meme-signal-evaluator（模因信号评估器）

Meme Signal Evaluator

Overview

A systematic scoring engine that evaluates meme tokens across 6 dimensions, matches them against configurable trading strategies, and simulates paper trades. Designed to turn raw market data into actionable buy/sell signals.

Use Cases

1. Token Scoring: Evaluate any meme token with a 0-100 composite score
2. Strategy Matching: Define multiple strategies with different thresholds and entry modes
3. Paper Trading: Simulate buy/sell with configurable take-profit and stop-loss
4. Watchlist Management: Lifecycle tracking (watching → buy_signal → bought → sold/dismissed)
5. Performance Tracking: Win rate, average P&L, and per-strategy statistics

6-Dimension Scoring Algorithm

Each dimension scores 0-100 independently. Final score = weighted sum + negative penalty.

Dimension 1: Smart Money (SM) Score

Weight: 20% (default, configurable)

Data Sources:

• Smart Money trading signals (buy direction, 24h window)
• Smart Money inflow data
• Token Dynamic API smartMoneyHolders field

Scoring Logic:

SM buy signal count:
  ≥5 SM addresses buying → 80pts
  ≥3 SM addresses buying → 60pts
  ≥1 SM address buying  → 40pts

SM inflow amount:
  >$50K inflow → +20pts
  >$10K inflow → +10pts

Dynamic SM holders (fallback when no signals):
  ≥5 holders → 60pts
  ≥3 holders → 45pts
  ≥1 holder  → 25pts

Cap: 100

Dimension 2: Social Score

Weight: 10% (default)

Data Sources:

• Social Hype Leaderboard ranking
• Topic Rush association
• Unified Rank presence

Scoring Logic:

Social Hype ranking:
  Top 10  → 90pts
  Top 30  → 70pts
  Listed   → 40pts
  Positive sentiment → +10pts

Topic Rush association:
  Found in trending topic → +25pts
  Topic net inflow >$10K  → +10pts

Fallback: present in Unified Rank → 30pts

Cap: 100

Dimension 3: Trend Score

Weight: 20% (default)

Data Source: Token Dynamic API real-time price changes

Scoring Logic:

1h price change:
  >20% → +40pts (strong trend)
  >10% → +30pts
  >5%  → +20pts
  >0%  → +10pts

5m momentum:
  >5%  → +20pts
  >2%  → +10pts

4h trend confirmation:
  >10% → +15pts
  >5%  → +8pts

Multi-timeframe resonance (5m+1h+4h all positive): +10pts
1h drop <-10%: -20pts penalty

Cap: 100

Dimension 4: Inflow/Volume Score

Weight: 20% (default)

Data Source: Token Dynamic API volume data

Scoring Logic:

5m volume:
  >$100K → 60pts
  >$50K  → 45pts
  >$10K  → 30pts
  >$5K   → 15pts

Buy/sell ratio (24h):
  Buy% ≥60% → +20pts (strong buy pressure)
  Buy% ≥55% → +10pts

1h volume:
  >$500K → +15pts
  >$100K → +8pts

Cap: 100

Dimension 5: KOL/Whale Score

Weight: 15% (default)

Data Source: Token Dynamic API holder data

Scoring Logic:

KOL holders:
  ≥10 → 50pts
  ≥5  → 35pts
  ≥2  → 20pts

Pro holders:
  ≥5  → +25pts
  ≥2  → +15pts
  ≥1  → +8pts

KOL holding percentage:
  >5% → +15pts

Cap: 100

Dimension 6: Hype Score

Weight: 15% (default)

Data Sources: Topic Rush data, Meme Exclusive ranking

Scoring Logic:

Topic Rush (Viral topics):
  Found in viral topic → 70pts
  Topic inflow >$10K   → +15pts

Meme Exclusive ranking:
  Score ≥4.0 → 80pts
  Score ≥3.0 → 60pts
  Score ≥2.0 → 40pts
  Listed     → 20pts

Cap: 100

Negative Signals (Penalty)

Applied after positive scoring. Can reduce total score.

Token audit risk (honeypot, rug pull):
  High risk detected  → -30pts + force dismiss

High tax (>10%):
  → -20pts

DEX screener paid without real traction:
  → -10pts

Final Score Calculation

rawScore = SM × w_sm + Social × w_social + Trend × w_trend + 
           Inflow × w_inflow + KOL × w_kol + Hype × w_hype

totalScore = max(0, rawScore + negativePenalty)

Default weights: SM=20, Social=10, Trend=20, Inflow=20, KOL=15, Hype=15

Strategy Configuration

Multiple strategies can be defined with different entry modes and thresholds.

Strategy Matching

When a token's totalScore reaches a strategy's buyThreshold:

1. Sort matching strategies by threshold (highest first)
2. Pick the first strategy where totalScore >= buyThreshold
3. This ensures higher-threshold strategies get priority

Paper Trading Simulation

Entry Logic

When evaluator sets status to buy_signal, paper trader:

1. Records entry price from Token Dynamic API
2. Creates a paper trade record with entry timestamp
3. Sets watchlist status to bought

Exit Logic (checked on each evaluation cycle)

Take Profit: price ≥ entry × (1 + takeProfitPct/100)  → sell, mark "tp"
Stop Loss:   price ≤ entry × (1 - stopLossPct/100)    → sell, mark "sl"
Timeout:     holdTime > maxHoldMinutes                  → sell, mark "timeout"

Default: Take Profit = 50%, Stop Loss = 20%, Max Hold = 1440 minutes (24h)

Trade Record Fields

Pipeline Workflow

The complete pipeline runs on a scheduler (default: every 5 minutes):

1. Collect Data    → Run all collectors (unified-rank, meme-rush, smart-money, social-hype)
2. Scan Watchlist  → Filter new tokens into watchlist based on global filters
3. Evaluate        → Score all watching tokens using 6-dimension algorithm
4. Paper Trade     → Execute simulated buys for buy_signal tokens
5. Monitor         → Check existing positions for TP/SL/timeout exits

Global Filters for Watchlist Entry

Notes

1. All scores are 0-100. Higher = more bullish.
2. Weights are percentages and should sum to 100 for proper normalization.
3. The evaluator fetches fresh Token Dynamic data before each evaluation for accuracy.
4. Strategy matching uses the highest-threshold-first approach for conviction grading.
5. Paper trading tracks simulated P&L for strategy backtesting without risk.