Explanation¶
Understand concepts and design decisions behind FLOX. These pages are language-agnostic; code samples use tabs where they help.
Overview¶
| Topic | What You'll Learn |
|---|---|
| Architecture | How components fit together |
| Bar Types | Time, tick, volume, range, bps-range, Renko, Heikin-Ashi — when to use which |
| Disruptor Pattern | Why we use ring buffers (system-level, C++ internals) |
| Memory Model | Zero-allocation event delivery (system-level, C++ internals) |
| Integration Flow | End-to-end data flow through the system |
| Indicators | What each indicator measures and when to use it |
| Venue types | How a venue type (CEX, AMM, hybrid) drives pricing and settlement routing |
| Per-symbol scale | How fixed-point price/quantity scale is chosen per symbol for DEX-range tokens |
| On-chain order lifecycle | Pending, reverted, and gas-replaced states for DEX orders |
| Liquidity-provision signals | Provide and withdraw signals for AMM pool positions |
| Position valuation | The hook for nonlinear (LP, option) position valuation |
| AMM pricing in backtests | Constant-product pool pricing, slippage, and price impact for DEX swaps |
| Market-making quoter | Two-sided quote ladder with inventory skew and requote tolerance |
| AMM DEX connector | Presenting a constant-product pool through the connector interface |
| Replay-equivalence gate | The CI check that defends deterministic backtest replay |
| Bar-close ordering | Which aggregator fires first when several bars close on the same instant |
| Connectors | The native exchange connectors module and the interfaces it implements |
| MCP control plane | Why the AI-agent control surface is shaped the way it is |
| Venue-stack reproducibility | What makes a backtest result reproducible across machines |
| Build feature flags | Every CMake option, what it costs, and what it depends on |
When to read these¶
- Before diving deep into customization
- When you want to understand design trade-offs
- If you're debugging performance issues
A note on language¶
Some explanation pages — Disruptor Pattern, Memory Model — describe internals of the C++ engine itself. Code in those pages is C++ because that is the implementation. If you only use the Python, Node.js, or Codon bindings you can read them as background; the bindings expose the relevant behaviour through their own APIs without you needing to touch C++.