05 · Algorithmic Trading and Compliance
Before 2023, algorithmic trading rules in China were "piecemeal": each exchange had its own abnormal-trade monitoring, each broker its own risk-control interpretations. In 2023–2024, new rules from the Shanghai/Shenzhen/Beijing exchanges plus the CSRC's administrative provisions on program trading in securities markets landed in succession — for the first time, algorithmic trading had a nationally unified framework. This article starts from definitions, unpacks the reporting regime, HFT supervision, and foreign experience, then lands on a compliance checklist for individual quant developers.
⚠️ Risk Warning
This article is an objective compilation of public knowledge, for study and research only, and does not constitute legal advice. Algorithmic trading regulation is being rapidly refined — reporting obligations, monitoring metrics, and penalty rules are all subject to the latest regulations. Before live deployment, confirm current practice with your broker/futures company.
1. What Is "Algorithmic (Program) Trading"
1.1 Definition (common framing, subject to the latest regulations)
| Dimension | Common Criteria |
|---|---|
| Core feature | Trading instructions automatically generated or executed by computer programs (not manual order-by-order entry) |
| Order speed | Programmatic order latency far below manual (milliseconds) |
| Batch submission | A single strategy batch-generating and submitting large volumes of orders within short windows |
| Common forms | Quant strategies, algorithmic order slicing (TWAP/VWAP), high-frequency market making, conditional orders / automated grids |
Plain judgment: as long as your buy/sell instructions are triggered by code rather than a human clicking through the keyboard, you most likely fall under program trading — regardless of how simple your strategy is or how low its frequency.
1.2 Why It Gets Special Regulation
| Risk | Explanation |
|---|---|
| Instant impact | Massive order piles causing violent price swings (e.g., programmatic amplification of the 2015 crash) |
| False signals | High-frequency cancellations fabricating fake liquidity, misleading other investors |
| Technical failures | Extreme orders from software bugs (fat fingers, runaway order loops) |
| Fairness | The arms race in infrastructure, speed, and data access widens the gap between institutions and retail |
2. China's Regulatory Landing (2023–2025 New Rules)
2.1 Timeline (common-sense summary)
| Date | Event |
|---|---|
| September 2023 | Shanghai/Shenzhen/Beijing exchanges issued draft implementation measures for program trading management, defining the reporting regime and monitoring requirements |
| 2024 | CSRC issued the Administrative Provisions on Program Trading in Securities Markets (Trial): the ministerial-rule-level framework for program trading regulation |
| 2024-2025 | Formal implementation measures took effect at the exchanges: reporting, monitoring, differentiated fees fully operational |
2.2 Core Points of the Administrative Provisions
| Point | Content |
|---|---|
| Reporting regime | Program traders must fulfill information reporting obligations to the exchange (after account opening, before trading) |
| Abnormal trade monitoring | Exchanges monitor in real time features like high-frequency order/cancel bursts and instant large orders, taking self-regulatory measures |
| Differentiated fees | High-frequency trading faces differentiated fees (higher order/cancellation costs), discouraging excessive trading |
| Prohibited conduct | Explicitly bans using program trading to manipulate markets (spoofing, ramping and slamming, etc.) |
| Gateway management | Management of brokers' in-house system access and trading gateways, preventing risk-control bypass via direct connections |
📖 Framing note
Operational details (reporting deadlines, indicator thresholds, fee schedules) are subject to the latest regulations — this section only explains the framework logic.
3. The Reporting Regime in Practice
3.1 Who Must Report
| Subject | Report? |
|---|---|
| Institutional program traders (quant private funds, broker prop desks, etc.) | Yes, and must designate a person responsible for reporting |
| Individual program traders (personal quant, automated grids, heavy conditional-order users) | Depends on rules: report once you meet the definition of program trading (subject to the latest regulations) |
| Ordinary investors only "clicking manually" in trading apps | Usually not program trading; no reporting needed |
Key common sense: "I'm a retail trader running some Python automation" still requires reporting — under the current framework, individual program traders are within scope too (subject to the latest regulations and broker practice).
🛑 Retail Python Automation Also Requires Reporting
"I'm a retail trader running some Python automation" still requires reporting. The current definition ignores identity and looks only at whether code triggers the orders — individual program traders are equally within the reporting obligation's scope.
3.2 What Gets Reported (common checklist)
| Item | Content |
|---|---|
| Basic information | Account, actual controller, sources of funds and securities |
| Strategy type | Strategy name and logic type (trend/arbitrage/market making/HFT etc.) |
| Server location | Where program trading servers are located, hosting arrangements (broker data center / own facility) |
| Technical parameters | Order frequency, cancellation ratio, maximum order volume etc. (per rule requirements) |
| Change reports | Changes to strategy, servers, controllers etc. require timely updated reports |
3.3 Reporting Process and Cooperation Duties
- Report to the exchange through your broker/futures company (brokers are the execution gateway).
- Brokers may conduct compliance reviews of program traders and restrict access when necessary.
- Consequences of non-reporting or false reporting: exchanges may impose self-regulatory measures such as trading restrictions; serious cases referred to the CSRC (subject to the latest regulations).
4. High-Frequency Trading Supervision
4.1 Recognizing HFT Characteristics (common framing)
| Indicator | Common Recognition Direction (subject to the latest regulations) |
|---|---|
| Order rate | Orders per second above a threshold (starting from several per second; exact thresholds per rules) |
| Cancellation ratio | Abnormally high proportion of rapid post-submission cancellations (many orders never filling) |
| Order-to-fill ratio | Submissions vs fills excessively skewed (place 100, cancel 95) |
| Daily cumulative volume | Single-day total orders reaching exchange monitoring thresholds |
4.2 Regulatory Tools
| Tool | Logic |
|---|---|
| Differentiated fees | Higher fees on high-frequency order/cancel flows — using cost to suppress meaningless traffic |
| Abnormal trade monitoring | Real-time detection of instant large orders, frequent cancels, ramping/slamming patterns |
| Restrictions | Position limits, trading restrictions, account suspensions (serious cases) |
| Look-through checks | Tracing back to actual controllers, devices, and strategies (echoes look-through supervision; see china-regulation.md) |
What it means for individual quants: ordinary personal strategies (minute-level, second-level frequencies) usually fall far below "high-frequency" thresholds — most individual strategies are outside HFT supervision's core range, but that does not waive the reporting obligation.
5. Foreign Experience
5.1 United States: Reg NMS and Fee Models
| Regime | Content |
|---|---|
| Reg NMS (2005) | National Market System rules: price protection (best-price priority), locked/crossed market prohibitions, order visibility requirements |
| Reg ATS | Alternative Trading System oversight: dark pools and other ATS must register with and report to the SEC |
| Maker-taker model | Exchanges pay rebates to liquidity providers (makers) and charge liquidity takers — fee structures incentivizing market making and passive HFT quoting |
| FINRA HFT surveillance | Surveillance of HFT firms' order traffic and cancellation behavior |
Key takeaway: US markets treat HFT as part of the market structure to be regulated rather than banned — regulatory focus targets abusive behavior (spoofing, layering), not speed itself.
5.2 EU: MiFID II Requirements for HFT
| Requirement | Content |
|---|---|
| Algo identification | Institutions engaging in algorithmic trading must report to regulators |
| Frequency threshold | Quantitative criteria define HFT (order rates, intraday order volume) |
| Market-making obligations | HFT market makers sign market-making agreements and must provide continuous liquidity |
| Circuit breakers & monitoring | Exchanges must have volatility interruptions and abnormal trade monitoring mechanisms |
5.3 United Kingdom
- The FCA applies the MiFID II framework to algo/HFT (core requirements retained post-Brexit).
- Emphasis on algorithm governance: institutions need complete internal controls over strategy development, testing, deployment, and rollback.
6. Impact on Individual Quants
6.1 Do Individuals Need to Report?
| Scenario | Judgment (subject to the latest regulations) |
|---|---|
| Conditional/grid orders via exchange/broker apps (cloud execution) | Mostly built-in software features executed by the broker's system; generally no separate reporting |
| Writing your own program to place orders via APIs/quant platforms | Is program trading; usually requires reporting |
| Strategies used only for backtesting, never connected to live accounts | No live trading involved; no reporting needed |
💡 Action advice: if unsure, just ask your broker's compliance department
When uncertain, directly ask the compliance department of your broker/futures company — the responsibility for whether to report sits with the trader; brokers must assist but won't decide for you.
6.2 Consequences of Violations (common-sense summary)
| Situation | Possible Consequences |
|---|---|
| Required but missing reports | Exchange/broker demands rectification; possible restriction of program trading permissions |
| Abnormal trade patterns detected | Inquiries, interviews, trading restrictions (days to months) |
| HFT order+cancel abuse found | Differentiated fees (rising costs) + restrictive measures |
| Manipulation via program trading | Administrative penalties + criminal liability (securities/futures market manipulation crime) — the red line among red lines |
7. Compliance Advice for Quant Development
7.1 Keep Audit Trails
| Trail Item | Content |
|---|---|
| Order logs | Complete timestamps (millisecond precision), price, quantity, status for every order/cancel |
| Strategy versions | Code version and deployment records for every strategy change |
| Account statements | Reconciled against broker statements, regularly |
| Parameter configs | Historical settings of risk parameters (max order size, max positions, daily loss limit) |
💡 Why trails matter: the strongest evidence of your own compliance
When questioned, complete logs and version history are the strongest evidence that you operated compliantly; unrecorded trading behavior equals "unexplainable conduct" in regulators' eyes.
7.2 Risk Control Thresholds (pre-trade controls)
| Control | Suggestion |
|---|---|
| Per-order limits | Hard caps on max amount/lots per order |
| Position limits | Max position per product, overall position cap |
| Cancellation ratio | Build a cancellation-ratio ceiling into the strategy; auto-stop when exceeded |
| Order frequency | Caps on orders per second/minute to prevent runaway loops |
| Loss circuit breaker | Auto-halt trading when daily losses hit the threshold |
| Network disconnect protection | No auto re-submission after disconnects, preventing duplicate orders |
7.3 Avoiding Abnormal Behavior Patterns
| Avoid | Why |
|---|---|
| Massive instant submissions followed by mass cancels | Classic "spoofing" suspicion pattern; can be deemed manipulation |
| High-frequency order/cancel loops | Raises surveillance hit rates and differentiated fee costs |
| Splitting across accounts to evade monitoring | Deliberate multi-account/small-order evasion = look-through priority (see china-regulation.md) |
| Running strategies without risk controls | One bug can turn your strategy into "abnormal trading" itself |
7.4 Individual Quant Compliance Checklist (self-check directly)
- Before going live with automation, confirm with your broker/futures company whether reporting is required; if so, complete it.
- Report truthfully: server location, strategy type, account info — update promptly on changes.
- Build pre-trade risk controls into every strategy: five gates for order size/positions/losses/cancel ratio/frequency.
- Keep complete order logs and version trails, reconciling regularly.
- Never do anything resembling manipulation: no spoofing, no wash trades, no splitting to evade surveillance.
- Follow the latest exchange and CSRC rules: detailed measures here are still updating fast.
8. Supervision Quick Reference
| Dimension | China (current framework) | US | EU |
|---|---|---|---|
| Reporting duty | Yes (via brokers) | Yes (FINRA/SEC) | Yes (MiFID II) |
| HFT definition | Order-rate / cancel-ratio thresholds | Watched by regulators, no unified frequency ban | Quantitative criteria |
| Fees | Differentiated HFT fees | maker-taker market mechanism | No unified differentiated fees |
| Manipulation ban | Explicit enumerated prohibited acts | Reg ATS/Reg NMS + anti-manipulation rules | Market Abuse Regulation |
Further Reading
- Engineering implementation of quant strategies: live-automation.md in Chapter 15 · Quant Practice
- Risk control architecture and audit trails: risk-systems.md in Chapter 10 · System Integration
- Look-through supervision and the penalty framework: china-regulation.md
- Spotting manipulative behavior (wash trades, spoofing in the wild): manipulation-detection.md in Chapter 12 · Market Ecosystem