Token Analysis¶

Chain Sentinel provides comprehensive token analysis powered by AI and machine learning models.

Overview¶

When you search for a token, Chain Sentinel performs a deep analysis using multiple data sources and AI models to determine if the token is legitimate or a potential scam.

Analysis Components¶

1. Basic Information¶

• Token Address: Unique identifier on Solana blockchain
• Token Name: Human-readable name
• Token Symbol: Trading symbol (e.g., BONK, WIF)
• Creation Date: When the token was deployed
• Creator Wallet: Address that deployed the token

2. Market Data¶

• Current Price: Real-time price in USD
• Market Cap: Total value of all tokens
• 24h Volume: Trading volume in last 24 hours
• Liquidity: Available liquidity in DEX pools
• Holder Count: Number of unique wallet holders

3. On-Chain Metrics¶

Transaction Activity¶

• Total transaction count
• Recent transaction patterns
• Transaction velocity (txs per hour)
• Unusual activity detection

Holder Distribution¶

• Top 10 holder concentration
• Whale wallet detection
• Distribution fairness score
• New vs old holder ratio

Liquidity Analysis¶

• DEX pool liquidity
• Liquidity lock status
• LP token distribution
• Rug pull risk indicators

4. AI Predictions¶

Chain Sentinel uses two advanced AI models:

XGBoost Model¶

• Accuracy: 95.59%
• Features: 20+ on-chain metrics
• Output: SCAM/LEGIT classification + confidence score
• Explainability: SHAP values show which features influenced the decision

GNN v2 Model (GraphSAGE)¶

• Accuracy: 96.06%
• Features: Network graph structure + node features
• Output: SCAM/LEGIT classification + confidence score
• Advantage: Captures relationships between wallets and tokens

5. Risk Indicators¶

Chain Sentinel checks for common scam patterns:

Detection Details¶

Confidence Score¶

The confidence score (0-100%) indicates how certain the AI model is about its prediction:

• 90-100%: Very high confidence
• 75-89%: High confidence
• 60-74%: Moderate confidence
• < 60%: Low confidence (requires manual review)

SHAP Explanations¶

For XGBoost predictions, SHAP (SHapley Additive exPlanations) values show:

• Which features contributed to the SCAM prediction (red bars)
• Which features contributed to the LEGIT prediction (blue bars)
• The magnitude of each feature's impact

Example:

holder_count: +0.15 (more holders = more legit)
top_10_concentration: -0.23 (high concentration = more scam)
liquidity_usd: +0.08 (more liquidity = more legit)

Network Graph¶

The network graph visualizes relationships between:

• Token (center node)
• Creator Wallet (who deployed the token)
• Top Holders (largest token holders)
• Related Tokens (other tokens from same creator)

Graph Features¶

• Node Size: Proportional to wallet balance or token value
• Node Color:
• 🔴 Red = SCAM
• 🟢 Green = LEGIT
• ⚪ Gray = Unknown
• Edge Thickness: Proportional to transaction volume
• Timeline Animation: See how the network evolved over time

Cluster Detection¶

Chain Sentinel identifies wallet clusters (groups of related wallets):

• Scam Rings: Multiple scam tokens from same creator
• Wash Trading: Coordinated buying/selling between related wallets
• Sybil Attacks: One entity controlling many wallets

How to Interpret Results¶

✅ Likely LEGIT Token¶

Prediction: LEGIT
Confidence: 92%
Model: XGBoost

Risk Indicators:
✅ High liquidity ($250K)
✅ Fair distribution (top 10: 28%)
✅ Active community (5,000+ holders)
✅ Verified project

❌ Likely SCAM Token¶

Prediction: SCAM
Confidence: 98%
Model: GNN v2

Risk Indicators:
❌ Honeypot detected
❌ Creator owns 85% of supply
❌ Low liquidity ($500)
❌ Created 2 hours ago
❌ No social media

⚠️ Uncertain (Manual Review Needed)¶

Prediction: SCAM
Confidence: 62%
Model: XGBoost

Risk Indicators:
⚠️ Moderate concentration (top 10: 55%)
⚠️ Low volume ($5K/day)
✅ Decent liquidity ($50K)
✅ 500+ holders

Best Practices¶

Before Buying¶

1. Check Chain Sentinel prediction - Start here
2. Review risk indicators - Look for red flags
3. Examine network graph - Check for scam rings
4. Verify social media - Real project or fake?
5. Test with small amount - Don't YOLO your life savings

Red Flags to Avoid¶

• Confidence > 80% SCAM prediction
• Multiple high-risk indicators
• Creator wallet in known scam cluster
• Honeypot or freeze authority enabled
• Anonymous team with no audit

Green Flags to Look For¶

• Confidence > 80% LEGIT prediction
• Verified project with audit
• Fair token distribution
• Locked liquidity
• Active community and development

API Access¶

You can integrate Chain Sentinel analysis into your own tools:

import requests

# Analyze token
response = requests.get(
    "https://api.chainsentinel.net/tokens/analyze",
    params={"address": "DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263"},
    headers={"X-API-Key": "your_api_key"}
)

data = response.json()
print(f"Prediction: {data['prediction']}")
print(f"Confidence: {data['confidence']}%")

See API Reference for full documentation.

Limitations¶

Support¶

Need help interpreting results?

• 📧 Email: support@chainsentinel.net
• 💬 Telegram: @chainsentinel_net
• 📖 FAQ: Frequently Asked Questions