' confidential information is involved;
3. who receives the information;
4. whether the system facilitates communication between competitors;
5. whether the algorithm recommends or implements coordinated conduct;
6. whether a dominant platform exploits sensing advantages to exclude rivals; and
7. whether the resulting conduct produces legally cognisable anti-competitive effects.

The major cases—from Topkins and Eturas to RealPage and Samir Agarwal—show the evolution from traditional human cartels implemented through software toward increasingly sophisticated forms of algorithmic coordination.

For China in particular, recent regulatory developments demonstrate that algorithmic pricing and data-driven platform conduct are receiving increasingly direct attention. SAMR's 2026 Huolala action is especially significant because it shows regulatory intervention where algorithmic systems were considered to have contributed to problematic pricing outcomes, even apart from a conventional competitor cartel.

Accordingly, the modern competition-law principle can be stated as:

Market sensing is ordinarily a competitive tool; market sensing that converts competitors' sensitive information into coordinated, exclusionary, or discriminatory conduct can become a competition-law problem.