Financial Prediction Verifier
Paste the following prompt into your AI chat to install this skill:
Please install @user_9894ba11/analyse into my AI assistant following https://skillhub.cn/install/skillhub.md.
About this skill
Problem
Financial prediction notes are often scattered and inconsistent, and manual review requires repeated checks against prices, returns, and time windows. @user_9894ba11/analyse turns that into a repeatable data check: take a simple CSV and produce a scored report plus a detail CSV.
How It Works
- Input: only 3 columns are required:
predictor,predict_date, andprediction_text. - Scoring: predictions are compared against real market data and scored; formulas are documented in
reference.md. - Output: the skill writes
{input}_report.mdand{input}_scores.csvnext to the input file. - Runtime: Node.js >= 18 is required, model access is provided through
KIMI_API_KEYor related variables, and no dependency installation is required.
Boundary: best for structured review and auditability of prediction text; if the text lacks quantifiable targets, the market data source is unclear, or the time window cannot be aligned, output quality depends on input quality.
Use Cases
- Normalize financial prediction notes into a CSV and score
- Review analyst market views and export a Markdown report.
- Merge prediction records into one input and detail CSV.
- Reformat forecasts into three columns before scoring.
Best For
- Research analysts who need to review prediction text and keep scoring evidence.
- Data engineers who normalize prediction records into a standard CSV.
- Ops editors who archive analyst views in a Markdown report.
- Risk or compliance staff who batch-check prediction quality.
Related Skills
Generate Markdown public opinion reports by calling an internal service with MIDU_API_KEY.
Extracts Google AI Mode answers, standard SERP, AI Overviews, and citations via Pangolin APIs, with multi-turn follow-ups and region support.
Provides break-even analysis frameworks and templates without code execution, outputting structured recommendations.
Generate web reports from existing analysis data with classic or PPT-style layouts, Chart.js charts, and keyboard/touch navigation.