Follow recent improvements to Pokémon card research, AI rankings, grading tools, Tracker accounts and website security. Newest versions appear first.
Latest version · 9.6.0Updated runtime and build dependencies
Updated affected software dependencies and retested the website before the Portfolio release.
- Applied compatible dependency patches while preserving the website's storage and account configuration.
- The checked release lockfile passed the current dependency audit with no known vulnerabilities.
Live slab values and personal grading analysis
Portfolio now distinguishes completed slabs from raw cards, refreshes company-specific graded values and keeps a separate record of your grading checks and returned results.
- Completed slabs show their grading-company logo, final grade and a distinct violet tile.
- Free public sold-price evidence refreshes actual company-and-grade values; unavailable evidence stays missing and previous successful quotes are dated.
- Raw-only totals and possible PSA 10 totals include ungraded cards only, while completed slabs have their own combined total.
- Added Grading Analysis to compare expected and returned grades and estimate future grades from condition checks.
- Personal learning uses checks saved before results arrive and activates only after enough later results beat the validation baseline.
- Preserved saved records, previous entered values, images, notes and costs, and restricted Content Studio to the configured owner account.
Action-based AI Advisor outlooks
Advisor cards now show a practical buying tier, target price, card-specific risk and modelled short-, medium- and long-term outlook instead of repeating a technical identity warning.
- Added Buy now, Buy at target price, Buy with caution, Buy later and Pass for now tiers.
- Added non-zero structural price outlooks when recent price history is thin, without inventing graded-card values.
- Separated raw-market analysis from grading-profit analysis so missing graded prices are shown as a limitation rather than the recommendation itself.
- Added guarded 90-day, 12-month and 24-month learning models trained only on verified later price outcomes.
- Added small, time-decayed preference signals from saved picks, Wishlist and Portfolio actions; these signals never count as proof of profit.
Personal public Wishlist links
Each Tracker account now has one stable public Wishlist link, with a simpler sharing process and cleaner public-page navigation.
- Fixed administrator and member links so they always open the Wishlist belonging to that account.
- Added one-click link copying from the account and private Wishlist areas.
- Simplified the public Wishlist menu without changing any saved Portfolio or Wishlist data.
- Restored the shorter homepage headline.
Higher-quality collector search
Removed repetitive promotional listings from public collector conversations so search results are more useful and easier to trust.
- Applied the same rolling 30-day window used by Popular Cards.
- Required card-query relevance before a Bluesky post can appear.
- Removed promotional posts and collapsed repeated links before displaying results.
Cleaner social signals and broader card research
Improved Popular Cards so recent, relevant collector conversation matters more while duplicated or promotional posts count less.
- Added a rolling 30-day freshness score, exact-card relevance checks and duplicate-post suppression to Popular Cards.
- Added Reddit public RSS as an optional, rate-limit-safe Collector Search source; it fails independently and never receives cookies or account credentials.
- Kept all new public research outside AI Advisor, model training and private Tracker data.
- Reviewed the Last 30 Days Codex plugin and reused only its safe research principles; the website does not install its browser-cookie, credential or system-tool integrations.
A clearer public changelog
Added this version-by-version update page so collectors can quickly understand how the website, AI tools and security have improved.
- Grouped recent improvements into a simple public timeline.
- Added the changelog to Discover and the website footer.
- Explained technical and security work without exposing unnecessary detail.
Clearer language and easier navigation
Improved the wording across the homepage and research tools so collectors can understand the evidence, risks and purpose of each feature more quickly.
- Rewrote key Homepage, Popular Cards, AI Advisor, Card Insight and methodology text.
- Kept useful market terms such as Price History and Pokémon Card Index.
- Moved Collector Search into the Discover menu and removed the separate site-wide search button.
Content Studio and individual Tracker accounts
Added protected content editing for the website owner and separate Tracker spaces for individual collectors.
- Added Demand Lab Studio for editing approved website text and images.
- Added username-and-password accounts with a separate Portfolio and private Wishlist for each collector.
- Added a shareable public Wishlist for each account while keeping private Tracker data separated.
- Improved the homepage card animation and moved card details outside the artwork.
Safer and more dependable software
Updated website dependencies and automated workflows after security alerts were reported. The fixes were tested with the market, search, grading, AI and Tracker features before completion.
- Updated affected production and development packages to supported versions.
- Updated the automated build actions used by the repository.
- Closed the resolved dependency alerts after validation and kept automated monitoring enabled.
More reliable AI Advisor and Popular Cards
Strengthened the daily ranking engines so they produce complete, varied lists while remaining honest when card evidence is incomplete.
- Kept AI Advisor at 16 cards when current graded-price evidence is temporarily limited.
- Prevented cards with weak identity or price evidence from receiving an unjustified machine-learning score.
- Improved Popular Cards rotation so visitors see more fresh candidates without exceeding 16 cards.
- Saved recommendation checkpoints in smaller batches so future outcome comparisons remain dependable.
Collector Search
Added one research search covering Pokémon card records, Demand Lab tools, public collector discussion and independent reference material.
- Connected several free public research sources without requiring visitor accounts or paid API keys.
- Kept research results separate from AI Advisor’s decision-making evidence.
- Added source labels, input limits and independent error handling for clearer and safer results.
Market experience and portfolio history
Made the market pages easier to explore and added a lasting daily record of Portfolio value.
- Added interactive graded-market-value scenarios and clearer card readouts.
- Reduced decorative motion while preserving useful typing, pointer and value animation.
- Added daily Portfolio value history without changing saved card-price records.
Trust and evidence upgrade
Rebuilt important grading and recommendation checks around exact card identity, transparent evidence and conservative costs.
- Added exact-card identity checks before graded prices can affect a decision.
- Added evidence confidence, anomaly handling and an Unknown risk result when information is too weak.
- Improved grading-profit calculations to include complete grade chances and common UK costs.
- Preserved recommendation inputs and later outcomes so past predictions can be reviewed fairly.
How changes are checked
Updates are tested locally before release. When a change affects market calculations, search, AI rankings, accounts or saved cards, those areas receive additional checks. Changelog entries describe completed work and do not promise a particular card price, grade or investment return.