Trader Risk & Anomaly Detection

**Role:** Risk Analyst / Fraud Analyst / Trading Data Analyst  
**Dataset:** Synthetic / anonymized demo data created for portfolio use.  
**Stack:** Python, pandas, scoring logic, risk metrics, matplotlib

## Business problem

A prop-trading or fintech platform needs to identify suspicious trader behavior, payout abuse risk and unstable profiles using interpretable risk metrics.

## What was built

Built a rule-based suspicious score from 0 to 100 using win rate, drawdown, payout amount, accounts per IP, average holding time and news-trading concentration. Produced risk tiers, top-risk accounts and visual diagnostics.

## Key outputs

- `results/trader_risk_scores.csv` — full scoring output

• results/risk_tier_summary.csv — summary by risk tier

• results/suspicious_score_distribution.png — score distribution

• results/drawdown_vs_payout.png — risk pattern visualization

  How to run

  python -m venv .venv
  source .venv/bin/activate  # Windows: .venv\Scripts\activate
  pip install -r requirements.txt
  python src/main.py

  Resume-ready bullets

  • Designed an interpretable trader suspicious-score framework using performance, payout and behavioral features.

• Segmented traders into low/medium/high risk tiers and produced risk breakdowns for fraud/risk analytics use cases.