How we rank AI trading tools

We start with a job, define the comparison before placing tools, and keep official claims separate from local test results.

1. Define the task.Research, strategy discovery, backtesting, paper trading, automation, developer extensibility, beginner setup, or overall self-hosted fit.
2. Set the evidence bar.Official repositories establish documented capabilities. Reproducible local runs establish what happened on our benchmark.
3. Compare fit, not returns.We compare access, workflow, repeatability, setup, and evidence quality. Investment returns are never a ranking factor.
4. Show limits.Unknown fields stay unconfirmed. A failed or partial test remains visible.