Caskey Engineering

← All Case Studies

Caskey Coding2024 – Present · Founder & Full-Stack Builder

Factor-First AI Investment Analysis Narrated by a Six-Persona Committee

Built a Quality + Valuation + Growth + Momentum + Health factor composite narrated by a six-persona committee — one consistent framework across every public equity.

The Problem

Standard brokerage tools expose raw financial metrics but do not synthesize them into a conviction-level judgment. Investors still do their own mental weighting, often inconsistently and without a documented framework.

The core gap is not data access; financial APIs are commodity. The real problem is a missing, repeatable scoring architecture that combines quantitative fundamentals with qualitative moat assessment.

The platform also had to be robust enough for real decision-making with personal capital, not a demo workflow.

The Approach

Moved from a Graham-centric parallel-scorer committee to a factor-first composite (ADR-011): Quality, Valuation, Growth, Momentum, and Health each score independently and aggregate into a single composite.

Built a six-persona investment committee that narrates the factor scores rather than recomputing them in parallel — each persona owns a factor lens and explains the reading in their philosophical voice.

Added Piotroski F-Score and Altman Z-Score as diagnostic gates that surface independently, not as factor bars, so a distress zone hard-caps the composite instead of being averaged away.

Pulled quantitative signals from financial data APIs and delegated committee narration to Claude via the Anthropic API (with Bedrock failover) using a constrained prompt that returns structured verdicts rather than open-ended commentary.

Grounded the LLM with factual financial context at inference time and added public no-login access with rate limiting at 10 requests per IP per hour.

The Impact

  • Returns a repeatable, structured verdict for any US-listed equity — a 0-100 composite and A+ to F grade with six-persona narration underneath — instead of raw metrics the investor has to weigh by hand
  • Distress gates (Piotroski F-Score, Altman Z-Score) hard-cap the composite rather than being averaged away, so a weak balance sheet can't be masked by a strong factor
  • One consistent factor framework across every ticker makes ticker-to-ticker comparison repeatable instead of ad-hoc, and surfaces which factor and which persona drove conviction
  • A public no-login demo narrates a fixed sample portfolio so anyone can see the committee in action — engineered for safe public exposure with a hard daily cost ceiling, per-session rate caps, and a default-off kill switch
AI/MLFinancial SystemsPlatformFull-StackAWS

Related

Tool

Investment Committee

Score any stock across five weighted dimensions and get a letter grade with a written committee verdict.

Read
Tool

Market's Best

The top-graded stocks from the latest market scan. No sign-in needed.

Read
Demo

Grade my portfolio

Run a sample portfolio through the investor committee.

Read
Post

Building a Personal Finance Reviewer: What Survived the Rewrite

A personal portfolio reviewer where the scoring is deterministic and the AI only narrates. The architecture that held up after I had to rewrite the model it was built on, and why that boundary is the whole point.

Read
Post

A Boring Design Let Me Run a Black Swan on a Tuesday

Two posts ago I bet that keeping my portfolio reviewer's engine deterministic and auditable was worth it. This is where that bet paid off: because the engine is replayable, I could run a simulated market crash through the real production code and catch a money-losing flaw on paper — before it could ever cost a real dollar.

Read
Post

When the Spec Was Wrong: Rewriting a Shipped Decision

Two weeks after I shipped a post about a scoring engine I'd built, I rewrote the spec it was based on. Here's what I learned, and why I had an AI agent do the literature review.

Read