> For the complete documentation index, see [llms.txt](https://docs.sire.bot/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.sire.bot/how-it-works/editor-1.md).
# Architecture
SIRE’s architecture runs as a sequence of interlocking processes, each one refining, weighting, and executing intelligence in real time.
From intake to optimisation, every layer builds on the last, creating a continuous adaptive loop.
#### System Flow
**1. \ Information Intake**
Multi-source contributor packets enter the system, including Score Vision (Subnet 44) data, proprietary datasets, live match feeds, and market odds.\
Each packet contains unique predictive signals that seed the forecasting engine.
**2. Model Processing**
The SIRE LLM processes all packets, interpreting statistical and visual relationships, generating structured representations of match context, and aligning them with model priors.
**3. Multiple Prediction Runs**
The engine performs ensemble inference: multiple independent predictions across different packet combinations, to estimate fair value, edge, and volatility-adjusted confidence intervals.
**4. Outcome Tracking**
Every prediction is recorded against real-world outcomes.
Market prices, settlement data, and live match results are logged to benchmark and recalibrate performance continuously.
**5. Performance Scoring**
Each contributor packet is evaluated using statistical metrics such as calibration, Sharpe ratio, and information value.
Contributors are ranked based on predictive quality and long-term reliability.
**6. Dynamic Weighting**
High-performing contributors are up-weighted; underperforming or correlated ones are reduced or replaced.
Sizing logic applies fractional Kelly scaling with adaptive exposure limits to maintain consistent risk profiles.
**7. Iterative Optimisation \**
The engine retrains on new data, updates weights, and redeploys refined models automatically.
Sharper, verified signals flow to **aLink** for visibility and **aVault** for autonomous execution.
#### Why It Works
* Unified pipeline connecting data ingestion, modelling, and execution.
* Continuous performance feedback ensures adaptive learning.
* Modular structure allows new models, datasets, or signals to integrate without redesign.
* Verifiable on-chain transparency at every step.
Read more in [The Multi-Source Prediction Engine](/research-and-development/the-multi-source-prediction-engine.md) section.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://docs.sire.bot/how-it-works/editor-1.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.