Avg CAGR (overall value in the matrix): The overall CAGR is calculated in two stages: First, a time-weighted average per asset (longer backtests count more). Then, a time-weighted average across all assets. This prevents a short, accidentally good run from skewing the overall result.

Deduplication: Identical backtests (same asset, same period, same parameters) are counted only once — even if 100 users ran the same backtest.

Effective time span: Overlapping periods are merged (union). Two runs covering 2018–2024 and 2020–2023 result in 6 effective years, not 9.

Parameter variants: Different settings (e.g., RSI 14 vs. RSI 7) are broken down separately. The overall CAGR is calculated from all variants combined, weighted by time period.

Buy & Hold comparison: The B&H CAGR shows what would have happened if you simply bought and held the asset — no active trading. The strategy is only "better" if it beats B&H.