CaligoApp.com

The problem

Retail traders are full of ideas (“buy when RSI is oversold and price reclaims the 50-day,” “fade gaps that fill by noon”), but almost none of them can test whether those ideas actually work. Validating a strategy means writing Python, sourcing clean market data, computing indicators, and building a backtest loop. That’s a wall most traders never get over, so they trade on untested intuition instead of evidence.

The result is a gap between having a hypothesis and knowing if it holds. Existing tools force a choice: rigid point-and-click screeners that can’t express a real strategy, or full code platforms that assume you’re an engineer. Nothing let a non-technical trader go from a plain-English idea to a rigorous, data-backed answer.

The approach

Caligo is an AI-assisted research terminal that turns a strategy described in natural language into a structured, reproducible backtest, with no code required.

A trader describes an idea in plain English. An AI analysis agent interprets it, maps it onto a library of technical indicators and pattern conditions, and assembles a runnable strategy from reusable building blocks (factors, indicators, entry/exit conditions). Backtests run against institutional-grade market data from Polygon.io, and results come back as performance metrics the trader can actually interrogate.

Crucially, the platform doesn’t stop at a single equity curve, which is where most retail backtesting goes wrong. A strategy that looks great on one set of parameters is usually just overfit. So Caligo layers in the analysis a quant would insist on: parameter-sensitivity sweeps to show whether an edge survives small changes, robustness analysis to separate signal from luck, and configurable exit conditions so the test reflects how the strategy would really be traded. The AI agent stays in the loop throughout, explaining results in language a trader understands rather than dumping raw numbers.

Under the hood it’s a full-stack system: a Python backend handling data ingestion, indicator computation, and backtest execution, a React/TypeScript front end for the research workflow, PostgreSQL for storage, Stripe for subscriptions, all deployed on AWS.

The outcome

Caligo collapses the path from idea to evidence from “learn to code first” down to a single conversation. A trader can validate (or kill) a strategy in minutes instead of never, and the built-in robustness and sensitivity analysis pushes them away from overfit strategies toward edges that actually hold up out of sample.

More broadly, it lowers the barrier to entry for retail traders moving from discretionary, gut-feel trading toward a systematic, evidence-based process, making the kind of rigor that used to require an engineering background accessible to anyone with an idea worth testing.