1.7.2

Pairwise Glicko Ranking 1.7.2

A patch release focused on matchmaking correctness and the stability progress bar.

Fixes

• Matchmaker now uses real stats from the start of a session: the arena preloads note IDs for the whole cohort before selecting the first pair. This lets the information-gain matchmaker immediately use existing ratings and uncertainty (instead of treating notes as "unknown" until they are opened), so the first few comparisons are chosen more accurately.

• Stability progress bar no longer overestimates convergence: very stable notes no longer "mask" unstable ones in the progress calculation. Sigma is now clamped at your stability threshold when computing overall stability, preventing the bar from reaching 100% too early while some notes are still highly uncertain.

Full changelog

1.7.1...1.7.2

1.7.1

Pairwise Glicko Ranking 1.7.1

A reliability, polish, and bugfix release.

Improvements

• Autosave cohort settings: The cohort options modal will now autosave when in "edit" mode, aligning with user expectations and the way Obsidian settings tabs work.
• More efficient frontmatter rank updates during sessions: when writing frontmatter for the currently displayed pair, ranks are now computed only for the changed pair (instead of recomputing ranks for the whole cohort every time).
• Added "Uncertainty" (σ) frontmatter option: you can now choose to write the Glicko uncertainty value to frontmatter (alongside rating, rank, matches and wins). This is configurable in global defaults and can be overridden per cohort.

Fixes

• Rounding consistency: Rating is rounded before calculations, so that ranks accurately reflect the rating you see in frontmatter.
• Deleted note handling during sessions: fixed a bug when a user deleted a note currently being compared resulting in a different note being shown in the session bar and the leaf that contained the deleted file.
• Prevent accidental repeated undo: Backspace is now included in the key repeat guards, so holding Backspace will not trigger multiple undo actions.

Full changelog

1.7.0...1.7.1

1.7.0

Pairwise Glicko Ranking 1.7.0

This release adds optional post-session reports, generating a markdown summary of what changed during your rating session (gains/losses, surprises, new entrants, leaderboard snapshots, match log, and more).

What's new

• Post-session reports (per cohort): when a session ends, the plugin can now generate an Obsidian note summarising the session.

  • Includes session duration, match count, unique players, draw rate, new entrants, stability change (sigma delta), biggest gains/losses, most surprising results, leaderboard snapshots, and a match log.
  • Reports are disabled by default - you can enable them per cohort in Cohort options.

• Customisable report templates (placeholders): reports are generated from a template using {{glicko:...}} placeholders.

  • Supports both scalar placeholders (like {{glicko:match-count}}) and pre-rendered markdown blocks (like {{glicko:leaderboard-table}}).
  • Many list/table placeholders support limiting with :N (for example {{glicko:gains-list:5}}).

Improvements

• Cohort creation/edit flow extended: Cohort options now include a "Generate post-session report" toggle and per-cohort overrides for folder/name/template.

Breaking changes / migration notes

• Plugin ID rename: the plugin ID has changed to pairwise-glicko-ranking.
  • Obsidian will treat this as a new plugin, which can make your existing settings/ratings appear "missing" until migrated.

Migraton

1. Copy (or rename) the plugin data folder:
   • from .obsidian/plugins/pairwise-elo-ranking/
   • to .obsidian/plugins/pairwise-glicko-ranking/
2. Remove the WilliamsJack/pairwise-elo-ranking entry in BRAT settings
3. Add the new URL https://github.com/WilliamsJack/pairwise-glicko-ranking to BRAT

Full changelog

1.6.0...1.7.0

1.6.0

Pairwise Glicko Ranking 1.6.0

A major migration from the classic Elo rating system to the Glicko rating system.

About Glicko-1

Unlike Elo, Glicko tracks both a rating and an uncertainty value ("sigma"). This means the plugin can move new or under-compared notes quickly (high uncertainty), while keeping well-established notes stable (low uncertainty) without you having to tune a K-factor. That uncertainty also feeds directly into the information-gain matchmaker.

The result is that your cohorts converge on a stable ranking much faster.

What's new

• Glicko: Glicko replaces Elo, adding uncertainty ("sigma"/rating deviation/RD), which controls how quickly a note's rating can move.
• Simplified settings: With Glicko-1, new/uncertain notes move more, and stable notes move less. Uncertainty replaces the need for a tunable K-factor and other convergence heuristics.
• Rebrand: the plugin UI, wording, and defaults have been updated to reflect the Glicko rating system. Commands are now named "Pairwise Glicko Ranking: ...".
• Reset note rating: a new command lets you reset the active note back to a fresh baseline ($1500$ rating, full uncertainty), either for one cohort or across all cohorts that contain it.
• Configurable note ID property name (with migration): you can now change the name of the property or HTML comment used for note IDs (e.g. glickoId). When you change it, the plugin will scan your vault and migrate existing IDs so your rating progress is kept.

Improvements

• Progress bar settings are easier to find: "Highlight surprising results" and "Stability threshold" have been moved out of the accordion and into a dedicated "Progress bar" section in settings.

Breaking changes / migration notes

• CSS class renames: UI classes have been renamed from the elo-* prefix to glicko-* (e.g. .elo-session-bar - .glicko-session-bar). If you have custom CSS targeting the old classes, you will need to update it.
• Defaults renamed to glicko*: default frontmatter property names and the default note ID property now use glicko... naming (for new users / new configurations).
• Backwards compatibility for existing vaults: to avoid breaking existing ratings, upgraded installations default the note ID property name to "eloId" if you do nothing. You can migrate to "glickoId" from Settings when ready.

Full changelog: 1.5.0...1.6.0

1.5.0

Pairwise Elo Ranking 1.5.0

This release introduces a new information-gain matchmaker, a progress bar, and a better undo flow.

What's new

• Information-gain matchmaking: pair selection now explicitly targets the most informative comparisons by tracking per-note uncertainty ("sigma") and choosing the pair with the highest expected information gain.
• Stability progress bar in the arena: the session bar now includes a thin progress bar showing how close the cohort is to stabilising overall.
• Surprise highlight: when you pick an unexpected outcome (relative to current ratings), the progress bar can briefly wobble to flag the surprise (optional), alerting you that you made an unexpected choice (giving you the option to undo it if it was a mistake).

Improvements

• Undo now restores the previous pair: when you undo a match, the arena re-opens that same left/right pair so you can immediately re-evaluate and play the rematch.

Changes to matchmaking settings (breaking)

• The old "Matchmaking heuristics" settings have been removed (similar-ratings sampling, low-matches bias, upset probes), as this functionality is replaced by information-gain matchmaking.

Full changelog: 1.4.1...1.5.0

1.4.1

Pairwise Elo Ranking 1.4.1

A small reliability release focused on improved handling of note deletions during an active session.

Fixes

• The arena will no longer suggest comparisons with empty panes if a note in the cohort is deleted during a session.
• The session will now end automatically if deletions leave fewer than two notes in the cohort.
• If either the currently displayed left or right note is deleted, the arena immediately picks a fresh pair to continue.

Full changelog: 1.4.0...1.4.1

1.4.0

Pairwise Elo Ranking 1.4.0

This release brings Obsidian Mobile app support, with a session layout and controls designed for small screens.

What's new

• Mobile/phone support: the plugin is no longer desktop-only and can now run on Obsidian Mobile.
• Phone-friendly arena controls: on phones, the session bar switches to a stacked layout with larger tap targets and a single title line showing which note you are currently viewing.
• "Switch" + contextual "Win" buttons (phone mode):
  • Switch toggles between the left and right note.
  • Win chooses the winner for whichever note is currently visible.
  • Draw, Undo, and End remain available as dedicated buttons.

Improvements

• Session layout is automatically adapted on phones: the arena uses a two-tab workflow on phones (your desktop layout setting is preserved for desktop/tablet).
• More robust pop-out handling: when pop-out windows are not available (e.g. mobile), choosing "new window" now falls back cleanly to tabs instead of erroring.

Full changelog: 1.3.5...1.4.0

1.3.5

Pairwise Elo Ranking 1.3.5

A small performance and polish release focused on tag cohorts and UI stability.

Improvements

• Faster tag cohort creation: the tag picker now uses Obsidian's MetadataCache instead of scanning notes, making the "New: from #tags..." flow much snappier in larger vaults.
• Tag hierarchy support in cohort matching: nested tags like #a/b/c are now treated as #a, #a/b, and #a/b/c when resolving tag-based cohorts. This aligns with how Obsidian treats nested tags and better matches user expectations.
• Smoother Cohort options warning UI: the warning about auto-scroll + synchronised scrolling now slides in and out smoothly, rather than immediately appearing and causing the modal layout to jump.
• Automated releases: to provide extra confidence that the published release files are built directly from the open-source code in this repository, releases are now built by a GitHub Actions workflow when a version tag is pushed.

Breaking Changes

• CSS class rename: .cohort-item-info has been renamed to .elo-cohort-item-info so it can be better targeted by custom CSS. If you have custom snippets or themes targeting the old class, update them to the new name.

Full changelog: 1.3.4...1.3.5

1.3.4

Pairwise Elo Ranking 1.3.4

A workflow and safety release focused on smoother side-by-side comparisons and keeping templates out of your cohorts.

Improvements

• Synchronised scrolling (per cohort): you can now keep the left and right notes scrolled together during a session, making it much easier to compare long notes line-by-line. Enabled per cohort in "Cohort options".
• Templates folder exclusion: set a "Templates folder" in settings to automatically exclude those notes from all cohorts (vault, folder, tags, Bases, manual). Prevents Elo IDs and frontmatter stats from being written to templates by accident.

Full changelog: 1.3.3...1.3.4

1.3.3

Pairwise Elo Ranking 1.3.3

A reliability and data-integrity release focused on Elo IDs - making it easier to keep cohorts clean when notes are copied, and safer to switch where IDs are stored.

What's new

• Optional Elo ID migration when changing "Elo ID location": when you switch between storing IDs in frontmatter vs the end-of-note HTML comment, the plugin can now scan your vault and offer to move existing IDs for you.

• Duplicate Elo ID detection and guided resolution before sessions start: if two or more notes in your chosen cohort share the same Elo ID (common after duplicating/copying a note), the session will pause and prompt you to choose which note keeps the ID. The ID is removed from the others so they can be assigned fresh IDs as needed.

Improvements

• More robust HTML comment ID handling: better parsing and cleanup of end-of-note <!-- eloId: ... --> markers, including handling multiple markers and preserving surrounding whitespace more carefully when removing or rewriting them.

Full changelog: 1.3.2...1.3.3