AgentSkillsCN

statistics

适用于量化交易研究中假设检验、估计不确定性、模型验证以及推断质量的统计工作流。当任务涉及显著性检验、置信区间、分布检验，或对统计有效性的生产监控时，可选用此类工作流。

SKILL.md
--- frontmatter
name: statistics
description: "Statistics workflows for hypothesis testing, estimation uncertainty, model validation, and inference quality in quantitative trading research. use when tasks involve significance testing, confidence intervals, distribution checks, or production monitoring of statistical validity."

Statistics

objective

Apply robust statistical inference to quantify uncertainty, avoid false discoveries, and validate model assumptions.

workflow

  1. •define hypotheses, sampling assumptions, and test-selection criteria.
  2. •estimate parameters and uncertainty intervals with robust methods.
  3. •run assumption validation for distribution, dependence, and variance stability.
  4. •evaluate multiple-testing and false-discovery risk in signal screening.
  5. •publish inference only when validation and robustness checks pass.

required validation

  • •p-value calibration and type-i/type-ii error balance.
  • •confidence-interval coverage and estimator stability.
  • •residual normality, autocorrelation, and heteroskedasticity validation.
  • •multiple-testing adjustment impact on selected signals.
  • •outlier and leverage-point sensitivity analysis.

risk controls

  • •enforce pre-registered testing rules for repeated experiments.
  • •enforce minimum sample-quality and power thresholds.
  • •enforce robust fallback estimators when assumptions fail.

outputs

  • •run python scripts/statistics_validation.py input.csv --output validation.json and keep the json artifact.
  • •write an implementation memo using references/statistics-playbook.md with assumptions, tests, limits, and rollout plan.

resources

  • •use scripts/statistics_validation.py for deterministic validation.
  • •use references/statistics-playbook.md for the domain checklist and delivery structure.