From strategy idea to operation¶

The QuantDinger strategy lifecycle covers hypothesis, code, validation, backtest, versioning, deployment, and monitoring. Each stage is a separate checkpoint so that “it runs” is never treated as “it is ready for live trading.”

1. Define the contract¶

Before coding, define the market, instruments, frequency, entries, exits, position limits, costs, direction, and stop conditions. Declare direction capability for Crypto swaps. Design spot and currently supported non-Crypto strategies as long-only.

2. Implement in the strategy editor¶

Start from a Strategy API V2 template. Keep initialize, subscriptions, universe setup, and handle_data responsibilities clear. Use full market identifiers and never infer a venue or product from a display name.

See the Strategy API V2 development guide for the programming surface, manifest, and sandbox rules.

3. Validate and version¶

Run source validation and resolve every error and warning. Record the strategy purpose and important parameters when saving. Create a new version when universe, frequency, sizing, or exit behavior changes so results remain comparable and reversible.

4. Backtest¶

Choose a range that includes warmup and multiple market regimes. Check coverage, benchmark, costs, slippage, funding, and liquidity constraints. Use the Backtest Center guide to review results.

5. Create a stopped deployment¶

Create deployments in the stopped state. Confirm that credential, market, settlement currency, and product type match. Use signal mode to validate notifications and state restoration. Do not mix incompatible markets or accounts in one deployment.

6. Start and monitor¶

Complete the live-trading safety checklist. After startup, monitor runtime state, orders, position ownership, unknown deltas, notifications, and errors. Stop new entries first when behavior is uncertain; stop the strategy or use emergency stop when required.

Release checklist¶

• Source and manifest validation pass.
• Backtest history and warmup are complete.
• Costs, slippage, funding, and minimum-order rules are understood.
• The credential matches the strategy market.
• Signal mode has been observed for an adequate period.
• Stop, recovery, and manual incident paths have been tested.