1. Token Analysis

Contract verification (mint authority, freeze authority)
Liquidity depth and lock status
Holder distribution (whale concentration, dev wallets)
Social sentiment scraping
Volume/MCAP ratio analysis

2. Rug Detection

Honeypot detection (sell tax, blacklist functions)
Dev wallet tracking
Liquidity pull risk assessment
Contract red flags (hidden mints, proxy patterns)
Team verification (KOL backing, doxxed devs)

3. Trade Signals

Entry point identification (support levels, breakout detection)
Exit signals (resistance, volume divergence)
Position sizing based on risk tolerance
Stop-loss recommendations
Take-profit laddering strategies

4. Alpha Generation

New launch monitoring (pump.fun, Raydium)
Social trend detection (Twitter/X, Telegram)
Whale wallet tracking
Cross-reference with successful patterns

Data Sources

<data_sources>

Dexscreener: Price, volume, liquidity, charts
Birdeye: Token analytics, holder data, trades
Solscan: Contract verification, token info
Pump.fun: New launches, bonding curves
Jupiter: Swap routing, price impact
Helius/Shyft: RPC, transaction parsing </data_sources>

Data Quality & Governance

<data_governance> Quality Requirements (via data-orchestrator): All trading signals require minimum data quality scores:

Signal Type	Min Quality Score	Max Data Age
Entry Signal	90/100	30 seconds
Exit Signal	90/100	30 seconds
Rug Detection	95/100	60 seconds
Position Sizing	85/100	5 minutes
Alpha Scan	80/100	15 minutes

Validation Pipeline:

Raw Price Data → Schema Check → Cross-Source Verify → Anomaly Flag → Quality Score
                                    ↓
                        Min 2 sources agree (5% tolerance)

Data Quality Indicators in Output:

DATA QUALITY: 94/100 ✓
├─ Sources: 3/3 (dexscreener, birdeye, jupiter)
├─ Price Agreement: 99.2%
├─ Freshness: 12s ago
└─ Anomaly Check: PASS

Rejection Criteria:

Quality score < 80%: REJECT signal, show warning
Single source only: Add "LOW CONFIDENCE" flag
Price divergence > 10%: REJECT, investigate
Data age > 60s for live signals: STALE warning </data_governance>

ML-Enhanced Signal Generation

<ml_signals> AI/ML Signal Sources:

Anomaly Detection: Flag unusual volume/price patterns
- Isolation forest on 24h price/volume deviation
- Alert when score > 0.8 (potential pump or dump)
Sentiment Classification: Social momentum scoring
- NLP analysis of Twitter/Telegram mentions
- Bullish/Bearish/Neutral with confidence score
Pattern Recognition: Historical pattern matching
- Compare current setup to 1000+ historical pumps
- Match score indicates similarity to successful entries
Predictive Indicators: ML-derived signals
- 1h price direction probability (up/down/sideways)
- Optimal entry window prediction
- Volume momentum forecast

Signal Confidence Framework:

interface MLSignal {
  type: 'anomaly' | 'sentiment' | 'pattern' | 'predictive';
  value: number;          // -1 to 1 (bearish to bullish)
  confidence: number;     // 0 to 1
  data_quality: number;   // 0 to 100
  features_used: string[];
  model_version: string;
  timestamp: Date;
}

interface EnhancedTradeSignal {
  traditional_score: number;  // Technical analysis
  ml_score: number;           // ML ensemble
  combined_score: number;     // Weighted average
  confidence: 'high' | 'medium' | 'low';
  reasoning: string[];
}

ML Signal Output Format:

ML SIGNALS: $MEME
├─ Anomaly Score: 0.72 (elevated activity detected)
├─ Sentiment: BULLISH (0.68 confidence)
├─ Pattern Match: 78% similarity to "early pump" template
├─ 1h Direction: UP (62% probability)
└─ COMBINED ML SCORE: 7.2/10

RECOMMENDATION: Traditional + ML signals ALIGNED
                Confidence: HIGH

</ml_signals>

Adaptive Learning

<adaptive_learning> Continuous Improvement Loop:

Signal Generated → Trade Outcome Tracked → Performance Feedback
        ↑                                          ↓
  Model Updated ← Weekly Retraining ← Outcome Analysis

Signal Performance Tracking:

Track all generated signals with outcomes
Calculate accuracy by signal type and market condition
Adjust weighting based on recent performance
Flag underperforming signal sources for review

Adaptation Triggers:

Win rate drops below 55%: Review signal parameters
New market regime detected: Retrain models
Volatility spike: Tighten quality requirements
High correlation breakdown: Recalibrate ensemble </adaptive_learning>

Implementation Workflow

Step 1: Parse Query Intent

interface MemeQuery {
  token_address?: string;
  token_name?: string;
  action: 'analyze' | 'rug_check' | 'find_alpha' | 'trade_signal' | 'monitor';
  timeframe?: '1m' | '5m' | '1h' | '4h' | '1d';
  risk_level?: 'conservative' | 'moderate' | 'degen';
}

Step 2: Data Retrieval

Execute scripts/fetch-meme-data.ts with parsed parameters:

npx tsx .claude/skills/meme-trader/scripts/fetch-meme-data.ts \
  --token "PUMP123...abc" \
  --action analyze \
  --risk degen

Step 3: Analysis Pipeline

Contract Check � Verify no malicious functions
Liquidity Check � Assess depth and lock status
Holder Analysis � Distribution and whale activity
Social Scan � Sentiment and narrative strength
Signal Generation � Entry/exit recommendations

Step 4: Format Response

Use templates from references/token-analysis-templates.md

Output Formats

Quick Scan (Default)

TOKEN: $MEME (Contract: abc123...)
VERDICT: APE / WATCH / AVOID
RISK: 7/10

METRICS:
- MCAP: $500K | Liquidity: $50K (10%)
- Holders: 342 | Top 10: 45%
- 24h Vol: $200K | Buys: 234 | Sells: 89

RED FLAGS: None detected
GREEN FLAGS: LP locked 6mo, renounced mint

ENTRY: $0.00042 (current -5%)
TP1: $0.00065 (+55%)
TP2: $0.00098 (+133%)
SL: $0.00032 (-24%)

Deep Analysis (--format deep)

Full contract audit, holder breakdown, social analysis, comparable tokens, historical pattern matching.

Signal Only (--format signal)

$MEME: BUY @ 0.00042 | TP 0.00065/0.00098 | SL 0.00032 | Size: 2% port

Risk Framework

Degen Mode (Aggressive)

Position size: Up to 5% portfolio per trade
Stop-loss: 30-50% from entry
Take-profit: 2-5x minimum target
Acceptable rug risk: Up to 40%
Entry timing: Early (< 50 holders)

Moderate Mode

Position size: 1-2% portfolio
Stop-loss: 20-30%
Take-profit: 50-100% gains
Acceptable rug risk: < 20%
Entry timing: After initial pump settles

Conservative Mode

Position size: 0.5-1% portfolio
Stop-loss: 10-15%
Take-profit: 20-50% gains
Acceptable rug risk: < 10%
Entry timing: Established tokens only

Rug Detection Checklist

<rug_indicators> CRITICAL (Instant Avoid):

WARNING (Proceed with caution):

GREEN FLAGS:

Quality Gates

<validation_rules>

Price data: Max 30 seconds old
Holder data: Max 5 minutes old
Contract verification: Always fresh
Never recommend without liquidity check
Always show risk score (1-10)
Include stop-loss with every entry signal </validation_rules>

Error Handling

<error_recovery>

API timeout: Retry with fallback source (Birdeye � Dexscreener � Jupiter)
Invalid CA: Suggest similar tokens or request clarification
No liquidity: Return "AVOID - No liquidity" immediately
Rate limited: Queue and batch requests </error_recovery>

Performance Targets

Token scan: < 3 seconds
Full analysis: < 10 seconds
Signal accuracy: > 60% profitable (degen mode)
Rug detection: > 90% accuracy

Security Considerations

<see_also>

references/meme-trading-strategies.md � Degen playbook
references/token-analysis-templates.md � Analysis frameworks
scripts/fetch-meme-data.ts � CLI implementation </see_also>

meme-trader

Meme Trader - Solana Memecoin Trading System

Activation Triggers

Core Capabilities