Scoring System

MangroveTrader ranks traders using a composite score that balances return, consistency, and risk management. Scores are recomputed nightly at midnight UTC.

Composite Score Formula

Composite Score = (0.50 × Return Percentile)
               + (0.30 × Consistency Percentile)
               + (0.20 × Risk Management Percentile)

Each component is a percentile rank (0–100) computed against all qualified traders.

Score Components

Return (50% weight)

• Metric: Average return % per closed trade
• Calculation: Mean of per-trade return percentages (relative to entry price)
• Also tracked: min return %, max return %, standard deviation
• Why average: Rewards efficiency per trade, not volume of trades

Consistency (30% weight)

• Metric: Sharpe ratio (annualized, 252 trading days)
• Formula: (mean_daily_return / std_dev) × sqrt(252)
• Minimum data: 30 daily returns required for meaningful calculation
• Why 30%: Consistent returns indicate repeatable strategy, not luck

Risk Management (20% weight)

• Metric: Maximum drawdown % (inverted — lower drawdown scores higher)
• Formula: ((peak - trough) / peak) × 100
• Why 20%: Managing risk prevents blowups and ensures longevity

Qualification

Traders must meet the following criteria to appear on the leaderboard:

• Minimum 1 closed trade — a complete round-trip (entry + exit)
• Open positions alone do not count toward qualification

The qualification threshold is intentionally low to encourage participation. As the trader base grows, the percentile-ranking system naturally differentiates skilled traders from casual ones.

Scoring Pipeline

The pipeline runs as a scheduled job:

1. Midnight UTC — APScheduler triggers compute_all_scores()
2. Fetch traders — all traders with 1+ closed trades
3. Compute metrics — total return %, Sharpe ratio, max drawdown per trader
4. Compute percentiles — rank each metric across the cohort
5. Composite score — weighted sum of percentile ranks
6. Sort & persist — update scoring_components table and traders rank
7. Cache leaderboard — write to Redis (5-minute TTL) and snapshot table

Leaderboard Timeframes

Performance Report

The free trader_performance_report tool returns a detailed breakdown:

{
  "twitter_handle": "toptrader",
  "composite_score": 87.4,
  "rank": 3,
  "total_return_pct": 24.7,
  "avg_return_pct": 0.53,
  "min_return_pct": -3.2,
  "max_return_pct": 8.1,
  "return_stddev": 2.14,
  "sharpe_ratio": 1.82,
  "max_drawdown_pct": 8.3,
  "win_rate": 0.68,
  "trade_count": 47,
  "scoring_weights": {
    "return": 0.5,
    "consistency": 0.3,
    "risk_management": 0.2
  }
}