Model evaluation for finance
Measure what matters—latency, accuracy, cost.
Claude Code workflows
Earnings digests and 10-K extractions that work.
def extract_earnings(filing): structured = parse_10k(filing) return { 'revenue': structured.revenue, 'guidance': structured.guidance }Compliance-safe implementation
Logging and audit trails your risk team will approve.
Works with your stack
Python, SQL, and notebooks—no breaking changes.
From 10-K to CSV
Before/after table extraction
# Before (PDF Table)Quarter | Revenue | GrowthQ1 2024 | $60.9B | +262%# After (Clean CSV) quarter,revenue_billions,growth_yoyQ1_2024,60.9,2.62Trusted by practitioners across the investment lifecycle
Workflow Gallery
Production-ready templates that save hours of development time
Earnings Call → Analyst Brief
Convert earnings calls into structured analyst briefs with key metrics and insights.
10-K Tables → Clean CSV
Extract and clean financial tables from 10-K filings into usable CSV format.
Backtest Builder with Tests
Generate backtesting code with built-in validation and compliance checks.
Model Output Compliance Check
Validate model outputs against regulatory requirements and internal policies.
Earnings Transcript Analysis
Extract sentiment, key topics, and quantitative mentions from earnings transcripts.
Regulatory Filing Monitor
Monitor and categorize regulatory filings with automated alert system.
Release Cadence
Consistent delivery of tools, templates, and insights to keep you ahead
Weekly Workflow Drop
Hands-on workflow with inputs, outputs, and checks
Model Eval Notes
Sampling harness + regression snapshots
Compliance Template
Logs, model card snippet, SOP redlines
Live Build Session
From 10-K to usable tables in minutes
All content delivered directly to your inbox and available in the workflow gallery
Research & Analysis
Quantitative insights, model benchmarks, and systematic approaches to AI-driven finance
Earnings Sentiment Signal Extraction
Quantitative framework for extracting sentiment alpha from earnings calls using transformer models and sector-specific lexicons.
10-K Financial Table Automation
Production system for parsing SEC filings into structured datasets with 99.2% accuracy using multimodal LLMs.
Options Flow Pattern Recognition
AI-powered detection of unusual options activity patterns with sub-second latency for systematic strategies.
Dynamic Risk Factor Monitoring
Real-time portfolio risk assessment using LLM-derived factor models with regulatory compliance frameworks.
Market Regime Classification Models
Ensemble approach combining traditional econometrics with LLM-based news sentiment for regime detection.
Credit Risk Signal Processing
Alternative data fusion for corporate credit assessment using earnings transcripts, SEC filings, and news sentiment.
What your risk team cares about
Built-in compliance features that address regulatory requirements and internal policies from day one
Sample SOP Template
# Standard Operating Procedure: LLM Model Usage## 1. Input Validation- Verify data sources and lineage- Check for PII/sensitive information- Validate against approved schema## 2. Model Execution- Log all prompts and responses- Track token usage and costs- Monitor for bias indicators## 3. Output Review- Human review of critical outputs- Automated compliance checks- Version control for model cards## 4. Audit Trail- Maintain complete execution logs- Store model metadata and versions- Document reviewer actionsBe first to the workflows
Short, useful emails. No spam.