All projects
LiveTrading · 2026

AI Personal Trading Platform

An experimental AI-assisted market research & paper-trading lab.

invest.basilan.us — Research Paper-only

Regime signal

Neutral

Confidence 62%
Strategy research · backtest · paper
AI analysis
Multi-agent
Paper simulation
Simulated
Risk controls
Enforced

The idea

Markets are noisy and full of opinions. I wanted a private space to study them on my own terms — to test ideas with real discipline, keep the research honest, and separate genuine edge from wishful thinking. The result is a personal research lab, not a product sold to anyone.

The problem

Most tooling either hides the reasoning behind a black box or drowns you in raw data. I wanted something in between: clear signals, explainable logic, and guardrails that keep experiments safe while I learn what actually works.

What I built

A local-first dashboard that pulls read-only market data, runs strategy research and backtesting, and simulates paper trades with enforced risk controls. It is deliberately paper-only — no live brokerage orders are placed by the system.

How it works

A safe research loop

Data flows in read-only, research runs locally, and every action stays inside a paper environment. Credentials and keys live only on the local machine — nothing sensitive is published or exposed.

AI-assisted market analysis

Multi-agent workflows summarize context and surface study candidates — always as research, never as instructions.

Strategy research & backtesting

Regime-gated strategies tested against historical data with conservative, explainable scoring.

Paper trading engine

Simulated fills and monitored positions starting from a fixed test balance, with strict stop/target logic.

Deterministic risk controls

Position limits, exposure caps, and a kill-switch hierarchy. Hard safety rules never depend on the AI layer.

Read-only broker integration

Market and account data arrive through read-only APIs. The system cannot submit live orders.

Monitoring & alerts

Scheduled checks and Telegram notifications keep the research loop running without constant manual attention.

AI & automation

AI assists. Rules decide.

Independent AI workers help with market-context summarization, candidate research, and coaching-style notes on closed paper trades. They are observational only.

The final decisions — entries, exits, sizing, and risk — are made by deterministic code and explicit controls. The AI layer can suggest and explain, but it cannot relax a stop, submit an order, or change the safety posture. This separation is the whole point of the design.

Automation handles the repetitive loop: scheduled research refreshes, monitoring, and notifications — so the lab keeps working even when I'm not watching it.

Strategy & research

Several strategy styles are explored in parallel — from broad-universe, regime-gated approaches to more focused momentum and volatility research. Each is backtested, paper-tested, and tracked with conservative readiness gates before it earns any trust.

Performance is treated as a hypothesis to be challenged, not a number to be advertised. Small samples are labeled inconclusive, and no claims of guaranteed results are ever made.

Always improving

The platform is a living experiment. New strategies, better research tooling, and tighter controls are added over time, with each change validated before it can affect anything.

It remains a personal research environment — a place to learn how markets behave and how automation can help, without risking real capital.

Technology

Built with

PythonSchwab Trader API (read-only)Multi-agent AIBacktestingPaper TradingRisk ControlsTelegram Alerts

The system is implemented in Python for research, data, and automation, with a web dashboard for exploration. Market data is integrated read-only, and automation runs through scheduled jobs with monitoring and alerting.

Explore AI Personal Trading Platform