NBA · WNBA · EuroLeague
Know what's covered.
Connected data, a research projection and an official pick are three different things. This page explains which paths exist—not whether every game has complete data today.
A source-backed path exists. Check the game's source time and missing-data messages.
Useful to investigate and measure. Not permission to publish a pick or place a bet.
Existing NBA/WNBA records keep their own gates. They are not results for the new native model.
Feature parity, without hiding gaps
What can I do?
| Tool | WNBA | NBA | EuroLeague |
|---|---|---|---|
| Games, rosters and player historyWP01 / WP03 | Implemented | Implemented | Implemented |
| Injuries and starting lineupsWP01 / WP09 | Implemented | Implemented | Implemented |
| Recorded player-prop offersWP02 | Implemented | Implemented | Implemented |
| Saved quote checkpoints and book historyWP02 / WP05 | Implemented | Implemented | Implemented |
| Native gameplay projectionsWP04 | Research only | Research only | Research only |
| Player-out and starting-five hypothesesWP04 / WP08 | Research only | Research only | Research only |
| Native model versus frozen baselineWP05 / WP08 | Research only | Research only | Research only |
| Five human and AI research flowsWP06 | Paper only | Paper only | Paper only |
| Paper decisions and recorded actualsWP05 / WP06 | Paper only | Paper only | Paper only |
| Official published recommendationsWP07 / WP08 / WP09 | Separate legacy lane | Separate legacy lane | Not eligible |
| My Reads and exact-quote insightsWP03 / WP05 | Implemented | Implemented | Implemented |
| Backsolve and scenario arenaWP03 / WP04 / WP08 | Separate legacy lane | Separate legacy lane | Not yet connected |
| Market Memory and exact-player odds historyWP02 / WP03 | Implemented | Implemented | Implemented |
| Published formulas and public resultsWP03 / WP06 / WP07 | Separate legacy lane | Separate legacy lane | Not yet connected |
| Standalone player analysis and briefsWP03 / WP07 | Separate legacy lane | Separate legacy lane | Not yet connected |
| Interactive analyst chatWP03 / WP06 | Separate legacy lane | Separate legacy lane | Not yet connected |
| Ask question and split libraryWP01 / WP03 | Separate legacy lane | Not yet connected | Not yet connected |
| Public season archive and final participantsWP01 / WP03 | Implemented | Implemented | Implemented |
| Player directory and recorded performancesWP01 / WP03 | Implemented | Implemented | Implemented |
| Game emails and newsletter deliveryWP07 / WP09 | Separate legacy lane | Separate legacy lane | Not yet connected |
What each tool does—and does not prove
- Games, rosters and player history
- Native game/player identities and history are connected. A current roster is not proof that a player is available, starting, or appeared in a historical game.
- Injuries and starting lineups
- Each game must show source time, completeness and expected versus confirmed status. Missing injury rows never mean everyone is healthy.
- Recorded player-prop offers
- Exact-game quote inventory is connected for all three. A listed market key, missing quote response or supported stat does not establish a currently offered sportsbook line.
- Saved quote checkpoints and book history
- Saved checkpoints and opening/latest book anchors are read where recorded. An opening/latest pair is not a complete tick history, and missing past checkpoints cannot be reconstructed as pregame evidence.
- Native gameplay projections
- Eight stat families share a market-blind native research path. Source-ready output may be studied before it is validated; no native forecast is authorized as an official recommendation.
- Player-out and starting-five hypotheses
- Deliberately applied what-if scenarios compare player minutes and stat means with the retained source snapshot. They preserve source-reported OUT status, require explicit replacement starters, and never enter the canonical comparison or official recommendation ledger. New source reads can also affect the observed differences.
- Native model versus frozen baseline
- Prospective capture/comparison paths keep leagues, lineup certainty, checkpoints and model artifacts separate. Empty or provisional samples cannot establish a winning model or betting profit.
- Five human and AI research flows
- Shared paper-only guided flows preserve market blinding, human decisions and deliberate AI execution. Every selected quote still needs exact captured evidence; no agent places bets.
- Paper decisions and recorded actuals
- Twelve count-stat families are in scope. Complete final-game/participant evidence and the captured grading policy are required. Missing counts are unknown, not zero; paper grades are not sportsbook settlement guarantees.
- Official published recommendations
- NBA/WNBA legacy publications remain a separate gated population. Their presence does not promote the new native model. EuroLeague official publication remains ineligible pending its own validated model, priced evidence and publication gates.
- My Reads and exact-quote insights
- All three leagues can retain an exact immutable pregame quote receipt with a private human note. A real recorded offer is still required; no offer means nothing to bookmark. Account-save time remains separate from market time, and saved quotes are not graded calls or official recommendations. EuroLeague legacy graded calls remain unsupported.
- Backsolve and scenario arena
- The existing Backsolve/Arena routes remain separate NBA/WNBA tools. A EuroLeague game projection panel does not provide equivalent market inversion or arena scenarios.
- Market Memory and exact-player odds history
- All three leagues share the twelve-stat history route. EuroLeague preserves native IDs and distinguishes no recorded rows, an unbuilt summary artifact, and a failed source. Its first actual offered-market history remains an acceptance gap; an empty route does not prove complete sportsbook coverage.
- Published formulas and public results
- EuroLeague's historical odds route does not provide legacy formula records, official results or serving scorecards. Native model comparisons answer a different question; official public results remain gated.
- Standalone player analysis and briefs
- EuroLeague opens native game/player research. Its standalone saved player-analysis API, legacy analyst page and morning brief still do not have parity.
- Interactive analyst chat
- The existing NBA/WNBA analyst tools remain separate from the guided Lab agents. EuroLeague needs native identity, analytics and exact saved-evidence adapters before that tool bundle can be enabled.
- Ask question and split library
- The current Ask library is WNBA-specific. NBA and EuroLeague need their own verified identity and split-query contracts; the shared navigation must not present WNBA answers as another league's research.
- Public season archive and final participants
- EuroLeague's archive begins with the verified 2025–26 season. Historical games use exact final-game participants and actual minutes, not today's roster or injuries. A final boxscore is retrospective evidence; original pregame forecasts appear only where recorded.
- Player directory and recorded performances
- EuroLeague's directory verifies current registered rosters against the current season's clubs, preserving native player IDs. Profiles show bounded recorded game history and actual minutes for twelve stat families. Current membership does not prove past membership, availability, or complete career coverage.
- Game emails and newsletter delivery
- EuroLeague is still excluded by the current game-email and newsletter contracts. The shared website styling does not imply notification delivery parity.
EuroLeague official recommendations remain ineligible. All three leagues' new native forecasts are research-only; none has been promoted by this capability map.
Same stat contract across all three leagues
12 recorded stat families. 8 native forecasts.
History and paper actuals require the underlying counts and verified participants. Forecast support is separate from sportsbook availability.
| Stat | History / paper actuals | Native forecast |
|---|---|---|
| Points | Source required | Research only |
| Rebounds | Source required | Research only |
| Assists | Source required | Research only |
| 3-pointers | Source required | Research only |
| PTS + REB | Source required | Research only |
| PTS + AST | Source required | Research only |
| REB + AST | Source required | Research only |
| PTS + REB + AST | Source required | Research only |
| Steals | Source required | Not implemented |
| Blocks | Source required | Not implemented |
| Turnovers | Source required | Not implemented |
| STL + BLK | Source required | Not implemented |
EuroLeague's default provider request scope remains eight candidate markets. Steals, blocks and turnovers require a separate opt-in and observed provider coverage. Steals + blocks has no supported provider key in the current collector. No extra polling or paid scope is enabled by this page.
Additional market gaps we are keeping visible
These NBA source markets can appear as recorded rows but do not yet have a shared three-league history, native-forecast and paper-grading contract:
- Field goals made
- Free throws made
- Free throws attempted
- 3-point attempts
They remain parity requirements. A generic recorded row is not equivalent to a supported predictive model.
Before you trust a number
Evidence still decides.
- For an offer: verify the exact player, game, sportsbook, stat, line, side, price and observation time. Discovery alone is not a quote.
- For a projection: inspect lineup certainty, missing inputs, projected minutes, outcome ranges and the captured model version. A mean is not an over probability.
- For a result: keep final participants, DNP handling, source completeness and correction policy. Missing stats are not zeroes.
- For a better model: require prospective paired samples, calibration and a registered promotion rule. Do not pool leagues or compare expected lineups with confirmed-lineup results.
Implementation registry: three-league-capabilities.v1. This page performs no live health check or provider poll. Current game coverage and timestamps belong to the connected research and market pages.