By name: nb-bias-detector-challenge description: Builds solutions aligned with National Bank Bias Detector Challenge requirements, including data input, bias detection logic, personalization, and judging criteria.
National Bank Bias Detector Challenge
Purpose
Use this skill to implement or review features so they match the official challenge brief for the National Bank Bias Detector project.
When to Use
- Defining scope for the Bias Detector prototype
- Designing CSV/Excel import and trade-entry workflows
- Implementing behavioral bias detection rules or models
- Generating user-facing feedback and recommendations
- Checking judging-readiness (performance, creativity, behavioral insight, personalization)
Required Product Capabilities
1. Trading History Input (support one or more)
- File upload for CSV/Excel trading records
- Simple UI form for manual sample trades
Expected core fields:
- Timestamp
- Buy/Sell
- Asset
- Quantity
- Entry price
- Exit price
- P/L
- Account balance
2. Mandatory Bias Detection
Overtrading
Detect patterns such as:
- Excessive trade count relative to account balance
- Frequent position switching
- Trading shortly after large losses or wins
- Time-clustered overactivity (for example, too many trades in one hour)
Loss Aversion
Detect patterns such as:
- Letting losing trades run too long
- Closing winning trades too early
- Unbalanced risk/reward
- Average loss size larger than average win size
Revenge Trading
Detect patterns such as:
- Larger trade sizing immediately after losses
- Increased risk-taking after negative P/L streaks
3. Feedback and Recommendations
Output should include:
- Bias summaries in plain language
- Graphical insights (charts, timelines, heatmaps)
- Personalized recommendations, such as:
- Daily trade limits
- Stop-loss discipline
- Cooling-off periods
- Journaling prompts for trading psychology
Optional Enhancements
- Detect additional behavioral biases
- Portfolio optimization suggestions
- Sentiment analysis on trader notes
- Risk profile scoring
- Predictive alerts for likely bias-triggering situations
- AI trading coach chatbot
- Stress/emotional state tagging
Judging-Aligned Quality Checklist
Performance
- Analysis is fast enough for practical use
- Design scales to larger datasets (target at least 20x mock-data size)
Creativity
- UX and visualizations are intentional and engaging
- AI/ML or rules are used in meaningful, explainable ways
Behavioral Finance Insight
- Bias definitions and signals are behaviorally sound
- Explanations are understandable to non-experts
- Interpretation goes beyond shallow metrics
Personalization
- Feedback adapts to each trader's history
- Recommendations are specific, not generic
- Outputs update dynamically with new trade data
Implementation Guardrails
- Prefer transparent, auditable signal logic over black-box claims
- Keep recommendations actionable and tied to observed behavior
- Separate detection confidence from recommendation severity
- Handle missing or dirty data gracefully and report data quality issues