8/ Fraud Detection Algorithms

Dash integrates advanced fraud detection algorithms designed to protect users from the increasingly complex and deceptive tactics within the Base ecosystem. By leveraging cutting-edge AI and on-chain data analysis, Dash identifies and mitigates high-risk activities before they can harm users.

1. Smart Contract Analysis

Dash conducts a thorough examination of token smart contracts, identifying vulnerabilities and malicious functionalities.

• Unrenounced Ownership: A key red flag is when developers retain control over the contract, allowing them to modify tokenomics or execute rug pulls.

  • Detection Formula:

    Or={1if ownership functions are active0if ownership is renouncedO_{r} = \begin{cases} 1 & \text{if ownership functions are active} \\ 0 & \text{if ownership is renounced} \end{cases}Or​={10​if ownership functions are activeif ownership is renounced​

    Example: If a contract retains an active minting function, Or=1O_{r} = 1Or​=1, triggering a fraud alert.

• Suspicious Functions: Dash flags contracts containing dangerous functions, such as:

  • Minting Functions: Used to create unlimited tokens.

  • Hidden Transfer Mechanisms: Allow unauthorized wallet transfers.

• Code Complexity Assessment: Uses static analysis tools to detect obfuscated or poorly documented code. For instance, a contract with 80% or more unused functions is flagged.

2. Behavioral Wallet Analysis

Dash employs Graph Neural Networks (GNNs) to analyze wallet interactions and detect coordinated activities.

• Anomaly Detection: Monitors abnormal wallet behaviors, such as:

  • Large, sudden transactions (TsT_{s}Ts​) exceeding a predefined threshold:

    Ts=VtAmT_{s} = \frac{V_{t}}{A_{m}}Ts​=Am​Vt​​

    Where:

    • VtV_{t}Vt​: Transaction volume.

    • AmA_{m}Am​: Average market volume.

    Example: If a token’s average market volume is 10,000 tokens and a transaction involves 500,000 tokens, Ts=50T_{s} = 50Ts​=50, signaling an anomaly.

• Wallet Clustering: Identifies groups of wallets acting in coordination. GNNs map wallet interactions to find:

  • Pump-and-Dump Schemes: Coordinated buys and sells to manipulate prices.

  • Liquidity Drains: Large withdrawals by multiple wallets over short periods.

3. Sentiment-Based Risk Assessment

Dash Natural Language Processing (NLP) models analyze sentiment around tokens by scraping social platforms like Twitter, Telegram, and Discord.

• Hype-to-Engagement Ratio: Flags tokens with disproportionate hype compared to genuine engagement. Formula:

  He=HcErH_{e} = \frac{H_{c}}{E_{r}}He​=Er​Hc​​

  Where:

  • HeH_{e}He​: Hype-to-Engagement Ratio.

  • HcH_{c}Hc​: Hype count (mentions, hashtags).

  • ErE_{r}Er​: Engagement rate (comments, replies).

  Example: A token with 10,000 mentions but only 200 comments has He=50H_{e} = 50He​=50, a sign of artificial hype.

• Toxicity Analysis: Detects aggressive language or bots promoting a token, reducing credibility.

4. Holder Distribution Analysis

Dash examines token holder distributions to identify potential risks.

• Centralization Risk: Tokens with high concentration in a few wallets are flagged. Formula:

  Cr=TwThC_{r} = \frac{T_{w}}{T_{h}}Cr​=Th​Tw​​

  Where:

  • CrC_{r}Cr​: Centralization Ratio.

  • TwT_{w}Tw​: Total tokens held by the top 5 wallets.

  • ThT_{h}Th​: Total circulating supply.

  Example: If the top 5 wallets hold 80% of the supply, Cr=0.8C_{r} = 0.8Cr​=0.8, indicating a high centralization risk.

• Holder Activity: Tracks the activity of token holders:

  • Dormant wallets (inactive for >30 days).

  • Wallets executing large trades during pump events.

5. Dynamic Risk Scoring

Dash assigns a dynamic risk score to each token based on real-time metrics:

• Risk Formula:

  Rs=w1Cr+w2He+w3TsR_{s} = w_{1}C_{r} + w_{2}H_{e} + w_{3}T_{s}Rs​=w1​Cr​+w2​He​+w3​Ts​

  Where:

  • RsR_{s}Rs​: Total risk score.

  • w1,w2,w3w_{1}, w_{2}, w_{3}w1​,w2​,w3​: Weighted factors for centralization, hype, and transaction anomalies.

  Example:

  • Cr=0.7C_{r} = 0.7Cr​=0.7, He=40H_{e} = 40He​=40, Ts=20T_{s} = 20Ts​=20.

  • With weights w1=0.5w_{1} = 0.5w1​=0.5, w2=0.3w_{2} = 0.3w2​=0.3, w3=0.2w_{3} = 0.2w3​=0.2:

    Rs=(0.5)(0.7)+(0.3)(40)+(0.2)(20)=12.35R_{s} = (0.5)(0.7) + (0.3)(40) + (0.2)(20) = 12.35Rs​=(0.5)(0.7)+(0.3)(40)+(0.2)(20)=12.35

    Tokens with Rs>10R_{s} > 10Rs​>10 are flagged as high risk.