backtest-engine

Replay L1 NBBO data with deterministic fills and latency modeling.

Updated Mar 2, 2026
One-click install
npx skills add https://github.com/Leiisawesome/feelies --skill backtest-engine-leiisawesome
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: backtest-engine
Source: https://github.com/Leiisawesome/feelies/tree/main/.cursor/skills/backtest-engine
Command: npx skills add https://github.com/Leiisawesome/feelies --skill backtest-engine-leiisawesome

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Backtest Engine provides a deterministic, event-driven replay of L1 NBBO data with realistic fill and latency modeling, enabling researchers and engineers to validate strategies, explore execution realism, and test for lookahead bias and queue dynamics.

Core Features & Use Cases

  • Deterministic replay: bit-for-bit identical signals, orders, and PnL across runs
  • Latency injection and realistic fill models to quantify execution realism
  • Integration-ready: orchestrator-driven backtests with live-data compatibility
  • Use cases: building backtest infrastructure, evaluating cost models, testing for lookahead bias, simulating transaction costs

Quick Start

Run a backtest by wiring MarketDataSource events into the Orchestrator and observe deterministic PnL and signal outputs.

Frequently Asked Questions about backtest-engine

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a deterministic NBBO backtest with realistic latency modeling?

Run a deterministic NBBO backtest by wiring MarketDataSource events into an Orchestrator, which drives event-driven replay with configurable latency profiles and integrity checks to produce bit-for-bit identical signals and PnL.

What causes lookahead bias in an event-driven backtest and how do I test for it?

Lookahead bias in an event-driven backtest occurs when future data leaks into current signals; integrity checks across the replay pipeline detect causality violations and ensure strategies only access historical data.

Can I use this backtest engine for multi-symbol order routing simulation?

Yes, the backtest engine supports multi-symbol order routing simulation by replaying L1 NBBO data across multiple symbols simultaneously with consistent fill and latency models.

How does a SimulatedClock ensure deterministic backtest results across multiple runs?

A SimulatedClock ensures deterministic backtest results by advancing time strictly through replayed market events, guaranteeing bit-for-bit identical signal generation, order execution, and PnL across multiple runs.

What market data do I need to perform an NBBO replay and transaction cost simulation?

To perform an NBBO replay and transaction cost simulation, you need historical L1 NBBO data and trade data, alongside a SimulatedClock and an orchestrator-based execution backend to process orders.

Why are my backtest results changing on every run instead of staying deterministic?

Non-deterministic backtest results indicate pipeline integrity issues like lookahead bias or causality violations; running built-in integrity checks verifies event ordering and prevents data leakage.