Work / Builds

Algorithmic Paper-Trading Platform

A Python system with three paper-trading engines (classic strategies, Kalshi event contracts, and NOAA-driven weather markets), a browser command center, and a documented research library.

Role: Designer and builder
Context: 2026 · ongoing
Team: Personal project
Deliverable: Software build + strategy inventory

3paper engines implemented

22strategy families documented

124hypotheses in the research gate

Summary

The question

Apparent trading edges often disappear after spreads, fees, liquidity, contract rules, and execution are counted. The goal was a system that could test ideas under those frictions and refuse to promote anything that had not earned it.

What I did

Built three engines on a $1,000 paper bankroll each: five classic strategies with a TWAP benchmark, an event-contract scanner that requires an exact rules match and executable quotes, and a weather model that compares NOAA/NWS forecasts with contract prices.
Added a command center that launches the engines, logs source health and data freshness, and keeps live order routing disabled.
Documented 22 strategy families and 124 event-to-security hypotheses, each with what it would take to prove or kill it before promotion.

What it showed

The most useful output so far is discipline: the latest eight-hour run found zero qualifying event trades and a weather result of −$0.20 on $1,000. The system is doing its job by saying no.

From the work

The platform is suitable for controlled paper experimentation. It is not ready for live-money use: no order route is active, the latest full run is only one evidence window, event candidates were zero, and modeled costs absorbed most of the classic sleeve’s gross result.
Strategy and system inventory, executive map

What exists today
Layer	What it does	Status
Command center	Starts the engines, monitors the portfolio and logs	Implemented
Classic engine	Five alpha strategies plus a TWAP benchmark	Implemented, paper
Event-contract engine	Scans Kalshi quotes for exact rule matches and validated candidates	Implemented, sparse evidence
Weather engine	NOAA/NWS probability model compared with contract prices	Implemented, paper
124 research pairs	Event-to-security hypotheses, each with a kill switch	Research backlog

In my words

I built this platform to turn investment ideas into testable hypotheses instead of relying on intuition alone. It gives me a structured place to paper trade strategies across traditional markets, event contracts, and weather markets. Before I would trust any strategy with real money, I would want repeated out-of-sample testing, realistic execution assumptions, and a long enough paper track record to challenge the original thesis.

Documents