🔥
FireForge Documentation
  • Overview
  • Origin
  • Methodology
  • Getting started
  • Strategy Finder
    • Settings
      • CEM - Define your objective
      • Metrics explained
      • Expiration & Selector
    • Results
      • Filters
      • Top Lists
      • CEM Graph
      • Heatmaps
  • Backtest
    • General
    • Strategy Info
      • Compute Dependencies
      • Backtest Settings
    • Results
Powered by GitBook
On this page
  • Custom Evaluation Metrics
  • Presets
  • ✨ Advanced Settings
  1. Strategy Finder
  2. Settings

CEM - Define your objective

PreviousSettingsNextMetrics explained

Last updated 7 months ago

Custom Evaluation Metrics

The Custom Evaluation Metric (CEM) is FireForge’s primary tool for evaluating strategies based on user-defined performance criteria. CEM serves as the core benchmark for ranking strategies according to individual goals.

  • Predefined Formulas: FireForge offers several CEM presets designed for common trading styles. For instance, the “Income Trading” preset focuses on strategies that provide steady returns, high win rates, and low drawdowns.

  • Advanced Customization: For advanced users, FireForge allows custom CEM formulas. Users can combine and weight various metrics (e.g., win rate, risk-of-ruin weight, drawdown) to fit specific trading styles and goals, giving complete flexibility in evaluating strategies.

Presets

Income Trading

Prioritizes steady, high win-rate returns with minimal risk.

Compounding 1

Rewards steady returns that support compounding with low BPR.

Compounding 2

Similar to Compounding 1 but favors strategies with stable, median returns.

Growth Under Stress

Focuses on high returns during volatile (high VIX) periods.

Steady Growth

Emphasizes consistent growth with low drawdowns and risk.

Hill Climber

Targets high peak capital growth, disregarding risk constraints.

Favorable Lottery

Seeks high upside potential with positive tail risk.

✨ Advanced Settings

The advanced settings allow you to provide your own, custom CEM formula based on the computed metrics for each strategy. This way, you can create your ver own, personalised results tailored to your needs.

You can use python's numpy functions for more complex operations.