SimpleRebalance
Rules-based target-weight rebalancing.
Java 21 · Interactive Brokers · live for ~5 years
A live, rules-based algorithmic trading engine.
Runs in Docker against the Interactive Brokers TWS API, manages real and paper accounts, and has been placing real orders unattended on a daily cron for ~5 years.
Framing: rules-based. No prediction, no discretionary calls, no returns or performance claims. The signal is engineering and operational rigor — a live system I run.
What it does
Trades automatically across real accounts.
Connects to Interactive Brokers TWS and runs rule-based rebalancing,
signal, and exit strategies against 3+ live and paper accounts,
every day.
Feeds FamilyFund.
Pushes positions and prices over REST into
FamilyFund, the book of record.
Reports daily.
Generates HTML email reports with JFreeChart charts — the operating
visibility for the engine.
How it works
- Engine
- Java 21, Maven, runs in Docker; daily cron.
- Broker
- Interactive Brokers TWS API; live + paper across 3+ real accounts.
- Strategies
- Two independent rule-based strategies (SimpleRebalance, LinearRegression) behind a common execution interface.
- Data in
- Positions, prices, and executions read from Interactive Brokers over the same TWS connection.
- Data out
- Pushes positions & prices to FamilyFund via
REST (
POST /api/asset_prices_bulk_update,POST /api/portfolio_assets_bulk_update). - Reporting
- HTML email reports with JFreeChart charts.
Strategies
Mechanical and rule-driven — no AI in the signal path.
LinearRegression
Trend-based signal.
Investment strategy
The thesis the engine implements: start from the broad index, treat leverage as capital efficiency rather than speculation, split that exposure across All Seasons-style buckets, and let the rebalancer hold the targets. Mechanical, not predictive — no calls on direction, no discretionary overrides.
Investment strategy
Starts from the broad index as the baseline, uses 3x leverage as capital efficiency instead of stock-picking, buckets exposure across All Seasons categories (growth, stability, crypto), and lets a quarterly rebalance hold the targets.
Rebalancing strategy
Why most strategies try to predict the market and this one doesn't — target weights, the drift problem, trading bands around each target, the sell-high / buy-low triggers, the tight-vs-wide band knob, and the time-based force rebalance every ninety days regardless of drift.
Trading bands
Per-asset target allocations with deviation triggers and live drift charts — how the rules look in FamilyFund, which is the book of record that DS Trader feeds.
Engineering & operations
- ~570 tests (44 TestNG classes, with data-provider row expansion) guarding the order logic.
- Daily scheduled execution (cron) in production for ~5 years, unattended.
- Multi-account orchestration against a real brokerage API (live and paper).
- HTML / charting report pipeline for daily operating visibility.
The integrated platform
DS Trader is one half of a two-app platform: it feeds FamilyFund, a Laravel portfolio app my family actually uses. End-to-end ownership across Java, PHP, Python, data modeling, API design, and operations.
About / contact
Built and operated solo. — jdtogni@gmail.com · GitHub · back to home