A pickleball rating anyone can check.
OPEN, the Open Pickleball Evaluation Network, is a free, open-source rating network that no company owns. We are building it in public, and we need players, organizers, and builders to help shape it.
Early development. No OPEN rating is live yet.
A rating decides which events you can enter, who you get matched with, and how you see your own progress. Something with that much influence should be open to inspection.
See it. Every rating will come with a clear explanation of how it was calculated.
Question it. When a number looks wrong, you will be able to challenge it and the method behind it.
Take it with you. You will be able to export your record and bring it to a different club, league, or app.
OPEN uses the familiar 2.0 to 8.0 scale and works alongside the ratings you already have, so nobody has to choose. Apps, leagues, clubs, and tournaments are partners, and anything they build can use OPEN for free.
A rating should show its work
OPEN's engine is designed as an Elo variant for doubles and singles. The engine and rating spec aren't published yet, so this sample shows the idea. Drag the final score and follow what happens to each player.
- Predict. Averaging each pair, Team 2 is rated 4.12 and Team 1 is rated 3.93, so Team 2 should win about 61% of the time.
- Compare. Team 1 won 11 to 7, an upset, so ratings move more than if the favorite had won.
- Weigh the margin. A 4-point game separates a close result from a decisive one.
- Update each player. Marek's rating is the least established of the four, so it moves the most.
What the engine is designed to account for
- Team rating
- Combines two partners' ratings to predict a match.
- Individual updates
- Gives each player their own rating change.
- Point differential
- Separates a close loss from a decisive one.
- Rating uncertainty
- Allows larger changes for new players.
- Recency
- Accounts for skill changing over time.
- Match weighting
- Can count verified tournament results differently from casual matches.
Every rating will also carry a reliability score from 0 to 100%. It will be earned through matches on OPEN, variety of opponents, and recent play. It can't be chosen.
A 70% favorite should win about 70% of the time.
Match outcomes are the ground truth. We will judge the engine by how well it predicts matches it hasn't seen, and new ratings will run in shadow mode before anyone relies on them, so early errors cost nobody.
- Predictive accuracy on future matches only, using Brier score and log loss
- Stability for established players and quick convergence for new ones
- Breakdowns by skill level, recreational and tournament play, new players, and separate groups of players
- Both teams will confirm a score or a verified organizer will submit it. Impossible scores will be rejected, duplicates caught, and outliers flagged.
Know this kind of math? Challenge our assumptions.
Five promises, written into a charter
Everyone who takes part agrees to the mission charter. It can be clarified or strengthened. It can't be weakened.
- Free
- No subscriptions, paid tiers, advertisements, or premium features, ever.
- Transparent
- Open-source code and a published, versioned rating spec.
- Community-run
- No person, company, or sponsor owns or controls it, including the founder.
- Portable data
- Players and leagues can export their data at any time.
- One standard
- Anyone can read and fork the code, but only the governing board changes the official spec, and only conforming tools can show an OPEN rating.
Players never pay
- Ratings are free forever.
- Players are never asked for money, in the app or at events.
- Donors, grant makers, and sponsors get no say over ratings.
- Every dollar in and out is published.
- No advertisements.
Where that stands today
Community control is the goal, and it is not true yet. The five-seat governing board has not been seated, and the founder is the only administrator of the GitHub organization, acting as interim steward.
The steward can't declare a rating official, can't sell or transfer the project to anyone but the board or an entity it controls, and must seat the first board before any OPEN rating goes live. Control of every project asset then passes to the board or a legal entity it controls. Read the governance plan.
What comes next, in rough order
No dates, because the work depends on volunteers.
-
Now
- Publish the first versioned rating spec and reference engine
- Test the engine on real match results in shadow mode
-
Before ratings go live
- Seat the governing board and hand over control
- Publish validation results
- Publish a privacy and data policy
- Define conformance rules and a policy for the OPEN name
-
Next
- Results API for platforms, clubs, and tournaments
- Ratings and stats API, free for any app
- Organizer tools to upload results, confirm starting ratings, and review flagged matches
-
Later
- Public app on web, Android, and iOS with player profiles, clubs, rankings, match history, data export, and a clear explanation of how each rating was calculated
Think something is out of order? Question a priority.
OPEN needs you, and you don't need to write code
It's early, so your input can shape what gets built. You don't need to apply or join the organization. Pick the line that sounds like you. The links open on GitHub, which needs a free account.
-
Players
Tell us what you need from a rating and what would make one easier to trust.
Send feedback -
Clubs and organizers
Share how you run events and record results.
Share how you run events -
App and league builders
Tell us what you'd need to send results to OPEN or show OPEN ratings in your product.
Send feedback -
Statisticians and rating enthusiasts
Review the methods, challenge the assumptions, and bring evidence for a better spec.
Propose a spec change -
Developers and designers
Help build the engine, APIs, and tools. The repositories are new, so early contributors help set their direction.
Read the contributing guide
Not sure where you fit? Ask a question on GitHub Discussions. OPEN is run by volunteers, so replies may take a few days. Please keep real player names, contact details, and match histories out of public posts.