Thread by @Wayne145591

The B-Book Model in Cryptocurrency Exchanges: Mechanisms, Arbitrage Opportunities, and Systemic Risks
Author: Waynecoin
Affiliation: Alpha 9 Ventures
Contact: X:@Wayne145591
Date: August 2025

Abstract
Against the backdrop of the rapid development of the cryptocurrency trading market, exchange business models have become increasingly diversified. Among them, B-Book exchanges have quickly risen due to their ability to offer high leverage, strong liquidity provisioning, and low operating costs. Unlike traditional A-Book exchanges, the B-Book model does not route orders to external liquidity providers. Instead, the exchange itself acts as the market maker, directly taking the opposite side of users’ trades. This model enables exchanges to capture user trading losses directly, creating high profit potential, but it also exposes them to significant risk during strong one-sided market trends.
This paper takes B-Book exchanges as the research subject and conducts a systematic study on three levels:

It explains the operational logic and historical background of the B-Book model, drawing comparisons with similar mechanisms in the forex market, contracts for difference (CFDs), and the gambling industry.

It analyzes the arbitrage opportunities and technical loopholes in B-Book exchanges, including latency arbitrage, extreme order execution flaws, risk control detection loopholes, promotional capital arbitrage, reverse control arbitrage, and high-leverage hedging with rebate farming. Mathematical formulas, operational models, and practical case studies are provided to illustrate these points.

It explores in depth the profit-and-loss structure and risk-control mechanisms of B-Book exchanges, showing how they generate stable profits in sideways markets through “long–short double-kills” and transaction fees, while in trending markets they may suffer heavy losses due to risk-control delays or unexpected retail traders being “on the right side” of the move.

The findings indicate that although arbitrage opportunities exist in B-Book exchanges, their sustainability is constrained by risk-control systems and back-end intervention. The key to long-term survival for arbitrageurs is to conceal arbitrage patterns by randomizing and dispersing trading behavior, making strategies appear closer to those of regular users, thereby reducing the risk of being flagged or switched to A-Book execution. This study not only reveals the dealer-style logic and arbitrage mechanisms underpinning B-Book exchanges, but also provides a useful reference framework for quantitative traders, risk-control modelers, and market regulators.

Introduction Research Background Since 2017, the cryptocurrency market has experienced explosive growth, with exchanges playing a central role as critical infrastructure. As the market has matured, exchange models have diverged into two main categories: 1.A-Book Model – User orders are routed to external liquidity providers (such as market makers or major exchanges), with spreads and fees as the primary revenue source. 2.B-Book Model (Dealer/Counterparty Exchanges) – The exchange itself acts as the counterparty, profiting directly from user losses. In traditional financial markets, the B-Book model is common in Forex and CFD platforms and is often criticized as “the house betting against retail.” However, in the crypto market, B-Book adoption has become widespread due to high leverage demand, insufficient liquidity in smaller tokens, and cost pressures on small exchanges. Some exchanges even default to B-Book execution for derivatives products, routing only a minority of profitable accounts to A-Book—creating a “hybrid model.” Research Motivation The essence of the B-Book model shapes a profit-and-loss structure that is markedly different from traditional exchanges: In sideways markets, most users lose money due to high leverage and frequent stop-outs, allowing exchanges to profit from “long–short double-kills.” In trending markets, if retail traders happen to take the right side, exchanges may incur heavy losses if unable to hedge in time. This duality creates reflexivity in B-Book exchanges: they are not only influenced by market trends but also feed back into market dynamics through their internal matching logic. Analyzing their operational framework thus helps arbitrageurs identify opportunities and allows researchers to better understand crypto market microstructure and systemic risks. Literature Review and Comparisons Forex B-Book: Academic studies show that forex B-Book brokers use statistical models to predict retail loss probabilities, achieving long-term profitability rates exceeding 80%. Gambling Industry: Casinos design games with mathematical expectations that systematically disadvantage players. Similarly, B-Book exchanges rely on high leverage and fees to push retail traders toward losses. Crypto Exchange Uniqueness: Unlike forex, liquidity in crypto is fragmented, and significant delays and price discrepancies exist between exchanges—creating additional technical arbitrage opportunities. Research Questions Based on the above, this paper aims to address the following questions: How does the B-Book model operate, and what are its economic and technical foundations? What exploitable loopholes and opportunities exist for arbitrageurs under this model? How do B-Book exchanges design risk-control systems and back-end interventions to restrict arbitrage, and how can traders avoid them? How do profit-and-loss characteristics evolve under different market structures (sideways vs. trending)? Research Methodology This study employs a combination of theoretical modeling, case analysis, and strategy simulation: Theoretical: Establish profit-loss functions and arbitrage condition formulas for B-Book exchanges. Practical: Analyze real-world cases (e.g., latency arbitrage, promotional bonus arbitrage). Strategic: Propose sustainable evasion methods to simulate long-term arbitrage survival strategies. Research Contributions Provides an academic-level, systematic study of the B-Book model within crypto markets. Offers arbitrageurs a comprehensive framework for strategies and risk-control evasion. Reveals systemic risks of the B-Book model to exchanges and regulators.

Chapter 1 – Operating Principles of B-Book Exchanges

1.1 What Is a B-Book Exchange
A B-Book exchange operates similarly to the B-Book model in Forex and CFD platforms. The core logic is that the exchange itself acts as the market maker, directly becoming the counterparty to user trades, rather than routing orders to real markets or external liquidity pools.

This means that when a trader opens a long position on a B-Book exchange, they are effectively “betting against the house”: if the trader profits, the exchange loses; conversely, if the trader loses, the exchange gains directly. This stands in fundamental contrast to the A-Book model, where the exchange acts only as an intermediary, passing orders to external liquidity providers.

1.2 Why Do B-Book Exchanges Exist?
The main reasons are:

1. Profit Maximization – By directly capturing user losses, exchanges are not fully dependent on transaction fees for revenue. In the A-Book model (routing orders to external liquidity providers), exchanges must pay spreads or fees to outside market makers. B-Book keeps orders internal, avoiding these costs.

2. Lower Liquidity Costs – For assets with low trading volume and poor external liquidity (e.g., obscure altcoins, small-cap contracts, leveraged tokens), A-Book execution is expensive and inefficient. B-Book ensures that users can always transact, as the exchange provides “low-cost liquidity” internally without relying on third-party providers.

3. Controllable Risk – By applying risk-control models, exchanges can route profitable traders to A-Book, while retaining losing traders in B-Book.

4. High Leverage & Fast Execution – For highly leveraged products (50x, 100x perpetuals), hedging externally is costly and difficult to execute quickly. B-Book allows instant internal matching, ensuring smoother user experience.

5. Casino-Style Model – Like gambling houses, exchanges adopt the dealer role to maintain long-term profitability.
1.3 Comparison with the A-Book Model

In traditional finance, the key difference between A-Book and B-Book lies in the source of profit and loss:

- A-Book:
- Revenue → Transaction fees, spread rebates
- Profit/Loss → Between users (buyers vs. sellers), with the exchange acting purely as intermediary
- Risk → Nearly zero, as the exchange does not bear market risk

- B-Book:
- Revenue → User losses (plus fees)
- Profit/Loss → Users vs. Exchange directly
- Risk → Highly correlated; if retail traders are net profitable, the exchange incurs losses

Mathematically:
- A-Book: Π_exchange = Σ Fees_i
- B-Book: Π_exchange = -Σ PnL_users + Σ Fees_i

Here, PnL_users denotes user profits and losses. This illustrates why B-Book exchanges are heavily dependent on retail traders’ long-term losses.

1.4 Mathematical Model of the B-Book

Let:
- N = number of users
- p = probability of a user winning a single trade
- q = 1-p = probability of a user losing a single trade
- L = average loss amount
- G = average gain amount

Then, the exchange’s expected profit is:
E[Π] = q * L - p * G

Since most users employ high leverage and trade irrationally, typically q >> p. This makes long-term exchange profitability almost inevitable, mirroring the probabilistic edge of casinos.

In practice, exchanges also use dynamic order routing:
- Profitable accounts → sent to A-Book (reducing risk).
- Losing accounts → kept in B-Book (amplifying profits).

Thus, the practical model is closer to:
E[Π] = (q * L - p * G) * (1 - α)

Where α is the proportion of “winning” traders redirected to A-Book.

1.5 Similarities to the Gambling Industry

- House Edge: Just as casino games (roulette, blackjack) are mathematically designed to favor the dealer, B-Book exchanges structure leverage, spreads, and fees to systematically disadvantage retail traders.
- Illusion of Winning: Retail traders often overestimate their chances of success and overtrade, just as gamblers believe they are “due for a win.”
- Long-Term Outcome: The exchange, like a casino, absorbs capital consistently through its probabilistic edge.
1.6 Case Extensions
- Case 1: Illiquid Altcoin Contracts
A small exchange lists a niche altcoin with no external liquidity. Under A-Book, it would incur high costs paying market makers. Under B-Book, trades can be matched internally at low cost, meeting user demand efficiently.

- Case 2: High-Leverage Perpetuals
Suppose a retail trader goes long BTC/USDT with 100x leverage, starting with $100 USDT. The notional position = $10,000. If BTC drops just 1%, the position is liquidated, and the trader loses $100—this loss becomes direct revenue for the exchange. This is one of the most common profit sources in B-Book models.

1.7 Summary
The existence of B-Book exchanges stems from their high profitability, low costs, and controllable risks. They combine the market-making logic of finance with the dealer logic of gambling. For traders, understanding these mechanics is a prerequisite for designing effective arbitrage strategies.

Chapter 3 – Profit and Loss Characteristics of B-Book Exchanges in Ranging vs. Trending Markets

3.1 Introduction
Market structure has a decisive influence on the profit-and-loss characteristics of B-Book exchanges. Traditional financial research shows that the profitability of a dealer-style model depends heavily on player behavior and market conditions. In ranging markets, retail traders frequently trigger stop-losses and use high leverage, leading to consistent losses and stable exchange profits. In trending markets, however, if retail traders unexpectedly align with the trend, the exchange faces substantial risks.

3.2 Advantages and Risks in Ranging vs. Trending Markets
Advantages of Ranging Markets for B-Book Exchanges:
1. High leverage increases the probability of stop-outs, creating long–short “double-kills.”
2. Frequent user entries and exits generate more fee income.
3. Nearly 99% of users fall into repeated stop-loss and re-entry cycles due to cognitive biases.

Risks of Trending Markets for B-Book Exchanges:
1. Retail traders unexpectedly align with the correct trend, especially under high leverage, amplifying profits.
2. Risk-control and hedging delays—particularly in rapid markets—make it impossible to offset losses in time, as spreads widen sharply across all venues.
3. Reverse stop-loss chains reinforce trend continuation.
4. The inability to exploit oscillations means the exchange loses its “two-sided harvesting” advantage.

3.3 Supplementary Analysis: Ranging Markets
3.3.1 Mathematical Model
Assume the market is ranging, with price fluctuation ΔP within [−x, x], where x is typically smaller than the liquidation threshold for retail traders.
Let:
- λ = user leverage multiple
- M = user margin
- L = 1/λ = liquidation threshold percentage
If |ΔP| ≥ L, the user is liquidated, losing M.
Because prices oscillate back and forth in ranges, most traders hit stop-losses within short intervals, creating long–short double-kills.
The exchange’s expected profit:
E[Π_bb] = Σ P(liquidation_i) ⋅ M_i + Σ Fees_i
3.3.2 Behavioral Finance Notes
- Retail traders’ “loss aversion” and “chasing gains/cutting losses” behaviors are magnified in ranging markets.
- With no clear trend, traders re-enter frequently, raising the exchange’s harvesting efficiency.
3.3.3 Case Study
- In 2023, a small exchange’s BTC/USDT contract ranged between $27,000–$27,800 for a week.
- Retail traders widely used 50x–100x leverage (liquidation thresholds <2%).
- Since the market fluctuated more than 3%, nearly all long and short positions were wiped out. The exchange netted over $5 million USDT in profit from this single product within one week.

3.4 Supplementary Analysis: Trending Markets
3.4.1 Mathematical Model
Let ΔP represent the one-sided market movement (e.g., price increase).
If retail traders are predominantly long, the exchange’s losses are:
Loss_bb = Σ (ΔP ⋅ λ_i ⋅ M_i) − Σ Fees_i
Where:
- λ_i = leverage multiple
- M_i = margin
When retail traders concentrate on the correct trend, losses scale as linear accumulation + leverage amplification, potentially overwhelming the exchange’s liquidity reserves in a short period.
3.4.2 Risk Transmission
Hedging Delay Risk: During rapid moves, a B-Book exchange attempting to hedge through A-Book routing may fail due to latency, resulting in overexposure.
Stop-Loss Chain Effect: As one group of users takes profit, the opposite group is forced to stop out, reinforcing the trend and worsening exchange losses.
3.4.3 Case Studies
- In May 2021, during a crypto market crash, one B-Book exchange failed to hedge in time. As a result, 80% of users with short positions profited massively, creating a $200 million USDT funding gap. The exchange was forced to halt withdrawals temporarily.
- Similar cases exist in Forex markets. For example, during the 2015 Swiss Franc “black swan” event, several B-Book Forex brokers went bankrupt when retail traders were collectively positioned correctly.

3.5 Comparing P&L Structures: Ranging vs. Trending
Ranging Markets → Retail traders suffer long–short double-kills and churn trades frequently. Exchanges profit from liquidations + fees, with minimal risk unless whales manipulate the market.
- Trending Markets → Retail traders may profit massively, and transaction fees cannot offset losses. Sustained one-sided price moves cause large-scale overexposure and liquidity shortages.

Formally:
E[Π_range] = High-Leverage Liquidation Profits + Fee Income
E[Π_trend] = −Retail Profits + Fee Income

Since retail profits usually far exceed fee income, exchanges suffer severe losses in strong trends.
3.6 Extensions: Market Control and Structure
- Manipulation Space: In ranging markets, exchanges can further harvest users via stop-hunts (“wicks”) or by adjusting matching logic. In trending markets, however, such control is limited.
- Market Structure Influence: If the user base is mostly retail, B-Book profits remain stable. If whales or quantitative traders dominate, one-sided markets may “reverse kill” the exchange.

3.7 Summary
Ranging markets are the most favorable for B-Book exchanges, as retail traders’ high leverage and cognitive biases lead to repeated liquidations. Trending markets, on the other hand, pose existential risks—if retail traders collectively align with the trend, liquidity reserves can be depleted rapidly.

Mathematically, B-Book expected profits are positive in ranging conditions, but may turn sharply negative in trending conditions. This asymmetry of profits and losses is the greatest challenge of the B-Book model—and a critical entry point for arbitrageurs.

Chapter 2: Arbitrage Opportunities and Exploitable Loopholes in B-Book Exchanges

2.1 Introduction
Arbitrage has long been regarded in financial markets as an “efficiency correction mechanism.” In traditional markets, arbitrage often arises from pricing discrepancies (e.g., spot–futures basis, cross-market exchange rate differences). However, in B-Book Exchanges, arbitrage opportunities typically stem from technical loopholes and structural delays.
This chapter analyzes six major categories of arbitrage methods and exploitable loopholes, extending into mathematical modeling, case simulations, and approaches to evade risk controls.

2.2 Latency Arbitrage
Exploiting price update delays on B-Book exchanges while hedging on major exchanges.
Multi-layer latency chains: Not just “major exchange → B-Book,” but also chaining multiple weaker B-Book exchanges with longer delays.
Order book snapshot delays: Some B-Book exchanges have WebSocket feeds that are not synchronized with their matching engines. By cross-checking different API routes (REST vs. WebSocket), discrepancies can serve as signals.
Principle:
B-Book exchanges often copy prices from major exchanges but with delays (ranging from milliseconds to seconds). Exploiting this, you can open a position on the B-Book at stale prices, then hedge on the major exchange once the B-Book price catches up.
Advanced Techniques:
Multi-layer latency chains (e.g., Binance → Small Exchange A → Smaller Exchange B).
Cross-checking WebSocket vs. REST order books to detect desynchronization.
Implementation Steps:
Build a low-latency monitoring module: monitor ticks from Binance/OKX and compare with a B-Book exchange simultaneously.
When the major exchange price updates but the B-Book has not yet refreshed → enter the B-Book at stale price, hedge on the major exchange.
Profit comes from the B-Book following the updated price.
Risk Control Evasion:
Avoid large single trades; otherwise accounts may be flagged as arbitrage.
Randomize order intervals (not every delay should trigger an order).
Distribute trades across multiple accounts to prevent an account showing abnormally high win rates.
Supplementary Analysis:
Mathematical condition:
Π=(ΔP−C)×Q,with condition ΔP>C
Case Simulation: ETH delay of 800ms, major exchange price rises +0.4%, arbitrageur enters on small exchange, profit ≈ 0.35%.
Risk Response: Randomize order timings instead of always exploiting exact delay points.

2.3 Extreme Order Execution Loophole
Some systems execute orders based solely on price triggers, ignoring actual order book depth.
Cross-market impact: In certain small exchanges, different trading pairs share the same matching engine. An extreme order in one pair can trigger anomalies in related pairs.
Multiple IOC Splits: Instead of one massive order, submit multiple IOC (Immediate-Or-Cancel) small extreme orders to simulate liquidity and amplify the anomaly.
Principle:
Certain B-Book exchanges only check whether a price threshold is reached, not whether there is sufficient depth. This allows traders to “punch through” levels using extreme IOC orders.
Advanced Techniques:
Cross-market exploitation in shared matching engines (e.g., ETH/USDT vs. BTC/USDT).
Multi-split IOC sequences to trick the system into recognizing fake liquidity.
Combine with maker GTC hedging on major exchanges.
Implementation Steps:
Test with extreme IOC orders far from market price (e.g., ±5%).
If executed → confirms “price-trigger only” logic.
Hedge positions on a major exchange.
Risk Control Evasion:
Limit the frequency and size of exploit use.
Rotate accounts to avoid long-term flagging.
Supplementary Analysis:
Math condition: If true depth = 100 USDT, but IOC = 10,000 USDT → normal execution ≤100 USDT; loophole execution fills full 10,000 USDT.
Case: Small exchange LTC/USDT filled at 49.0; hedge on Binance, arbitrage succeeded.
Risk Mitigation: Multiple small IOC instead of one big order to avoid retroactive cancellation.

2.4 Risk-Control Evasion via Loss Fabrication
Traders deliberately create artificial losses to be classified as “losers” (kept in B-Book), avoiding migration to A-Book.
For example: forcing small losses to reduce win rate <20%, or triggering liquidations to skew PnL history.
Supplementary Analysis:
Motivation: Avoid being switched to A-Book.
Strategy: Intentionally incur losses early → keep win rate <30% → later execute large arbitrage trades.
Mathematical Model:
WinRate=Wins/(Wins+Losses) <30%

2.5 Promotional Capital Arbitrage
Exploiting bonuses or fee vouchers for leveraged high-volatility trades.
Cross-platform bonus cycling: Exploit timing gaps between promotional campaigns across exchanges.
Leverage stacking: Since activity requirements often use notional trading volume, high leverage enables rapid completion with minimal risk.
Principle:
B-Book exchanges frequently offer campaigns like deposit bonuses, trading rebates, or fee vouchers. These can be gamed for arbitrage.
Advanced Techniques:
Use the same principal capital across multiple exchanges’ promotional campaigns.
Apply high leverage (e.g., 100x) to reach notional trading targets quickly.
Implementation Steps:
Deposit 1,000 USDT, open 100x leverage small trades.
Accumulate millions in notional volume rapidly → collect rewards.
Rotate the same capital across multiple platforms.
Risk Control Evasion:
Mix bonus-driven trades with real trades to mask behavior.
Spread activity across multiple accounts to reduce flagging risk.
Supplementary Analysis:
Math condition:
Capital=Vreq/Leverage
Case: Requirement = 1,000,000 USDT volume. With 100x leverage, only 1,000 USDT needed → extremely low actual risk.

2.6 Reverse Arbitrage via Market Microstructure
Monitoring internal matching patterns and exploiting them externally.
Single-market observation: Identify timing of delayed liquidity replenishment after large orders.
Liquidity lock: Place orders that lock liquidity in the B-Book, then move prices on major exchanges before B-Book reacts.
Principle:
Some small exchanges exhibit delayed responses or predictable liquidity replenishment patterns. Exploit these by positioning first on major exchanges, then arbitraging the delay.
Advanced Techniques:
Log latency of replenishment after large trades.
Lock liquidity with pending orders before triggering external moves.
Implementation Steps:
Monitor order book depth long-term.
Move prices on a major exchange.
Arbitrage against the lagging small exchange.
Risk Control Evasion:
Randomize behavior, avoid repeating in the same trading pair.
Limit order sizes to avoid overt manipulation.
Supplementary Analysis:
Math condition:
Price_main(t) ≠ Price_bb(t+Δt_b)。
Case: Small exchange liquidity replenishment delay = 2s. Arbitrageur enters major exchange first, then offsets on the small exchange.

2.7 High-Leverage Hedged Wash Trading
Using dual accounts or partners to farm bonuses and rebates while neutralizing risk.
Task-based arbitrage: Complete missions (e.g., trade counts, holding time) through self-matching.
Internal cooperation networks: Spread trades across accounts to lower detection probability.
Principle:
Use bonus funds or high leverage in paired trades to farm rebates with little/no risk.
Advanced Techniques:
Paired accounts open mirrored positions (long/short).
One account takes the bonus loss, the other captures the rebate.
Implementation Steps:
Account A uses 1,000 USDT bonus to go long 50x.
Account B shorts 50x simultaneously.
Account A liquidates (bonus lost), Account B breaks even but earns rebates.
Risk Control Evasion:
Rotate counterparties regularly.
Operate across multiple IPs/devices to reduce detection.
Supplementary Analysis:
Case: Account A’s 1,000 USDT bonus long 50x liquidates; Account B’s short nets out and collects rebate → risk-free arbitrage.
Mitigation: Use distributed accounts to avoid linking by exchanges.

2.8 Conclusion
The six major arbitrage strategies share common features:
They rely on internal mechanisms or structural delays rather than pure price discrepancies.
They generate short-term high returns, but sustainability depends on the evolution of risk control systems.
3. They are highly replicable and can be applied across multiple small B-Book exchanges.

Chapter 3 – Profit and Loss Characteristics of B-Book Exchanges in Ranging vs. Trending Markets

3.1 Introduction
Market structure has a decisive influence on the profit-and-loss characteristics of B-Book exchanges. Traditional financial research shows that the profitability of a dealer-style model depends heavily on player behavior and market conditions. In ranging markets, retail traders frequently trigger stop-losses and use high leverage, leading to consistent losses and stable exchange profits. In trending markets, however, if retail traders unexpectedly align with the trend, the exchange faces substantial risks.

3.2 Advantages and Risks in Ranging vs. Trending Markets
Advantages of Ranging Markets for B-Book Exchanges:
1. High leverage increases the probability of stop-outs, creating long–short “double-kills.”
2. Frequent user entries and exits generate more fee income.
3. Nearly 99% of users fall into repeated stop-loss and re-entry cycles due to cognitive biases.

Risks of Trending Markets for B-Book Exchanges:
1. Retail traders unexpectedly align with the correct trend, especially under high leverage, amplifying profits.
2. Risk-control and hedging delays—particularly in rapid markets—make it impossible to offset losses in time, as spreads widen sharply across all venues.
3. Reverse stop-loss chains reinforce trend continuation.
4. The inability to exploit oscillations means the exchange loses its “two-sided harvesting” advantage.

3.3 Supplementary Analysis: Ranging Markets
3.3.1 Mathematical Model
Assume the market is ranging, with price fluctuation ΔP within [−x, x], where x is typically smaller than the liquidation threshold for retail traders.
Let:
- λ = user leverage multiple
- M = user margin
- L = 1/λ = liquidation threshold percentage
If |ΔP| ≥ L, the user is liquidated, losing M.
Because prices oscillate back and forth in ranges, most traders hit stop-losses within short intervals, creating long–short double-kills.
The exchange’s expected profit:
E[Π_bb] = Σ P(liquidation_i) ⋅ M_i + Σ Fees_i
3.3.2 Behavioral Finance Notes
- Retail traders’ “loss aversion” and “chasing gains/cutting losses” behaviors are magnified in ranging markets.
- With no clear trend, traders re-enter frequently, raising the exchange’s harvesting efficiency.
3.3.3 Case Study
- In 2023, a small exchange’s BTC/USDT contract ranged between $27,000–$27,800 for a week.
- Retail traders widely used 50x–100x leverage (liquidation thresholds <2%).
- Since the market fluctuated more than 3%, nearly all long and short positions were wiped out. The exchange netted over $5 million USDT in profit from this single product within one week.
3.4 Supplementary Analysis: Trending Markets
3.4.1 Mathematical Model
Let ΔP represent the one-sided market movement (e.g., price increase).
If retail traders are predominantly long, the exchange’s losses are:
Loss_bb = Σ (ΔP ⋅ λ_i ⋅ M_i) − Σ Fees_i
Where:
- λ_i = leverage multiple
- M_i = margin
When retail traders concentrate on the correct trend, losses scale as linear accumulation + leverage amplification, potentially overwhelming the exchange’s liquidity reserves in a short period.
3.4.2 Risk Transmission
Hedging Delay Risk: During rapid moves, a B-Book exchange attempting to hedge through A-Book routing may fail due to latency, resulting in overexposure.
Stop-Loss Chain Effect: As one group of users takes profit, the opposite group is forced to stop out, reinforcing the trend and worsening exchange losses.
3.4.3 Case Studies
- In May 2021, during a crypto market crash, one B-Book exchange failed to hedge in time. As a result, 80% of users with short positions profited massively, creating a $200 million USDT funding gap. The exchange was forced to halt withdrawals temporarily.
- Similar cases exist in Forex markets. For example, during the 2015 Swiss Franc “black swan” event, several B-Book Forex brokers went bankrupt when retail traders were collectively positioned correctly.

3.5 Comparing P&L Structures: Ranging vs. Trending
Ranging Markets → Retail traders suffer long–short double-kills and churn trades frequently. Exchanges profit from liquidations + fees, with minimal risk unless whales manipulate the market.

- Trending Markets → Retail traders may profit massively, and transaction fees cannot offset losses. Sustained one-sided price moves cause large-scale overexposure and liquidity shortages.
Formally:
E[Π_range] = High-Leverage Liquidation Profits + Fee Income
E[Π_trend] = −Retail Profits + Fee Income
Since retail profits usually far exceed fee income, exchanges suffer severe losses in strong trends.

3.6 Extensions: Market Control and Structure
- Manipulation Space: In ranging markets, exchanges can further harvest users via stop-hunts (“wicks”) or by adjusting matching logic. In trending markets, however, such control is limited.
- Market Structure Influence: If the user base is mostly retail, B-Book profits remain stable. If whales or quantitative traders dominate, one-sided markets may “reverse kill” the exchange.

3.7 Summary
Ranging markets are the most favorable for B-Book exchanges, as retail traders’ high leverage and cognitive biases lead to repeated liquidations. Trending markets, on the other hand, pose existential risks—if retail traders collectively align with the trend, liquidity reserves can be depleted rapidly.
Mathematically, B-Book expected profits are positive in ranging conditions, but may turn sharply negative in trending conditions. This asymmetry of profits and losses is the greatest challenge of the B-Book model—and a critical entry point for arbitrageurs.

Chapter 4 – Risk Control Rules and Enforcement Mechanisms of B-Book Exchanges
4.1 Introduction
The core revenue model of B-Book exchanges is derived from user losses and transaction fees. While this “dealer model” can be profitable most of the time, it also carries significant risks—particularly when users collectively take the correct side of a strong trending market. To mitigate these risks, B-Book exchanges generally implement a set of risk-control rules and enforcement measures, applied at the system level, account level, and trade level.
This chapter retains the original content and adds supplementary analyses under each rule, including:
Technical implementation methods
Mathematical or statistical models
Real-world case studies
Possible countermeasures for arbitrageurs

4.2 Account and Fund Flow Risk Control
(1) Multi-Account Association Detection
Principle: Correlation via IP, device fingerprints, KYC data, wallet address links.
Trigger: Multiple accounts opening the same position direction in a short time or acting as direct counterparties (e.g., Account A places an order, Account B takes it, repeatedly). Such activity increases anomaly scores, such as “arbitrage/wash trading,” and triggers manual review.
Impact: Arbitrage teams risk position freezes or bonus clawbacks.
Supplementary Analysis:
Technical Logic: Exchanges often build a behavioral similarity matrix, using IP, device fingerprints, order timestamps, counterparties, etc., to measure correlation. High similarity = flagged as “team” or “linked accounts.”
Model Principle: Multi-account wash trading often produces clustered trades around the mid-price and abnormal concentration. A “correlation score” >0.8 (out of 1.0) usually enters manual audit.
Case Study: A 3-person team executed 50 self-matched trades in 1 second. Despite small order sizes, the system flagged them as wash trading, and all accounts were frozen.
Countermeasures:
Use residential IP pools, not cloud VPS.
Randomize device fingerprints with anti-fingerprint browsers + VMs.
Add random order delays (50ms–3s).
Mix in external counterparties to avoid pure self-matching.

(2) Internal Fund Flow Monitoring
Principle: Detects frequent transfers or withdrawals to the same address.
Trigger: High-frequency small transfers or circular flows.
Impact: Funds frozen or flagged as high risk.
Supplementary Analysis:
Technical Logic: Combines on-chain tracking with internal transfer stats, checking for closed loops.
Model Principle: Graph analysis identifies tightly connected transfer paths (e.g., triangular loops).
Case Study: A 5-account team recycled funds back into the same cold wallet, flagged as “money laundering.”
Countermeasures:
Multi-hop transfers (2–3 wallets), cross-chain swaps.
Delay returns (2–5 days, not seconds).
Mix with real spot trading for natural-looking flow.

(3) Fund Flow Speed Limits
Principle: Restricts abnormal withdrawals or repeated deposits/withdrawals.
Trigger: Withdrawal volumes inconsistent with trading volumes.
Impact: Slows capital turnover.
Supplementary Analysis:
Technical Logic: Compares withdrawal-to-trading ratios.
Model Principle: Standard deviation detection; e.g., daily withdrawal 10x above 7-day mean (Z > 3σ).
Case Study: A team withdrew 95% of its account balance in hours, blocked 48h.
Countermeasures:
Withdraw ≤30% of average daily balance.
Rotate withdrawals across accounts.
Match withdrawals with trading activity.
📌 Summary: In account/fund flow controls, the key is avoiding detection as a clustered group or abnormal flow pattern. Randomization, obfuscation, and distribution help extend survival.

4.3 Trading Behavior Risk Control
(4) Abnormal Order Behavior
Principle: Monitoring for high-frequency order placement/cancellation and extreme price orders.
Trigger: Excessively frequent order placement and cancellation, or placing orders at prices significantly deviating from the market.
Impact: Easily flagged as wash trading or market manipulation.
Supplementary Analysis:
Technical Logic: The exchange’s matching engine calculates the order duration distribution for each account. If the average order duration is less than 100ms, or if the cancellation rate exceeds 90%, it is flagged as “abnormal ordering.”
Model Principle: A Poisson process combined with skewness detection is used to evaluate whether the order frequency exceeds the limits of normal human operation.
Case Study: An arbitrage team used an API to place and cancel 500 orders within 1 second; as a result, their API access was restricted.
Countermeasures:
Set a minimum order duration threshold (e.g., ≥1 second) to avoid being identified as pure wash trading.
Keep the cancellation rate ≤70%, while mixing in some genuine trades.
For extreme price orders, use a step-by-step approach that gradually moves closer to the target price, instead of submitting a single, unreasonable order.

(5)Wash Trading Detection
Principle: Identifying self-trades or team-based matched trades through counterparty analysis.
Trigger: Excessive concentration of counterparties or trade prices consistently deviating from reasonable ranges.
Impact: Classified as wash trading, leading to revoked rewards or even frozen accounts.
Supplementary Analysis:
Technical Logic: The system tracks the distribution of trading counterparties. For example, if 80% of Account A’s trades are with Account B, the accounts are flagged as associated wash trading.
Mathematical Model: The Herfindahl-Hirschman Index (HHI) is used to measure counterparty concentration. If HHI > 0.5, this indicates excessive reliance on a single counterparty and a high level of risk.
Case Study: A two-person team engaged in wash trading to earn fee rebates; eventually, the platform canceled three months’ worth of their rebates.
Countermeasures:
Increase the proportion of external counterparties; ensure at least 50% of trades are executed with random counterparties.
Use random delays and cross-trading across multiple accounts to avoid fixed trading patterns.
Avoid wash trading in illiquid tokens; instead, blend activities into high-liquidity markets.

(6) Extreme Leverage Usage
Principle: Frequent use of the maximum leverage multiplier.
Trigger: Multiple full-leverage positions with extremely short holding periods.
Impact: Flagged as a high-risk account, which may trigger additional margin requirements.
Supplementary Analysis:
Technical Logic: The system tracks the leverage distribution of each account. If usage is heavily concentrated in the 50x–125x range and the average holding period is less than 10 seconds, the account is categorized into the “abnormal group.”
Model Principle: Extreme leverage detection commonly employs K-means clustering to separate normal traders from extreme-risk traders.
Case Study: An arbitrage team repeatedly used 125x leverage and closed positions within just 1 second. Due to excessively high risk markers, the account was forcibly restricted to a maximum of 10x leverage.
Countermeasures:
Adjust leverage dynamically—for example, occasionally use 5x or 10x to avoid consistently operating at maximum leverage.
Maintain an average holding period ≥ 1 minute to create a “real trading” footprint.
For arbitrage purposes, distribute trades across multiple accounts instead of concentrating them in a single account.

(7) Abnormal Liquidation Rate
Principle: Detecting whether the number of liquidations on an account is abnormally high.
Trigger: Multiple liquidations or forced closures within a short period.
Impact: The account is flagged as a “high-risk player” or “manipulated account.”
Supplementary Analysis:
Technical Logic: The platform monitors the liquidation rate, calculated as:
Liquidation Rate=Number of Liquidations/Number of Open Positions
If the liquidation rate exceeds 20%, the account is suspected of being an “artificial test account.”
Model Principle: Logistic regression models are often used to determine whether an account belongs to the “irrational high-risk group.”
Case Study: An arbitrage team deliberately triggered frequent liquidations to create the illusion of “newbie losses” and trick the platform into granting benefits. However, because their liquidation rate was too high, their accounts were flagged.
Countermeasures:
Keep the liquidation rate below 10%, using early closures to avoid forced liquidation.
Occasionally generate small profitable trades to lower the overall liquidation ratio.
Distribute extreme positions across multiple accounts to prevent abnormal risk concentration on a single account.

(8) Abnormal Profitability Detection
Principle: Profit rates or win rates that are excessively stable and unusually high.
Trigger: Sustained high win rates (>80%) or single trades generating extreme profits.
Impact: The account is suspected of using cheating tools or having access to insider information.
Supplementary Analysis:
Technical Logic: The platform tracks the profit-and-loss curve of each account. If it appears overly smooth with almost no drawdowns, the account is flagged under the “mechanized arbitrage” list.
Model Principle: Metrics such as the Sharpe Ratio and Sortino Ratio are used to measure the quality of returns. If Sharpe > 5, it is generally considered unrealistic.
Case Study: A high-frequency arbitrage team maintained a 95% win rate, with very small profits per trade. However, due to their overly stable performance, they were suspected of “algorithmic manipulation combined with latency arbitrage.”
Countermeasures:
Deliberately create “fake losses”, such as maintaining a win rate between 65–75%.
Occasionally execute non-optimized trades to avoid an overly smooth equity curve.
Keep monthly returns within a reasonable range (10–20%) to avoid raising suspicion with extraordinary results.
📌 Summary: For trading behavior control, the key is randomness. Orders, leverage, win rates must mimic humans, not algorithms.

4.4 A/B-Book Routing
(9)Profit and Loss Tagging Mechanism
Principle: Users are automatically classified based on their profit and loss records. Long-term losing traders remain in the B-Book, while consistently profitable traders are migrated to the A-Book for external hedging.
Trigger: Consecutive profits over a defined period, or when risk models indicate elevated risk.
Impact: Profitable accounts are routed into the A-Book, where the exchange hedges against them in external markets, thereby restricting arbitrage opportunities.
Supplementary Analysis:
Technical Logic: The platform calculates the account’s rolling PnL%. For example, if 30-day returns > +10% and Sharpe ratio > 2, the account is flagged as a “high-risk profitable trader.”
Model Principle: Clustering models are used to classify users into “loser group (B-Book)” vs. “winner group (A-Book).”
Case Study: An arbitrage team achieved an 80% win rate and a 25% return in 30 days, and their accounts were automatically migrated to A-Book, making their latency arbitrage strategy ineffective.
Countermeasures:
Avoid overly smooth equity curves; introduce “fake losses” to lower Sharpe ratios.
Keep monthly returns ≤ 10% to avoid accumulating profits too quickly.
Establish a multi-account profit matrix to spread arbitrage gains across multiple accounts, reducing the chance of a single account being flagged.

(10)Behavioral Pattern Recognition
Principle: Detecting obvious mechanized arbitrage traits, such as always executing at optimal price points.
Trigger: Unnaturally consistent order latency or highly fixed position-holding patterns.
Impact: Accounts are identified as high-frequency arbitrage traders and are either migrated to the A-Book or subject to traffic throttling.
Supplementary Analysis:
Technical Logic: By analyzing the latency distribution, if an account’s order latency always falls in the 3–5ms range, it is assumed to be API-driven algorithmic trading.
Model Principle: Markov chains + entropy indicators are used to measure behavioral randomness. If entropy < 0.3, the account is deemed excessively systematic.
Case Study: An arbitrage bot consistently executed at exactly 2ms latency, capturing 100% of Binance’s price propagation spreads, and was ultimately throttled and shifted to the A-Book.
Countermeasures:
Introduce randomized delays (e.g., 50–300ms) to avoid overly stable patterns.
Diversify holding periods instead of always closing positions in exactly 10 seconds.
Mix in non-arbitrage trades to add human-like randomness.

(11)Risk Tiering System
Principle: Assigning different routing tiers based on trading risk. For example, low-risk traders remain in the B-Book, while high-risk profitable traders are shifted into the A-Book.
Trigger: Excessive leverage, overly frequent order placement, or abnormally high win rates.
Impact: Users in different tiers experience differentiated trading conditions, reducing the effectiveness of arbitrage strategies.
Supplementary Analysis:
Technical Logic: The platform constructs a Risk Scoring Card, such as:
Leverage Score (higher leverage = higher risk)
Profitability Score (sustained gains = higher risk)
Behavioral Score (frequent order placement/cancellations = higher risk)
A final total score determines whether the account remains in B-Book or is shifted to A-Book.
Model Principle: Often implemented using a weighted linear model, e.g.:
Total Score=0.4×Leverage+0.3×Profitability+0.3×Behavior
Case Study: An account that consistently used 50x leverage accumulated a high risk score, was forcibly reduced in leverage, and shifted to the A-Book.
Countermeasures:
Dynamically switch leverage (e.g., 5x–20x at random) to avoid consistently high multiples.
Keep win rates in the 60–70% range to avoid suspicion.
Avoid concentrated bursts of frequent orders to reduce behavioral risk scores.

(12) Latency Arbitrage Detection
Principle: Comparing price propagation delays across markets to detect traders exploiting latency gaps.
Trigger: An account consistently executes within the first millisecond of a price move, always in the correct direction.
Impact: Accounts are flagged as latency arbitrage bots and either shifted to A-Book or throttled.
Supplementary Analysis:
Technical Logic: The system compares order matching times against external market price movements. If an account consistently executes within the 1ms delay window, it is flagged as “latency arbitrage.”
Model Principle: Granger causality tests are used to determine whether order placement is excessively dependent on external price transmissions.
Case Study: An arbitrage team exploited Binance’s price propagation delay at a smaller exchange, achieving a 95% win rate. Eventually, their accounts were moved into a dedicated “latency pool,” eliminating their arbitrage opportunity.
Countermeasures:
Add randomized execution delays instead of always trading at the very first millisecond.
Occasionally “miss opportunities” to avoid appearing too perfect.
Deliberately introduce small losses at times to maintain human-like trading behavior.

📌 Summary: A/B routing removes consistent winners. Counter = randomness + noise.

4.5 Risk Events and Interventions
(13) Post-Trade Adjustment / Order Cancellation
Principle: If a trade execution price is deemed an “obvious error,” the transaction can be retroactively canceled.
Trigger: System bugs or thin-order-book mis-executions.
Impact: Arbitrage profits may be erased instantly.
Supplementary Analysis:
Technical Logic: Exchanges define an Erroneous Trade Boundary, e.g., trades executed ±5% beyond fair market price may be invalidated.
Model Principle: Use VWAP (Volume-Weighted Average Price) plus standard deviation thresholds. For example, if the execution price deviates from VWAP by more than 3σ, it is considered invalid.
Case Study: An arbitrageur swept a thin-order-book gap at a small exchange, capturing a 10% spread. Minutes later, the exchange announced “abnormal pricing, trade canceled,” wiping out all profits.
Countermeasures:
Avoid capturing prices that are too far from fair value, e.g., >3% away from major exchange averages.
Use split orders instead of sweeping the entire book at once, reducing the abnormal trade signature.
If cancellation occurs, preserve trade snapshots (depth charts, order book records) as evidence for disputes.

(14)Profit Withdrawal Freezing Period
Principle: Part of profits must be held for a specified number of days before withdrawal.
Trigger: Promotional accounts or accounts trading with bonus funds.
Impact: Slower capital turnover and delayed cash flow.
Supplementary Analysis:
Technical Logic: For accounts that received bonuses, rebates, or fee vouchers, platforms often restrict profit withdrawals for N days.
Model Principle: Funds are separated into principal and bonus-based profits, with withdrawal subject to an unlocking schedule.
Case Study: A team used bonus funds for leveraged arbitrage, earning 100,000 USDT short-term. They were then told “profits must be locked for 30 days,” and subsequent market volatility cut the value in half.
Countermeasures:
Use clean principal accounts for main arbitrage, avoiding linkage to promotional funds.
If bonus arbitrage is unavoidable, adopt a quick in-and-out strategy, converting bonus-based profits into “normal profits” before lock-up applies.
Diversify across multiple accounts—one dedicated to promotions, others for standard trading.

(15)Account Risk Tier Monitoring
Principle: High-risk accounts are placed into special monitoring queues.
Trigger: Multiple risk signals triggered simultaneously.
Impact: Trading speed slows, withdrawals face increased review.
Supplementary Analysis:
Technical Logic: Platforms calculate a risk grade for each account, assigning them to different pools. Low-risk pools face routine monitoring, while high-risk pools require manual review.
Model Principle: Multi-factor scoring, e.g., abnormal IP + high leverage + high win rate + large withdrawals, when triggered together, mark an account as “high risk.”
Case Study: An arbitrage team repeatedly hit latency arbitrage opportunities and made abnormal withdrawals within three days. Their accounts were moved to the high-risk monitoring pool, with each withdrawal requiring manual review and delays of at least 24 hours.
Countermeasures:
Reduce the probability of multiple risk signals overlapping (e.g., avoid combining high leverage + synchronized orders + large withdrawals).
Apply a matrix account strategy, distributing different risk behaviors across multiple accounts.
Maintain camouflage: occasionally perform small, non-profitable trades to reduce abnormal indicators.

(16)Backend Forced Liquidation
Principle: When the platform determines that an account is engaged in abnormal trading or poses excessive risk, it can forcibly liquidate positions.
Trigger: Risk warnings or technical anomalies.
Impact: Positions are closed before stop-loss levels are reached, locking in immediate losses.
Supplementary Analysis:
Technical Logic: Once an account is flagged as a “high-risk arbitrageur,” the platform’s Risk Engine can forcibly liquidate its positions under the pretext of “system maintenance” or “protecting market liquidity.”
Model Principle: If the system detects large directional concentration combined with the inability to hedge externally, forced liquidation may be activated.
Case Study: An arbitrage team opened large positions for latency arbitrage on a small exchange. The platform forcibly closed their positions—even before hitting liquidation price—resulting in a direct 30% loss.
Countermeasures:
Avoid concentrating oversized positions in a single account.
Use laddered positions and multi-account distribution to spread risk.
Once targeted interventions are detected, immediately cease the strategy and shift operations to another exchange.

📌 Summary: These are the last defense for exchanges; once triggered, profits vanish. Arbitrageurs must withdraw quickly.

4.6 System-Level Restrictions
(17) API Rate Limits
Principle: Restrict the number of API requests (orders/cancellations) per second.
Trigger: Exceeding the limit leads to temporary IP/Key suspension.
Impact: Latency arbitrage and cross-exchange arbitrage frequency is curtailed.
Supplementary Analysis:
Technical Logic: Different exchanges set specific rate limits, e.g.:
Binance: 50 requests/sec
Bitget: 20 requests/sec
LBank: 10 requests/sec
Model Principle: The system uses counters to track API requests per unit of time. Exceeding the threshold returns error code 429 (Too Many Requests) or results in IP/Key suspension.
Case Study: An arbitrage team ran multiple latency strategies simultaneously on Bybit, exceeding the limit. Their API key was disabled for 1 hour, interrupting all strategies.
Countermeasures:
Use multiple keys + multiple IPs for traffic distribution (e.g., 5 keys at 10 requests/sec each, maintaining high throughput).
Introduce randomized delays to avoid request bursts.
Use WebSocket subscriptions + batch orders to minimize REST API requests.

(18) Maximum Single Order Size Limit
Principle: Prevent large orders from manipulating the market.
Trigger: Orders exceeding the per-trade maximum are automatically rejected.
Impact: Arbitrageurs cannot sweep order book depth in one go.
Supplementary Analysis:
Technical Logic: Platforms impose maximum order caps, e.g.:
BTC/USDT: max 2 million USDT
Small-cap pairs: max 50,000 USDT
Model Principle: Pre-trade risk checks reject oversized orders with an “Order Exceed Limit” error.
Case Study: An arbitrageur tried to sweep an entire small-cap pair’s sell wall but was rejected for exceeding the cap, missing the opportunity.
Countermeasures:
Use split-order strategies, breaking large orders into multiple small IOC orders.
Apply time-slicing, distributing orders over a few hundred milliseconds to mimic organic trading.
Employ multi-account coordination to bypass per-account caps.

(19) Matching Engine Frequency Limits
Principle: Limit the number of order matches an engine processes per account.
Trigger: Excessively frequent orders.
Impact: High-frequency strategies are throttled (“slowed down”).
Supplementary Analysis:
Technical Logic: Matching engines throttle accounts, e.g.:
Maximum 100 matches per second per account.
Excess orders are delayed or dropped.
Model Principle: The system uses priority queues, favoring low-frequency traders while relegating high-frequency accounts to secondary queues.
Case Study: A high-frequency team submitted 500 orders per second, but only 20% were processed; the rest were delayed hundreds of milliseconds, erasing their edge.
Countermeasures:
Reduce order quantity while improving per-order efficiency.
Use predictive matching models to reduce ineffective orders.
Distribute load across multiple accounts to balance engine pressure.

(20)Network Latency and Throttling
Principle: Control request latency for different users via gateway servers.
Trigger: High-frequency users or high-risk accounts.
Impact: Arbitrage order execution slows down, eliminating latency advantage.
Supplementary Analysis:
Technical Logic: At the API Gateway layer, the platform may introduce artificial latency, e.g., 20–50ms for regular users, while VIP or MM (market maker) accounts receive low-latency channels (1–5ms).
Model Principle: Uses traffic shaping, assigning different Quality of Service (QoS) levels based on account type.
Case Study: An arbitrage bot tested on LBank with API latency fixed at 40ms, while competitor MM accounts had only 3ms. Result: zero chance of winning latency arbitrage.
Countermeasures:
Strive to obtain VIP/MM account status to gain access to low-latency channels.
Deploy co-location servers near the exchange’s data centers.
Add predictive models at the strategy layer to reduce reliance on millisecond-level latency.

📌 Summary: System restrictions aim to weaken HFT arbitrage. Core response = distribution, slicing, and prediction.

Chapter 4 – Risk Control Rules and Enforcement Mechanisms of B-Book Exchanges

4.7 Backend Interventions and Post-Trade Adjustments
(21) Profit Adjustment
Principle: The platform may adjust or void abnormal profits.
Trigger: Abnormal pricing or API delays leading to irregular fills.
Impact: Arbitrage gains may be reduced or clawed back.
Supplementary Analysis:
Technical Logic: At end-of-day settlement, exchanges review accounts with high profits. If trades deviate significantly from mainstream market prices, they may be marked as “abnormal.”
Model Principle: Abnormal Deviation Rate =∣Ptrade−Pmarket∣/Pmarket If deviation > 5%, profits may be voided.
Case Study: An arbitrage team captured a 12% mispricing on a small exchange, initially showing $50,000 profit. At settlement, $45,000 was deducted as “abnormal price orders.”
Countermeasures:
Keep matching engine snapshots (e.g., order book screenshots and mainstream exchange comparisons) for appeal.
Avoid over-reliance on extreme mispricings.
Diversify arbitrage sources to reduce exposure from large, suspicious gains.
(22) Order Cancellation Review
Principle: Large or rapid cancellations are subject to manual review.
Trigger: Frequent order placement/cancellation mimicking wash trading.

Impact: Order matching delayed, reducing strategy efficiency.
Supplementary Analysis:
Technical Logic: Platforms monitor Cancel Ratio: Cancel Ratio=Canceled Orders/Total Orders If ratio > 80%, the account enters manual review.
Model Principle: High-frequency arbitrage often relies on “quote probing → instant cancel.” To prevent fake liquidity, exchanges restrict such activity.
Case Study: A market-making team maintained a 90% cancel rate in an illiquid pair. Their account was flagged as “high risk,” with order matching delayed by 500ms, making their strategy useless.
Countermeasures:
Keep cancel ratios below 60% (e.g., place 10 orders → leave 4 active → cancel 6).
Use iceberg/hidden orders to reduce visible cancel frequency.
Place long-tail resting orders as cover, avoiding all-or-nothing cancellations.
(23) Account Freezing
Principle: Accounts flagged for abnormal arbitrage or risky fund flows may be frozen.
Trigger: High win rate + high leverage + frequent withdrawals.
Impact: Funds become inaccessible, sometimes permanently seized.
Supplementary Analysis:

Technical Logic: Once flagged as a “high-risk arbitrage group,” the platform may require enhanced KYC or “proof of profit source.” Failure to comply risks long-term freezing.
Model Principle: System evaluates win rate + ROI + fund velocity. If values exceed 99% of the user base, activity is deemed abnormal.
Case Study: An arbitrage team earned $500,000 in one week and quickly withdrew. Their account was frozen under suspicion of “money laundering/arbitrage.” After negotiation, they recovered only 30%.
Countermeasures:
Avoid “perfect” metrics (e.g., win rate of 99% raises suspicion).
Make fund movements appear natural, avoiding large short-term withdrawals.
Use multi-account, multi-wallet layers to spread risk.
(24) Post-Settlement Profit Recalculation
Principle: Platforms may recalculate PnL after trading hours, citing updated logic.
Trigger: Price source updates or fee recalculations.
Impact: Reported profit may not match final withdrawals.
Supplementary Analysis:

Technical Logic: Exchanges may claim “price source errors” and adjust PnL based on TWAP/VWAP.
Model Principle: Final PnL=Current PnL+Δadjustment whereΔ includes fee adjustments or retroactive price corrections.
Case Study: A team reported $100,000 daily profit, but the next day saw only $80,000 after the platform “recalculated” due to price feed anomalies.
Countermeasures:
Keep exchange announcements, candlestick data, and mainstream exchange references for appeals.
Avoid large arbitrage positions on smaller exchanges, where retroactive adjustments are more common.
Hedge by holding offsetting positions on mainstream exchanges to mitigate post-settlement risks.
📌 Summary: Backend adjustments are opaque and discretionary. Arbitrageurs must assume clawbacks and hedge accordingly.

Chapter 5 –Why Betting Exchanges See Maker Orders as Their Enemy
(with Mathematical Models & Control-Based Arbitrage Principles)

1. Nature of Betting Exchanges: The B-Book Model
Unlike mainstream exchanges (A-Book), which hedge or pass orders externally, betting exchanges (B-Book) internalize risk:

- A-Book: Exchange only earns transaction fees, does not bear PnL.
- B-Book: Exchange becomes the counterparty to user trades.

Thus, exchange profit Π equals the negative of total user PnL:

Π = - Σ P_i

Π: Exchange profit.
P_i: Net PnL of trader i.

👉 Users lose → exchange wins; users win → exchange loses. This is effectively the house model.

2. Taker vs. Maker Order Expectation
Two types of players dominate order flow:

- Takers (market/IOC orders): Impatient, often retail chasing pumps or panic selling. On average negative expectation E[L_T] > 0 (user losses). Good for exchange profits.

- Makers (limit orders): Patient, calculated entries, often arbitrageurs or professionals. On average positive expectation E[L_M] > 0 (user gains). Bad for exchange profits.

Expected exchange profit can be modeled as:

E[Π] = α·E[L_T] - (1-α)·|E[L_M]|

α: Proportion of taker volume.
1-α: Proportion of maker volume.

👉 If maker proportion or maker edge grows too large, E[Π] < 0 and the exchange loses money long-term.

3. Why “Maker Orders Always Fill (100% Fill)”

On mainstream exchanges: Pr(fill) < 1 (depends on depth).
On betting exchanges: To preserve the illusion of a real order book, any limit order must fill if touched:
Pr(fill) = 1 when Price_t ≥ LimitOrder

👉 A maker order is effectively a zero-cost American option:
- If price touches → automatic execution.
- If not touched → no loss.

Thus for professionals, maker orders = risk-free optionality.

4. Maker Arbitrage Mathematical Model
Suppose reference price at major exchange is S_t, and betting exchange lags by Δt. Arbitrageur places a limit order at price L:

E[π] = Pr(S_{t+Δt} ≥ L) · (S_{t+Δt} - L) - C

Pr(S_{t+Δt} ≥ L): Probability the reference price crosses L.
S_{t+Δt} - L: Arbitrage capture if filled.
C: Costs (fees, slippage).

Condition for profitable arbitrage: E[π] > 0.

Since betting exchanges enforce Pr(fill) = 1, these orders provide more stable arbitrage returns.

5. Exchange Risk Exposure Model (with Control-Based Arbitrage)
Exchange exposure to maker orders:

E[Loss_exchange] = Σ Pr(fill_j) · E[π_j]

M: Number of maker orders.
π_j: Expected profit of order j.

Since Pr(fill) = 1, exposure = unavoidable loss.

The most dangerous class: Control-based (manipulative) arbitrageurs:
1. Push price at a major exchange (A-Book).
2. Pre-place maker orders at the betting exchange (B-Book).
3. When price follows, maker orders are guaranteed to fill.
4. Arbitrageur collects profit; betting exchange bears the loss.

Their profit expectation:
E[π_control] = (S_{A,t+Δt} - L) - C

Since manipulators influence S_{A,t+Δt} directly, E[π_control] >> 0.

👉 Control-based arbitrageurs exclusively rely on maker orders, because only makers guarantee conversion of manipulation into forced exchange payouts.
6. Final Conclusion

1. Betting exchanges profit from taker (retail) losses.
2. Maker orders represent professional arbitrage, usually positive EV for traders.
3. Because maker orders always fill, they expose betting exchanges to structural negative EV.
4. Control-based arbitrageurs exclusively use maker orders:
- They manufacture price moves on major exchanges.
- Pre-place maker orders on betting exchanges.
- Betting exchanges must honor the fills, creating guaranteed arbitrage extraction.

👉 Therefore:
The true enemy of betting exchanges is not ordinary retail traders, but maker-based arbitrageurs — especially control-based manipulators. Maker orders convert the exchange’s B-Book model into a guaranteed long-term cash drain.

Thank you for reading until the end.
Always remember one principle: the true enemies of betting exchanges are maker limit orders and limit IOC orders.
In this market, we deliberately create volatility and exploit it through back-to-back arbitrage. In essence, this damages the profitability of the exchange. Exchanges often claim that sudden crashes or pumps hurt retail traders, but the reality is that we exploit these betting exchanges just as ruthlessly as they exploit their users.

We and such exchanges are like ravens spotting carrion — if you don’t take a bite, another raven surely will.

There are far too many betting exchanges in this industry, but only a few are truly worth studying in depth. Most exchanges welcome you opening positions, but they don’t want you to size too big. They fear losing, yet love to gamble. Ultimately, their most lethal weapon is risk control — freezing accounts, delaying your capital flow cycle, or even withholding funds entirely.

I have always encouraged my employees and students to experiment with small-capital arbitrage. However, to generate meaningful short-term volatility in the market, at least 350,000–500,000 USDT is required. For those just entering crypto, such capital is unrealistic — which is why smaller players are better off focusing on the strategy outlined in my earlier paper:
“Latency Arbitrage Strategy Research: A Quantitative Trading Model Based on Dual-Exchange Quote Delays.”
Details on exchange risk-control mechanisms are also fully covered in that article.

This is a niche market that only a few can access. Still, I welcome capable individuals to reach out and exchange ideas. As for collaboration, my principle ...........