METHODOLOGY
How We Test AI Trading Tools
We buy our own subscriptions. We are not paid for placement. Tools are ranked by score, never by commission. If a tool is bad, it gets a low score and we tell you why.
The five pillars
| Pillar | Range | What we look for |
|---|---|---|
| Signal Quality | 0 to 20 | High score: AI signals that are accurate, timely, and actionable across real market conditions. Drags down: Generic alerts, delayed signals, or AI that amounts to a moving-average wrapper. |
| Transparency | 0 to 20 | High score: Clear methodology, disclosed data sources, and honest statements about limitations. Drags down: Black-box claims with no explanation of how signals are generated or what data feeds them. |
| Price to Value | 0 to 20 | High score: A subscription cost that matches the features, data quality, and depth delivered. Drags down: A high price tag for a thin feature set, or hidden upsells for core functionality. |
| Usability | 0 to 20 | High score: A clean interface, a manageable learning curve, and a mobile experience that works. Drags down: A cluttered dashboard, a steep setup process, or a mobile app that is an afterthought. |
| Community Verdict | 0 to 20 | High score: Consistently positive user ratings across Trustpilot, G2, and Google. Drags down: A pattern of complaints about billing, data quality, or customer support. |
What a score actually means
| Score | Meaning |
|---|---|
| 90 to 100 | Best in class |
| 75 to 89 | Strong, with known trade-offs |
| 60 to 74 | Works for a narrow use case |
| 40 to 59 | Proceed with caution |
| Under 40 | We would not spend our own money here |
How the finder ranks tools
You answer five questions. Each answer carries a set of tags, things like "real-time," "swing-trading," or "beginner-friendly." Every tool in our database has its own tags. We count how many of your tags each tool matches, sort by that count, and show you the top three. Ties break on Alpha Score, then on user rating.
No tool can pay to rank higher. The matching is pure tag overlap. A tool with no affiliate program and a tool with one are scored by the exact same criteria.
What we will not do
- − No predictions about what the market will do.
- − No return promises or implied performance claims.
- − No fabricated quotes, testimonials, or user reviews.
- − No paid placement. No tool can buy a higher rank.
- − No tools we have not used ourselves.
Rated, unrated and newcomer
Every tool in a finder sits in one of three states, and the page tells you which one.
- Rated
- At least one public platform (Trustpilot, G2, Google, the App Store or Reddit) holds 10 or more reviews of the tool. We show that platform's score and review count exactly as published, one badge per platform, and never average them.
- Unrated
- The tool has been around long enough to be reviewed but no platform has reached 10 reviews, or we have not yet verified its ratings. Below 10 reviews, one angry or one glowing customer can swing the number, so we show nothing rather than a score that means little.
- Newcomer
- The tool is too new to carry public ratings. It gets a NEWCOMER badge instead of a rating. In the matching, a tool with a real Alpha Score always ranks ahead of one without, so a newcomer never outranks a tested tool on an empty score.
When we get it wrong
Errors are fixed in place. We do not delete the mistake and pretend it never happened. When we correct a score, a price, or a feature description, we note the change with a date and describe what was wrong. The correction stays visible so you can see the history of our work.
If you spot an error, contact us and we will verify and correct it. Transparency matters more than looking infallible.