add_basketball_pick
ChatGPTOnly on explicit user request: log one pre-game match-winner prediction in your private Sportily prediction log. Bind forecastId from the displayed forecast to preserve identity/probability.
add_to_my_picks
ChatGPTSave a horse to the user's personal My Picks tracker so its result and ROI are tracked over time. Provide the Sportily raceId and horseId (from search_races / get_race_card / discover_* tools). The pick is auto-tagged with the source (claude or chatgpt) so the user can filter by who suggested it. Idempotent — saving the same runner twice will not duplicate it. Use only when the user explicitly asks to save, track, or add a pick.
analyze_my_picks
ChatGPTAnalyse the user's My Picks track record: overall strike rate, stake-weighted ROI, beat-SP (closing-line value), max drawdown and win/lose streaks, plus a performance breakdown grouped by a chosen dimension (going, course, country, distance, class, type, or tag) and plain-language best/weakest insights. Use when the user wants to understand patterns in their picks — e.g. 'how do my picks do on soft going' or 'where am I losing money'.
backtest_race_lens
ChatGPTBacktest one of the authenticated user's saved Race Lens filters, or a draft filterAst, over historical completed races. Race Lens model predicates use the base price-agnostic race model, not SmartPicks or Value Finder tipster percentages. Capped at 60 days for MCP (see the returned days / maxBacktestDays). Note: get_saved_lenses_summary reports metrics over each lens's own original lookback (which can be longer than 60 days), so its numbers will not exactly match this tool's — compare lenses using results computed by this tool over the same days. Returns performance, sample selections, and concrete filterAst improvement ideas to try next.
compare_runners
ChatGPTUse for horse-vs-horse comparison in a Sportily race. Returns focused evidence for selected runners; treat model percentages as directional analytics signals, not absolute truth.
delete_basketball_lens
ChatGPTOnly on explicit user request: delete your saved filter.
delete_basketball_pick
ChatGPTOnly on explicit user request: remove one of your recorded picks.
delete_tennis_lens
ChatGPTSoft-delete an owned Lens only on an explicit request. Supply its current version from list_tennis_lenses or get_tennis_lens; stale or missing Lenses are rejected.
delete_tennis_pick
ChatGPTDelete the user’s own unstarted tennis pick only on an explicit request. Supply its current version from list_tennis_picks; started picks cannot be deleted.
discover_racing_picks
ChatGPTUse by default when a user asks for horse-racing picks, good horses, value, SmartPicks, Value Finder, or Sportily analytics candidates. For a specific request, drill into that pick immediately. For a broad scan, score the candidates first and analyse the top 2-3 races in parallel. The goal is to test whether model% is realistic, not just restate it. Only return a bare list if the user explicitly asks for a list.
find_basketball_games
ChatGPTStored NBA schedules/results, grouped by America/New_York date; maximum 31 days. Experimental win probabilities are 0–1. This is not a live feed or bookmaker odds.
find_basketball_players
ChatGPTSearch stored player names and return IDs for player profiles. Case-insensitive literal name search, at most 100 results; no live roster feed.
find_basketball_teams
ChatGPTSearch stored team names and get IDs for profiles.
find_horses
ChatGPTUse by default when a horse-racing user names a horse, asks about a runner, or needs Sportily runner matching. Finds runners by partial horse name, or lists runners for a date so the MCP client can match a horse.
find_races
ChatGPTUse by default when a user asks what horse races are on, asks for a course without a race id, or needs Sportily race ids. Lists racecourses and race ids for a date.
find_tennis_events
ChatGPTFind stored tournament editions with matches in a London calendar-day range of at most 31 days. Includes event IDs, progress, remaining contenders and known finals. Results are stored coverage, not a complete tournament calendar.
find_tennis_matches
ChatGPTStored ATP/WTA singles fixtures and results by London calendar day (maximum 31 days). Prices include capture time and historical/prospective provenance. No live feed.
find_tennis_players
ChatGPTSearch stored player names; returns internal IDs for player and head-to-head tools.
get_basketball_game
ChatGPTStored game, ordered teams, box statistics, prior games, H2H and experimental forecast. Adds extras (roles and previous starts) and scheduled-game pregame rest, season strengths and last appearances; these are not confirmed lineups. Box statistics from this game are post-game information. Optional panel failures return null panels and researchWarnings while preserving the base game. No live feed.
get_basketball_player
ChatGPTStored game-by-game player box statistics and team history, plus a view with full-history season aggregates, recent form, notable nights and coverage exclusions. Both the existing games list and view.log return at most 100 recent games; view.logTotal and view.logTruncated describe the full log. Season aggregates use all stored completed games. Optional view failures preserve the base profile and return researchWarnings.
get_basketball_standings
ChatGPTEast/West standings calculated from stored completed regular-season games, with records, ranks and coverage. Optional season is the four-digit start year; defaults or falls back to the latest season with completed games. Inspect the returned season and fallbackFrom. Null means no completed season is stored; this is not a live or official standings feed.
get_basketball_team
ChatGPTStored games and observed players, plus a season view with records, scoring averages, quarter lines and averages, coverage caveats, upcoming games and player season averages. Optional season is the four-digit start year; unavailable seasons fall back to the latest played season. Player list is not a confirmed current roster. Optional season-view failures preserve the base profile and return researchWarnings.
get_external_race_context
ChatGPTFetch validated industry context for a Sportily race. Call automatically after get_race_card when externalResearchTargets are available. Returns status and extracted facts — fold insights into analysis without naming the source.
get_horse_profile
ChatGPTUse after finding a Sportily horse id when a horse-racing user asks for an opinion, profile, form view, or analysis of a named horse. Fetches career stats, upcoming races, and recent runs; open the race card before a race verdict.
get_race_card
ChatGPTUse by default when the user asks to analyse a horse race, Sportily race, runner, racecard, or named horse in a specific race. Fetches a pre-race analytics payload with race, runner, model, market price, stats, past-performance, response-shape, and external-research guidance. Target-race result fields are intentionally omitted so historical analysis can be run blind.
get_race_results
ChatGPTUse when the user asks who won, what happened, or wants explicit result data for a Sportily race. This is the post-race companion to get_race_card and exposes outcome fields intentionally hidden from the analysis payload.
get_saved_lenses_summary
ChatGPTList the authenticated user's saved Sportily Race Lens filters with compact backtest metrics and lens ids. Step 1 of the lens flow: use this before get_upcoming_lens_matches (to run lenses on upcoming races) or backtest_race_lens (to re-test over history) when the user wants to choose a specific saved lens, system, filter, or angle. Each lens's backtest metrics use its own stored lookback (backtest.days), which may exceed the 60-day MCP backtest cap.
get_tennis_event
ChatGPTRead a stored tournament edition, up to 400 matches, seeds, other edition IDs and captured main draw with byes and observation time when available. Surface ratings are current research, not historical pre-match features. A missing draw is null, never a fabricated bracket.
get_tennis_head_to_head
ChatGPTStored meetings, including separate retirement and walkover statuses. Without surface returns the existing meetings array. Supply surface for {matches, player1, player2, surface, view}, including a hypothetical current matchup and 52-week serve/return profiles on that surface. This is the stored corpus, not a claim of complete career coverage; hypothetical probabilities are not a forecast for a scheduled match.
get_tennis_lens
ChatGPTRead one owned Lens, its versions, run ledger and recorded forward selections with audited settlement.
get_tennis_match
ChatGPTStored match, pre-match form context, statistics, prices, H2H, derived research story and captured market/model chance history. Historical prices are retrospective, not evidence of an obtainable pre-match price. Forecast probabilities are 0–1, not percentages. Includes current result when known. No provider or model calls.
get_tennis_player
ChatGPTPlayer profile, stored ranking, surface records, serve/return counts and recent matches. Current profile aggregates must not be used as historical pre-match features.
get_tennis_rankings
ChatGPTLatest captured ATP or WTA rankings, with observation time.
get_upcoming_lens_matches
ChatGPTRun the authenticated user's saved Sportily Race Lens filters against upcoming races in the next 24 hours. Use when the user asks for their Race Lens results, saved-lens matches, custom systems, filters, angles, or upcoming runners found by their lenses. Race Lens model fields use the base price-agnostic race model and can differ from SmartPicks or Value Finder tipster percentages. User-specific: only the connected user's saved lenses are considered. Step 2 of the lens flow — call get_saved_lenses_summary first if you need lens ids or want the user to choose specific lenses.
list_basketball_lenses
ChatGPTRead your saved NBA probability filters.
list_basketball_picks
ChatGPTYour logged NBA predictions and how they turned out, from your private Sportily prediction log. Any price or weighting shown is what you entered manually.
list_my_picks
ChatGPTReturn the user's saved My Picks with live result and stake-weighted ROI for each, plus headline performance (strike rate, ROI points, beat-SP rate, drawdown, streaks). Use when the user asks what they have saved, how their picks are doing, or for their track record.
list_race_lens_filters
ChatGPTList the available Sportily Race Lens filter fields, operators, value types, and example predicates, plus the filterAst shape and rules. Use this when helping a user build or edit a lens. Model fields in this catalog use the base price-agnostic race model, not SmartPicks or Value Finder tipster percentages. Build flow: call this for the catalog, then backtest_race_lens with a draft filterAst to test the idea over history before saving. Note: a filterAst is a flat list of AND-ed predicates — there is no nested OR.
list_tennis_lenses
ChatGPTRead the connected user’s private saved tennis research rules and versions.
list_tennis_picks
ChatGPTRead the connected user’s private tennis prediction log: the player predicted for each match, notes and how each prediction turned out.
list_tennis_tournaments
ChatGPTDiscover enabled stored ATP/WTA tournaments, their internal IDs and countries.
preview_basketball_lens
ChatGPTFilter upcoming published forecasts. Returns frozen game/side/forecast identities. Preview does not record picks and is not a historical backtest.
record_basketball_lens_picks
ChatGPTOnly on explicit user request: atomically record the exact game/side/forecast IDs from a preview. Changed forecasts require a new preview. Never places bets.
remove_from_my_picks
ChatGPTRemove a previously saved horse from the user's My Picks tracker, identified by raceId and horseId. Use only when the user explicitly asks to remove or delete a saved pick.
save_basketball_lens
ChatGPTOnly on explicit user request: save a named NBA filter. Does not automatically track selections.
search_races
ChatGPTUse by default for horse-racing race lookup requests, including races, racecards, runners at a course, or requests that say "on Sportily". Finds Sportily race cards by date, course, and region.
update_basketball_pick
ChatGPTOnly on explicit user request: replace an owned prediction log entry’s notes and its permitted pre-game price/weighting. Omitted price/weighting remain unchanged; null clears them. Price and weighting lock at game start. Updates the Sportily prediction log only.