| name | ngat-strategies |
| description | Implement and refine strategy logic in strategies/ based on BaseStrategy contracts, currently centered on Supertrend + RSI signal generation. Use when adding new strategies, modifying signal criteria, or improving risk controls and dedup behavior. |
Ngat Strategies
Overview
Maintain reusable strategy modules that receive candles and emit trade signals.
Submodules
strategies/supertrend_rsi.py: SupertrendRSIStrategy.core/base.py::BaseStrategy: on_candle and reset contract.
Pros
- Keep strategy self-contained and simple to reason about.
- Deduplicate repeated signals with
last_signal memory.
Cons
- Keep no explicit stop-loss/target/risk model in signal payload.
- Keep single-strategy example with limited portfolio context.
- Depend on pandas-ta naming conventions that can shift by version.
Caveats
- Validate indicator column keys from pandas-ta before direct indexing.
- Avoid making strategy state non-serializable if moved across Ray workers.
- Ensure minimum warmup rows before signal generation.
Forward Features
- Add risk module fields (
stop_loss, take_profit, position_sizing).
- Add strategy registry and dynamic loading from config.
- Add portfolio-aware filters (exposure, correlation, daily loss cap).
