AlphaForge – System Context (C1)

Source: Notion | Last edited: 2025-11-21 | ID: 2b22d2dc-3ef...

---

Narrative

AlphaForge sits between quant researchers & AI agents on one side, and execution venues & data providers on the other.

Input Side

From quant researchers and AI agents

They submit:

• Strategy DSL specifications
• Experiment configurations
• Deployment / promotion requests
  • e.g. “promote this config to live trading” From data providers

Data providers (exchanges, index vendors, on-chain sources) feed:

• Raw market data
• Pre-aggregated OHLCV
• On-chain events and metrics

---

Inside AlphaForge

• The DSL and Compiler translate high-level strategy descriptions into executable DAGs.
• The Orchestrator manages backtests, simulations, paper trading, and live strategies.
• The Data and Feature Service provides consistent historical and live data views to all runs.
• The Execution Gateway connects to external execution engines and venues in a pluggable way.

---

Output Side

To execution venues

• Execution venues receive orders and manage positions. To humans & agents

Researchers, PMs, and AI agents consume:

• Performance metrics and curves

• Factor exposures and diagnostics

• Logs, artifacts, and experiment metadata via:

• APIs

• Dashboards

• Notebooks

---

System Context Diagram (Mermaid C4)

In Notion, add a Code block, set the language to mermaid, and paste the following:

C4Context
title AlphaForge / QuantOS – System Context

Person(researcher, "Quant Researcher", "Designs and tests strategies using the AlphaForge DSL.")
Person(ai_agent, "AI Research Agent", "Automates experiment generation and analysis.")
Person(pm, "Portfolio Manager", "Reviews performance, risk, and approves deployment.")
Person(devops, "DevOps / SRE", "Operates the AlphaForge platform.")

System_Boundary(af, "AlphaForge / QuantOS") {
    System(af_core, "AlphaForge Core", "DSL compiler, orchestrator, data and execution services.")
}

System_Ext(data_providers, "Market and On-Chain Data Providers", "Exchanges, data vendors, index providers.")
System_Ext(exchanges, "Trading Venues and Brokers", "Crypto exchanges, prime brokers, execution engines.")
System_Ext(storage, "External Storage and Object Store", "Long-term storage for raw and derived data.")
System_Ext(analytics, "Analytics and BI Tools", "Dashboards, notebooks, and reporting tools.")

Rel(researcher, af_core, "Submits strategies and experiments via DSL or API")
Rel(ai_agent, af_core, "Calls APIs to create and manage experiments")
Rel(pm, af_core, "Queries performance and risk metrics")
Rel(devops, af_core, "Deploys, monitors, and scales the platform")

Rel(data_providers, af_core, "Pushes or streams market and on-chain data")
Rel(af_core, exchanges, "Sends orders, receives fills and positions")
Rel(af_core, storage, "Reads and writes raw and cold data")
Rel(af_core, analytics, "Exposes metrics, logs, and result datasets")

---

Key Context Decisions

1. AlphaForge is a platform, not a single app

• It is multi-tenant and multi-user.
• It exposes a stable DSL and API surface for both humans and agents.

2. Execution engines are external and pluggable

• AlphaForge does not assume a specific engine.
• It connects to one or more engines via the Execution Gateway
  • e.g. Nautilus, custom engines.

3. Data providers are heterogeneous

• Exchanges, index vendors, and on-chain sources are heterogeneous.
• They are normalized through the Data and Feature Service.

4. Results are for humans and machines

• Outputs are designed for human consumption:
  • Dashboards
  • Notebooks
• And also for AI agents:
  • Structured APIs
  • Vector space over strategies and experiments