Quant Analyst

quant-analyst

Overview

This public intake copy packages plugins/antigravity-awesome-skills-claude/skills/quant-analyst from https://github.com/sickn33/antigravity-awesome-skills into the native Omni Skills editorial shape without hiding its origin.

Use it for quant research and review tasks such as:

• building or reviewing financial models
• backtesting trading strategies
• analyzing market and event data
• computing risk metrics
• evaluating portfolio optimization outputs
• reviewing statistical arbitrage or factor research

Keep the upstream workflow, copied support files, and provenance visible. Treat outputs as research, diagnostics, and scenario analysis rather than personalized investment advice or live trading instructions.

For higher-confidence execution, use:

• references/review-criteria.md for a compact review rubric
• references/troubleshooting-patterns.md for quant-specific diagnostic patterns
• examples/review-example.md for a worked review example

When to Use This Skill

Use this skill when the task is primarily a quant research or quant workflow review problem.

Use it for

• Reviewing whether a backtest is methodologically credible
• Checking for look-ahead bias, survivorship bias, revised-data contamination, or leakage
• Evaluating point-in-time data joins between prices, fundamentals, and events
• Assessing whether train/validation/test splits are time-safe and walk-forward aware
• Reviewing risk metrics such as drawdown, Sharpe, turnover, exposure, and concentration
• Examining optimizer setup, constraints, feasibility, and sensitivity
• Turning a naive strategy summary into a more decision-useful research report
• Preserving provenance from imported workflow files while improving operational quality

Do not use it for

• Personalized investment advice or recommendations to buy, sell, or short a specific asset
• Broker-specific order routing or live execution playbooks
• Legal, tax, accounting, or compliance sign-off
• Pure data engineering tasks with little quant judgment involved
• Production deployment design, monitoring infrastructure, or exchange connectivity

If the request drifts into data pipelines, deployment, or general financial education, route to a more suitable skill if available.

Operating Table

Workflow

1. Clarify the research question

   • Identify the asset universe, horizon, target variable, rebalance cadence, and decision context.
   • Ask whether the task is exploratory research, a comparison study, or a decision-support review.
   • State known limits up front: missing data, unavailable timestamps, limited history, or execution assumptions.

2. Validate data timing and point-in-time availability

   • Confirm what each timestamp means: observation time, publication time, exchange close, or vendor load time.
   • Normalize time zones and trading calendar assumptions before joining series.
   • Check whether fundamentals, macro data, or event data may include revised values or delayed publication.
   • Prefer point-in-time safe joins, including backward-looking as-of logic when event data and market data live on different clocks.
   • Explicitly ask: could this dataset contain survivorship filtering, revised-history contamination, or post-event timestamps?

3. Define the evaluation design before modeling

   • Use time-ordered train/validation/test windows unless a cross-sectional design is clearly justified.
   • Separate research, tuning, and final evaluation.
   • Fit scalers, imputers, encoders, and feature selection only on training windows.
   • Prefer walk-forward or rolling evaluation when market nonstationarity matters.
   • Treat a single aggregate performance metric as insufficient evidence.

4. Review strategy logic and feature realism

   • Verify that signals are available when trades are assumed to occur.
   • Check holding period, rebalance frequency, and turnover implications.
   • Ask whether the strategy depends on unrealistic fills, unlimited borrow, or unbounded liquidity.
   • Label outputs as exploratory if costs, slippage, turnover, or capacity are not modeled.

5. Evaluate backtest realism

   • Require explicit assumptions for fees, slippage, spread, latency where relevant, liquidity, borrow constraints, leverage, and position caps.
   • Report not only return metrics but also drawdown, turnover, concentration, exposure drift, and regime sensitivity.
   • Distinguish between paper performance and decision-useful evidence.
   • If assumptions are weak, say so directly rather than presenting precise but fragile metrics.

6. Assess portfolio optimization carefully

   • State the objective, inputs, and all constraints before interpreting weights.
   • Check feasibility, concentration, leverage, turnover, and shorting assumptions.
   • Do not accept optimizer outputs just because a solver returned a solution.
   • Ask how sensitive the solution is to expected returns, covariance estimation, and constraint changes.

7. Capture reproducibility artifacts

   • Record data snapshot dates, source names, symbols/universe rules, time windows, and parameter settings.
   • If randomization or simulation is used, log seeds or generator configuration.
   • Preserve enough information for another reviewer to reproduce or challenge the result.
   • Only escalate to containerized or tightly pinned environments if dependency drift or solver instability is blocking reliable reruns.

8. Produce a bounded conclusion

   • Summarize what is supported, what is only exploratory, and what still needs validation.
   • Highlight the most material methodological risks.
   • Keep conclusions framed as research findings, not trade recommendations.

Imported Workflow Notes

Imported: Instructions

• Clarify goals, constraints, and required inputs.
• Apply relevant best practices and validate outcomes.
• Provide actionable steps and verification.
• If detailed examples are required, open resources/implementation-playbook.md.