How the BotFinder score works
The score measures how much evidence stands behind a strategy's numbers. It does not predict returns, and a high score is never a recommendation to buy.
The five components
Every listing shows its own breakdown — hover the score on any strategy page. The five parts add up to 100:
- Performance data — up to 35. How many of the five core figures (win rate, max drawdown, return, profit factor, Sharpe) the source actually published. A vendor who discloses two of five scores lower than one who discloses all five, whatever the values are.
- Community — up to 25. The rating and the number of reviews collected from the source. Capped at 200 reviews: beyond that, more reviews tell us nothing further.
- Evidence — up to 20. Whether performance figures exist at all (12) and whether a substantive description of the strategy was found (8).
- Recency — up to 10. How long ago the listing was last collected. Under six months scores full; over a year scores 2.
- Verification — up to 10. The verification tiers recorded for the strategy. In practice this component is near zero for almost every listing today — see the honesty note below.
Verification statuses
Separate from the score, every strategy carries one status:
- Live Verified — the figures come from a third party that read the trading account itself. The strongest evidence we publish, and still historical.
- Backtest Only — a simulation over past data, published by the vendor. Backtests are trivially optimised after the fact.
- Publisher Claimed — the vendor states the figures; we recorded the claim and could not confirm it.
- Unverified — collected, with no evidence grade established.
- Demo — a generated row used to test the catalogue. Never indexed, always labelled.
These are never blended. A backtest and a broker statement do not become one averaged number, and they never share a label.
What we do not do
- We do not audit vendors. No account is inspected by us, no ownership is confirmed, no capital is checked.
- We do not hash-lock or tamper-proof trade histories. We record what a source published and when we read it.
- We do not test strategies against prop-firm rule sets. Where compatibility is shown, it comes from the source, not from a test we ran.
- We do not gate reviews behind a verified live position. We show the reviews we collected, and say so when there are none.
- We do not favour our own strategies. Some listings (publisher “H2WI”) are run by BotFinder’s publisher, on demo accounts, with figures computed from their own trade journals and verified by no third party. They are graded by the same rules as everything else and flagged as ours.
Where the data comes from
Listings are collected from public sources — vendor marketplaces, public performance directories and open-source repositories. Every row records its source, the exact URL and the collection date, all three visible on the strategy page so you can open the original and judge it yourself.
Some written analysis is produced with the assistance of language models from the collected data. It is a reading aid, it can be wrong, and it is never a recommendation.