TASK-FIN-103-03 · Task family
Market data integrity & corporate action handling. Data validation, adjusted prices and historical consistency.
Overview. Produce a validated, adjusted price and return history and a complete universe, with every correction documented.
A price history has unadjusted splits, missing dividends and no delisted securities. The agent validates the data against the corporate-action record, corrects it and documents every change.
A split that isn't adjusted shows up as a huge one-day loss. Deleting the outlier instead of adjusting for the split corrupts the series.
Value to the buyer
Bad data silently ruins research. Trains agents to validate against the record, not to smooth.
Specification. The task family, field by field.
- task_id:
- TASK-FIN-103-03
- domain:
- Quantitative Research
- parent_environment:
- ENV-FIN-103 · Systematic Alpha Research
- target_role:
- Data Scientist
- objective:
- Produce a validated, adjusted price and return history and a complete universe, with every correction documented.
- starting_context:
- A raw daily price file, the corporate-action record, the delisting record and the data-quality checklist.
- tools:
query_market_dataget_corporate_actionsexecute_pythonsubmit_research_artifacts
- required_actions:
- Detect jumps and gaps that match corporate actions
- Apply split and dividend adjustments
- Restore delisted securities with delisting returns
- Document each correction
- expected_outputs:
- Adjusted price and return series
- Corrected universe
- Data-quality report
- interaction:
- Single agent with sandboxed code execution.
- verification:
- Adjusted series are compared with the reference built from the corporate-action record.
- reward:
- Credit for each correctly handled event. Deleting valid data to remove outliers scores zero on that check.
- difficulty:
- Calibrated against reference models during acceptance testing
- task_horizon:
- Multi-step data work. Execution limits are agreed with the buyer.
- delivery:
- Task instances with seeded data, reference series and verifier code.
- commercial_unit:
- Per validated task instance
Sample prompt. What the agent is asked to do, as a buyer would see it.
The price file shows Corvid Semiconductor closing at $186.30 on June 13, 2025 and $62.40 on June 16, 2025, a one-day return of −66.5%. Several securities in the universe also stop appearing without a delisting return. Validate the data against the corporate-action and delisting records, correct it and document every change.
Verification criteria. What the verifier checks, and how.
Split found
A 3-for-1 split effective June 16, 2025 explains the drop.
Adjustment
Prior prices are divided by 3: the adjusted June 13 close is $62.10, so the return is +0.48%.
Dividends
Total returns include cash dividends on their ex-dates.
Delistings
Delisted securities stay in the universe until delisting, with their delisting returns.
Report
Every correction is listed with its source record.
Key failure modes. Where agents go wrong on this task.
- Deleting the −66.5% day as an outlier
- Adjusting prices but not volumes or per-share fundamentals
- Dropping delisted securities from the history
Other task families. The rest of Systematic Alpha Research, on the same FinanceWorld state.
- TASK-FIN-103-01
Point-in-time factor reconstruction & leakage audit
Temporal reasoning, data lineage and feature engineering.
Quantitative Researcher
- TASK-FIN-103-02
Strategy implementation & backtest repair
Quantitative coding, financial calculations and translating a strategy into code.
Quantitative Developer
- TASK-FIN-103-04
Strategy robustness & overfitting audit
Statistical testing, out-of-sample analysis and methodological validity.
Quantitative Researcher