Ranking policy

Last updated: 15 September 2026

This page explains how the order on the front page is formed, what moves a tool up or down, and what cannot move it at all.

1. How a tool gets into the selection

We start from everything that plausibly belongs in the category: bots that act during a hand, and study tools that calculate between hands. Candidates come from vendor sites, poker forums and public discussion.

Twenty-seven tools were screened for this edition and fifteen were published. A tool reaches the published list when three things are true:

• it is available to buy or use today, not announced or abandoned;
• the vendor publishes enough about it to describe it factually — at minimum what it runs on and what it covers;
• it is distinguishable from what is already on the list, rather than a rebrand of it.

A tool is excluded when it exists only as a landing page, when its supported rooms and platform cannot be established from any published source, or when it turns out to be the same product under a second name.

2. How the score is formed

The score is an editorial judgement, informed by the criteria groups below and expressed on a ten-point scale. It is not the output of an arithmetic formula, and we will not pretend otherwise: a spreadsheet would give the weights an appearance of precision that the underlying data does not support, because vendors publish different things and half of them publish no price at all.

What the weights below do describe is honest and checkable: the relative importance we gave each group when comparing tools, and therefore what a vendor would have to change to move.

3. Criteria and weights

The six groups hold 45+ individual criteria. The same number appears in the page description, in the lead under the H1 and in the methodology section of the ranking — if you ever find those three disagreeing, one of them is a mistake and we want to hear about it.

4. The scale

Scores run to one decimal place on a ten-point scale. The published range in this edition is 7.4 to 9.6.

Nothing scores 10. A perfect score would claim there is nothing left to check, and in a category where most vendors do not publish a price, that claim would be false.

Nothing below 7.0 is published either — a tool that scores lower has not cleared the selection rules in section 1, so it does not appear at all rather than appearing at the bottom.

Bots and study tools are scored on two separate ladders, because they do different jobs. A 9.0 in one column and a 9.0 in the other are not the same claim, and the ranking says so where the two lists meet.

5. Equal scores

Where two tools land on the same score, the one that publishes more verifiable detail ranks higher. Documentation that can be checked beats a claim that cannot.

If they are still level, the older and more actively maintained product goes first.

6. What does not affect a position

• Payment. Nothing here is for sale: not a position, not a score, not a sentence. This site earns nothing from any vendor in the ranking, so there is no commercial reason for one to sit above another.
• A vendor's preference. No vendor sees the text before publication, and no vendor has approval over it.
• Contact from a vendor. Being in touch with us changes nothing. Sending a correction with a source changes the fact, not the position.
• Popularity. Forum sentiment and follower counts are not criteria.
• Our own convenience. A tool that is hard to research does not lose points for that; it loses points only where the missing information is itself the problem, such as an unpublished price.

7. Limits of the method

We rank on what vendors publish. That has real consequences and we would rather state them than hide them:

• A vendor that documents itself well looks better than one that is equally good but secretive. We think that is the correct bias for a buyer, but it is a bias.
• Vendor performance figures are claims. Except for one benchmark result produced against an independent opponent, nothing in this category has been verified by anyone.
• We do not test the tools ourselves for this ranking, so nothing here describes how a product behaves at your stakes, in your room, against your pool.
• Prices and coverage change without notice. Every fact carries the date it was checked.

8. When the ranking is reviewed

Quarterly, and sooner when something material changes. Every revision is recorded in the "what changed" section of the ranking with the reason and the consequence.

The next scheduled review is December 2026.

9. Enquiries from vendors

Vendors are welcome to send corrections, updated documentation or a product we have missed, through the support page or the submission form at the foot of the ranking.

What we will do: check it against your own published material and correct what is wrong.

What we will not do: share a draft before publication, agree a position in advance, remove a documented shortcoming, or accept payment for any of the above.