tradedesk 0.2.0

Trading infrastructure library with backtesting

tradedesk

What tradedesk is

tradedesk is a lightweight Python framework for developing, backtesting, and running systematic trading strategies across multiple data providers and execution environments.

It provides:

• A consistent strategy lifecycle
• Explicit separation between market data, strategy logic, and execution
• Deterministic backtesting with identical strategy code
• Live and DEMO execution via provider-specific clients

It is designed for research, validation, and controlled deployment of trading strategies rather than high-frequency or ultra-low-latency trading.

What tradedesk is not

tradedesk is intentionally not:

• A portfolio management system
• A signal marketplace
• A turnkey trading bot
• A performance-optimised HFT engine

It REQUIRES the user to accept responsibility for strategy design, risk management, and operational controls.

High-level architecture

At a high level, tradedesk consists of four layers:

1. Providers – Interfaces to external data/execution sources (e.g. IG, historical backtest data)
2. Clients – Concrete implementations that fetch data and place orders
3. Strategies – User-defined trading logic responding to market events
4. Runner – Orchestrates lifecycle, subscriptions, warmup, and shutdown

Market data flows from the provider into the strategy, which emits execution decisions back through the client.

Core concepts

Strategy lifecycle

A strategy progresses through the following phases:

1. Construction
2. Subscription registration
3. Warmup (optional but strongly recommended)
4. Live or replayed market data handling
5. Order execution
6. Graceful shutdown

Subscriptions

Strategies explicitly declare their required data via subscriptions:

• MarketSubscription – Tick-level price updates (bid/offer)
• ChartSubscription – Aggregated candle data for a given timeframe

Only subscribed data is delivered to the strategy.

Warmup

Warmup allows a strategy to request historical data before live execution begins in order to initialise indicator state and internal windows.

A strategy that does not warm up must be robust to partially initialised indicators and delayed signal readiness.

Backtest vs live execution

The same strategy code can be run against:

• A live or DEMO provider
• A deterministic backtest client that replays historical data

Differences between environments are isolated to the client layer.

Supported providers

• IG – Live and DEMO trading via REST and streaming APIs
• BacktestClient – Deterministic replay of historical data

Historical data acquisition is handled by a companion project:

• tradedesk-dukascopy

Quick start

1. Install dependencies
2. Write or select a strategy
3. Choose a client (backtest or live)
4. Run via the tradedesk runner

Detailed tutorials are provided in the documentation guides listed below.

Project status and guarantees

• APIs may evolve as the framework matures
• Backwards compatibility is maintained on a best-effort basis
• The framework prioritises correctness and clarity over performance

Further reading

• docs/indicators.md – Indicator concepts and mathematical foundations
• docs/strategy_writing_guide.md – Step-by-step strategy tutorial
• docs/backtesting_guide.md – Methodology for rigorous backtesting

---

License

Licensed under the Apache License, Version 2.0. See: https://www.apache.org/licenses/LICENSE-2.0

Copyright 2026 Radius Red Ltd.