Legacy Architecture Context¶
This page remains for historical links and compatibility context.
It is not the primary architecture truth anymore.
Use These Pages First¶
If you are trying to understand the current runtime, start here instead:
Why This Page Changed¶
Older Cracktrader docs were organized around a simpler mental model:
- Cerebro as the center of the system
- feeds and brokers as the main runtime boundaries
- stores as the dominant integration surface
That framing is still useful when reading legacy compatibility code or migration-oriented pages, but it is no longer the preferred top-level explanation.
The current runtime direction is session-owned and contract-first:
- shared reference data and state coordination
- multi-strategy fanout over one shared snapshot per step
- execution contexts and stable route IDs
- central inventory, exposure, and risk
- unified execution adapters across backtest, paper, sandbox, and live
- post-trade and control-plane hooks around the runtime edge
How To Read Older Pages¶
When you encounter older terms, reinterpret them this way:
| Older emphasis | How to read it now |
|---|---|
| Cerebro-centric execution | important compatibility surface, not the whole runtime map |
| Feed + broker wiring | one layer inside a broader session-owned orchestration model |
| Store as architecture center | external IO and transport ownership, not runtime ownership |
| Risk manager near broker | one slice of a broader explicit risk engine |
Where Legacy Framing Still Helps¶
Older pages are still useful for:
- Backtrader compatibility behavior
- exchange/store implementation detail
- feed or broker-specific usage notes
- migration work where old and new abstractions coexist