Previous value: -"Ignored. Intelligence always reports the book the assessment was priced against. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, westgate, wynn, south_point, stations, hardrock, betonline, betr, betrivers."New value: +"Ignored. Intelligence always reports the book the assessment was priced against. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, westgate, wynn, south_point, stations, hardrock, betonline, betr, betrivers, lowvig."

Previous value: -"Player-prop rows: Owls prop category (1:1 counting stats). MLB: strikeouts_pitcher, hits, rbis, runs, strikeouts_batter, hits_allowed, earned_runs, outs_recorded. NFL/NCAAF: passing/rushing/receiving_yards, receptions. NBA/NCAAB: points, rebounds, assists, steals, blocks, threes_made. NHL: goals, hockey_assists, shots_on_goal. Main-line rows (bet_type set): the main-line family — h2h, spreads, or totals."New value: +"Player-prop rows: prop category (1:1 counting stats). MLB: strikeouts_pitcher, hits, rbis, runs, strikeouts_batter, hits_allowed, earned_runs, outs_recorded. NFL/NCAAF: passing/rushing/receiving_yards, receptions. NBA/NCAAB: points, rebounds, assists, steals, blocks, threes_made. NHL: goals, hockey_assists, shots_on_goal. Main-line rows (bet_type set): the main-line family — h2h, spreads, or totals."

Previous value: -"Ignored. Intelligence always reports the book the assessment was priced against. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline."New value: +"Ignored. Intelligence always reports the book the assessment was priced against. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, westgate, wynn, south_point, stations, hardrock, betonline, betr, betrivers."

Previous value: -"Natural-language matchup preview — form, context, what to watch. Search-backed context overlay on every event sport (MLB, NFL, NCAAF, NBA, NCAAB, NHL, soccer, tennis); sport-shaped queries. Null when no narrative has been generated."New value: +"Natural-language pre-game matchup preview — form, context, what to watch. Written before kickoff by a Search-backed overlay on every event sport (MLB, NFL, NCAAF, NBA, NCAAB, NHL, soccer, tennis); sport-shaped queries. The request path only reads the stored row. Null when the pre-kickoff job has not run or the fixture was not eligible."

Previous value: -"Event-level factual matchup chips (injuries, recent form, lineup/availability news — sport-shaped). Search-backed context overlay; may populate on first get_intelligence, later calls reuse the stored row."New value: +"Event-level factual pre-game matchup chips (injuries, recent form, lineup/availability news — sport-shaped). Written before kickoff; the request path only reads the stored row."

Previous value: -"Forecasted wagers for this event (same object as list_forecasts) — a model prediction, not a beat-the-market claim. Player-prop rows on team sports; tennis moneyline via bet_type ML_P1/ML_P2. selected marks the top-conviction rows on the daily slate; the rest of the event catalog is included with selected:false. Read p_hit as P(the picked outcome hits). How + field catalog: https://lumify.ai/docs/forecasts"New value: +"Forecasted wagers for this event (same object as list_forecasts) — a model prediction, not a beat-the-market claim. Player-prop rows on team sports; tennis main-line via bet_type ML_P1/ML_P2 (moneyline), SPREAD_P1/SPREAD_P2 (game handicap), OVER/UNDER (total games). selected marks the top-conviction rows on the daily slate; the rest of the event catalog is included with selected:false. Read p_hit as P(the picked outcome hits). How + field catalog: https://lumify.ai/docs/forecasts"

Previous value: -"Main-line token when this row is a main-line forecast — ML_P1/ML_P2/ML_HOME/ML_AWAY/ML_DRAW, SPREAD_P1/SPREAD_P2/SPREAD_HOME/SPREAD_AWAY, or OVER/UNDER (same vocabulary as get_intelligence bets[].bet_type and list_ev). Null on player-prop rows — that's the discriminator between the two row shapes. Tennis moneyline rows set ML_P1 / ML_P2."New value: +"Main-line token when this row is a main-line forecast — ML_P1/ML_P2/ML_HOME/ML_AWAY/ML_DRAW, SPREAD_P1/SPREAD_P2/SPREAD_HOME/SPREAD_AWAY, or OVER/UNDER (same vocabulary as get_intelligence bets[].bet_type and list_ev). Null on player-prop rows — that's the discriminator between the two row shapes. Tennis sets ML_P1/ML_P2 (moneyline), SPREAD_P1/SPREAD_P2 (game handicap), and OVER/UNDER (total games)."

Previous value: -"Forecasted wagers for this event (same object as list_forecasts) — a model prediction, not a beat-the-market claim. Player-prop rows today; schema also carries main-line rows via bet_type (none shipped yet). selected marks the top-conviction rows on the daily slate; the rest of the event catalog is included with selected:false. Read p_hit as P(the picked outcome hits). How + field catalog: https://lumify.ai/docs/forecasts"New value: +"Forecasted wagers for this event (same object as list_forecasts) — a model prediction, not a beat-the-market claim. Player-prop rows on team sports; tennis moneyline via bet_type ML_P1/ML_P2. selected marks the top-conviction rows on the daily slate; the rest of the event catalog is included with selected:false. Read p_hit as P(the picked outcome hits). How + field catalog: https://lumify.ai/docs/forecasts"

Previous value: -"Main-line token when this row is a main-line forecast — ML_P1/ML_P2/ML_HOME/ML_AWAY/ML_DRAW, SPREAD_P1/SPREAD_P2/SPREAD_HOME/SPREAD_AWAY, or OVER/UNDER (same vocabulary as get_intelligence bets[].bet_type and list_ev). Null on player-prop rows — that's the discriminator between the two row shapes. No sport ships a non-null bet_type here yet."New value: +"Main-line token when this row is a main-line forecast — ML_P1/ML_P2/ML_HOME/ML_AWAY/ML_DRAW, SPREAD_P1/SPREAD_P2/SPREAD_HOME/SPREAD_AWAY, or OVER/UNDER (same vocabulary as get_intelligence bets[].bet_type and list_ev). Null on player-prop rows — that's the discriminator between the two row shapes. Tennis moneyline rows set ML_P1 / ML_P2."

Previous value: -"Forecasted Over/Under player-prop wagers for this event (same object as list_forecasts). selected marks the top-conviction rows on the daily slate; the rest of the event catalog is included with selected:false. Read p_hit as P(side hits the posted line) from the player's shrunken rate. How + field catalog: https://lumify.ai/docs/forecasts"New value: +"Forecasted wagers for this event (same object as list_forecasts) — a model prediction, not a beat-the-market claim. Player-prop rows today; schema also carries main-line rows via bet_type (none shipped yet). selected marks the top-conviction rows on the daily slate; the rest of the event catalog is included with selected:false. Read p_hit as P(the picked outcome hits). How + field catalog: https://lumify.ai/docs/forecasts"

Added value: +{
+  "description": "Main-line token when this row is a main-line forecast — ML_P1/ML_P2/ML_HOME/ML_AWAY/ML_DRAW, SPREAD_P1/SPREAD_P2/SPREAD_HOME/SPREAD_AWAY, or OVER/UNDER (same vocabulary as get_intelligence bets[].bet_type and list_ev). Null on player-prop rows — that's the discriminator between the two row shapes. No sport ships a non-null bet_type here yet.",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Previous value: -"American price for the chosen side, keyed by bookmaker slug (e.g. {draftkings: -453}). Posted market price for the side the model picked."New value: +"American price for the picked outcome, keyed by bookmaker slug (e.g. {draftkings: -453}). Posted market price for the side/bet_type the model picked."

Previous value: -"Signed contributions that moved p_hit. Read input as the player's measured rate and effect as how far that rate lifts the chosen side versus a league-mean player."New value: +"Signed contributions that moved p_hit. Read input as the measured rate/rating behind the pick and effect as how far it lifts the chosen outcome versus a baseline."

Previous value: -"Signed shift in p_hit versus a league-mean player on the same line. 0.0 on yardage markets."New value: +"Signed shift in p_hit versus a baseline case on the same line. 0.0 on yardage markets."

Previous value: -"e.g. player.l10_rate — the recent-form rate that fed Poisson."New value: +"Prop rows: typically player.l10_rate — the recent-form rate that fed Poisson. Main-line rows would use a different namespace once a main-line forecast model ships."

Previous value: -"Measured rate in counting-stat units per game after shrinkage."New value: +"Measured rate/rating behind this driver, in native units."

Previous value: -"Posted Over/Under main line the model was scored against."New value: +"Posted line the model was scored against. Prop and totals main-line rows: Over/Under number. Spreads main-line rows: the spread number. Null on h2h/moneyline rows."

Previous value: -"number"New value: +[
+  "number",
+  "null"
+]

Previous value: -"Owls prop category (1:1 counting stats). MLB: strikeouts_pitcher, hits, rbis, runs, strikeouts_batter, hits_allowed, earned_runs, outs_recorded. NFL/NCAAF: passing/rushing/receiving_yards, receptions. NBA/NCAAB: points, rebounds, assists, steals, blocks, threes_made. NHL: goals, hockey_assists, shots_on_goal."New value: +"Player-prop rows: Owls prop category (1:1 counting stats). MLB: strikeouts_pitcher, hits, rbis, runs, strikeouts_batter, hits_allowed, earned_runs, outs_recorded. NFL/NCAAF: passing/rushing/receiving_yards, receptions. NBA/NCAAB: points, rebounds, assists, steals, blocks, threes_made. NHL: goals, hockey_assists, shots_on_goal. Main-line rows (bet_type set): the main-line family — h2h, spreads, or totals."

Previous value: -"Model probability (0–1) that side hits line, from the player's shrunken counting-stat rate (Poisson vs the posted Over/Under). Optional research may nudge this by at most ±0.03; the side stays."New value: +"Model probability (0–1) that the picked outcome (side on prop rows, bet_type on main-line rows) hits. Prop rows: built from the player's shrunken counting-stat rate (Poisson vs the posted Over/Under). Optional research may nudge a prop's p_hit by at most ±0.03; the chosen side stays the same."

Previous value: -"Player display name from the priced line."New value: +"Player display name from the priced line. Null on main-line rows with no player attribution (bet_type is a match-level token like OVER/UNDER/ML_DRAW)."

Previous value: -"string"New value: +[
+  "string",
+  "null"
+]

Previous value: -"Lumify player ID. Bound by exact / Jr-stripped name match."New value: +"Lumify player ID. Bound by exact / Jr-stripped name match. Null on main-line rows with no player attribution."

Previous value: -"integer"New value: +[
+  "integer",
+  "null"
+]

Added value: +{
+  "description": "'home'/'away' for team sports or 'p1'/'p2' for head-to-head sports, set only alongside a non-null bet_type. Null for player-prop rows and for match-level main-line tokens (OVER, UNDER, ML_DRAW).",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Previous value: -"Optional Deep Research overlay: {stance, conviction, note}. stance is validate / neutral / invalidate. Validate nudges p_hit up by at most 0.03 and lifts conviction; invalidate does the reverse. The chosen side stays. Null means read p_hit from the rate model alone."New value: +"Optional Deep Research overlay on high-probability wagers (p_hit ≥ 0.80): {stance, conviction, note}. stance is validate / neutral / invalidate. note is the wager-level why. Validate / invalidate rows always carry a note. Validate nudges p_hit up by at most 0.03 and lifts conviction; invalidate does the reverse. The chosen side stays. Null means read p_hit from the rate model alone."

Previous value: -"'over' or 'under' — the more likely side given the player's shrunken rate versus line."New value: +"Prop rows: 'over' or 'under' — the more likely side given the player's shrunken rate versus line. Null on main-line rows — read the picked outcome from bet_type instead."

Previous value: -"string"New value: +[
+  "string",
+  "null"
+]

Previous value: -"Natural-language matchup preview — form, context, what to watch. On predictive MLB may be a Search-backed context overlay. Null when no narrative has been generated."New value: +"Natural-language matchup preview — form, context, what to watch. Search-backed context overlay on every event sport (MLB, NFL, NCAAF, NBA, NCAAB, NHL, soccer, tennis); sport-shaped queries. Null when no narrative has been generated."

Previous value: -"Event-level factual matchup chips (starting pitchers, injuries, form). Event-level context overlay. Predictive MLB may populate on first get_intelligence; later calls reuse the stored row."New value: +"Event-level factual matchup chips (injuries, recent form, lineup/availability news — sport-shaped). Search-backed context overlay; may populate on first get_intelligence, later calls reuse the stored row."

Previous value: -"Highest price-gap pick across edges_by_book: {book, price, edge, quote_age_seconds}. Price gap ≠ EV; Tier C informational line-shopping."New value: +"Highest price-gap book across edges_by_book: {book, price, edge, quote_age_seconds}. Tier C informational line-shopping."

Previous value: -"Price gap versus fair.probability per soft book (fair.probability × decimal_odds − 1). Line-shopping only, NOT EV / NOT a recommend."New value: +"Price gap versus fair.probability per soft book (fair.probability × decimal_odds − 1). Line-shopping metric: rank soft books by gap."

Previous value: -"Sharp-consensus fair: {probability, books, n_books, is_consensus}. MLB/NFL: Pinnacle+Circa; MLS/tennis: Pinnacle. Price-gap reference for edges_by_book/best — not EV."New value: +"Sharp-consensus fair: {probability, books, n_books, is_consensus}. MLB/NFL: Pinnacle+Circa; MLS/tennis: Pinnacle. Price-gap reference for edges_by_book/best."

Previous value: -"Natural-language matchup preview — form, context, what to watch. On predictive MLB may be a Search-backed context overlay (not a recommendation). Null when no narrative has been generated."New value: +"Natural-language matchup preview — form, context, what to watch. On predictive MLB may be a Search-backed context overlay. Null when no narrative has been generated."

Previous value: -"Event-level factual matchup chips (starting pitchers, injuries, form). Context overlay only — not a pick. Predictive MLB may populate on first get_intelligence; later calls reuse the stored row."New value: +"Event-level factual matchup chips (starting pitchers, injuries, form). Event-level context overlay. Predictive MLB may populate on first get_intelligence; later calls reuse the stored row."

Previous value: -"False when no intelligence has been computed for this event yet — every other field is then null/empty and the request is not charged."New value: +"False when no predictive bets[] have been computed for this event yet. forecasts[] can still populate on MLB/NFL/NCAAF/NBA/NCAAB/NHL. Free only when available is false and forecasts is empty."

Added value: +{
+  "description": "Forecasted Over/Under player-prop wagers for this event (same object as list_forecasts). selected marks the top-conviction rows on the daily slate; the rest of the event catalog is included with selected:false. Read p_hit as P(side hits the posted line) from the player's shrunken rate. How + field catalog: https://lumify.ai/docs/forecasts",
+  "items": {
+    "properties": {
+      "away_team": {
+        "description": "Away team display name. Slate only.",
+        "type": [
+          "string",
+          "null"
+        ]
+      },
+      "books": {
+        "description": "American price for the chosen side, keyed by bookmaker slug (e.g. {draftkings: -453}). Posted market price for the side the model picked.",
+        "type": "object"
+      },
+      "commence_time": {
+        "description": "ISO-8601 UTC scheduled start. Slate only.",
+        "type": [
+          "string",
+          "null"
+        ]
+      },
+      "conviction": {
+        "description": "Board rank key on 0–1: |p_hit−0.5|×2×sufficiency×research. A high p_hit on a thin sample ranks below a moderate p_hit on a long one. Highest conviction first on the daily slate.",
+        "type": "number"
+      },
+      "drivers": {
+        "description": "Signed contributions that moved p_hit. Read input as the player's measured rate and effect as how far that rate lifts the chosen side versus a league-mean player.",
+        "items": {
+          "properties": {
+            "direction": {
+              "description": "up / down / neutral from the sign of effect.",
+              "type": "string"
+            },
+            "effect": {
+              "description": "Signed shift in p_hit versus a league-mean player on the same line. 0.0 on yardage markets.",
+              "type": "number"
+            },
+            "id": {
+              "description": "e.g. player.l10_rate — the recent-form rate that fed Poisson.",
+              "type": "string"
+            },
+            "input": {
+              "description": "Measured rate in counting-stat units per game after shrinkage.",
+              "type": [
+                "number",
+                "null"
+              ]
+            }
+          },
+          "type": "object"
+        },
+        "type": "array"
+      },
+      "event_id": {
+        "description": "Lumify event ID. Present on list_forecasts; omitted on get_intelligence.forecasts (the event is the argument).",
+        "type": [
+          "integer",
+          "null"
+        ]
+      },
+      "home_team": {
+        "description": "Home team display name. Slate only.",
+        "type": [
+          "string",
+          "null"
+        ]
+      },
+      "interval": {
+        "description": "[lo, hi] stated band around p_hit. Wider when sufficiency is low. A stated evidence band, not a statistical confidence interval.",
+        "items": {
+          "type": "number"
+        },
+        "type": "array"
+      },
+      "league": {
+        "description": "League slug, if any. Slate only.",
+        "type": [
+          "string",
+          "null"
+        ]
+      },
+      "line": {
+        "description": "Posted Over/Under main line the model was scored against.",
+        "type": "number"
+      },
+      "market": {
+        "description": "Owls prop category (1:1 counting stats). MLB: strikeouts_pitcher, hits, rbis, runs, strikeouts_batter, hits_allowed, earned_runs, outs_recorded. NFL/NCAAF: passing/rushing/receiving_yards, receptions. NBA/NCAAB: points, rebounds, assists, steals, blocks, threes_made. NHL: goals, hockey_assists, shots_on_goal.",
+        "type": "string"
+      },
+      "model_version": {
+        "description": "Scorer version, e.g. forecast-v0.",
+        "type": "string"
+      },
+      "p_hit": {
+        "description": "Model probability (0–1) that side hits line, from the player's shrunken counting-stat rate (Poisson vs the posted Over/Under). Optional research may nudge this by at most ±0.03; the side stays.",
+        "type": "number"
+      },
+      "player": {
+        "description": "Player display name from the priced line.",
+        "type": "string"
+      },
+      "player_id": {
+        "description": "Lumify player ID. Bound by exact / Jr-stripped name match.",
+        "type": "integer"
+      },
+      "reliability": {
+        "description": "Badge on this sport×market cell: emerging, moderate, high, or informational. v0 ships emerging for every wager.",
+        "type": "string"
+      },
+      "research": {
+        "description": "Optional Deep Research overlay: {stance, conviction, note}. stance is validate / neutral / invalidate. Validate nudges p_hit up by at most 0.03 and lifts conviction; invalidate does the reverse. The chosen side stays. Null means read p_hit from the rate model alone.",
+        "type": [
+          "object",
+          "null"
+        ]
+      },
+      "selected": {
+        "description": "True when this wager is on the daily board (top conviction). On get_intelligence the array includes selected and the rest of the event catalog.",
+        "type": "boolean"
+      },
+      "side": {
+        "description": "'over' or 'under' — the more likely side given the player's shrunken rate versus line.",
+        "type": "string"
+      },
+      "sport": {
+        "description": "Sport slug. Slate only.",
+        "type": [
+          "string",
+          "null"
+        ]
+      },
+      "sufficiency": {
+        "description": "Evidence weight on 0–1: n/(n+12), where n is the recent-form game count (typically L10). A player with 10 games has sufficiency ≈ 0.45. The wager still ships; this field tells you how much of p_hit is earned from this player's own boxes.",
+        "type": "number"
+      }
+    },
+    "type": "object"
+  },
+  "type": "array"
+}

Previous value: -"Bookmaker slug for market prices on recommended bets. Defaults to pinnacle. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline. No effect for probability-model sports (MLS soccer and MLB), which report the book their assessment was priced against."New value: +"Ignored. Intelligence always reports the book the assessment was priced against. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline."

Previous value: -"Attribution keys backing rationale, same order. Points-model sports only."New value: +"Not published."

Previous value: -"Points-engine score, 0-1, not a calibrated probability. Points-model sports only."New value: +"Not published. Use probability."

Previous value: -"Signal-coverage fraction, 0-1. Points-model sports only."New value: +"Not published."

Previous value: -"Per-bet narrative sentence. Points-model sports only."New value: +"Not published."

Previous value: -"Per-bet signal-strength bullets. Points-model sports only. Distinct from top-level rationale (event-level context chips)."New value: +"Not published. Top-level rationale is the event-level context overlay."

Previous value: -"Per-signal point breakdown. Points-model sports only."New value: +"Not published."

Previous value: -"Qualitative-validator verdict: {stance, confidence, delta, validated_at}. Points-model sports only."New value: +"Not published."

Previous value: -"Legacy points-engine probable-starter context. Not returned on the MLB predictive customer surface — use get_stats for pitcher/lineup Data."New value: +"Not returned. Use get_stats for pitcher/lineup Data."

Previous value: -"Event-level factual matchup chips (starting pitchers, injuries, form). Context overlay only — not a pick. Distinct from bets[].rationale on points-model sports. Predictive MLB may populate on first get_intelligence; later calls reuse the stored row."New value: +"Event-level factual matchup chips (starting pitchers, injuries, form). Context overlay only — not a pick. Predictive MLB may populate on first get_intelligence; later calls reuse the stored row."

Added value: +{
+  "description": "Beta. Customer-facing EV estimate: {beta: true, book, price, ev_pct, kelly_fraction, quote_age_seconds, n_books}. Re-packages the same fair price gap for direct display. Moneyline only. n=1 Pinnacle is a valid fair (soccer/tennis). Null when the gap is ≤0, above 25% (stale-line), or only suppressed books remain (MLB ML × DK/FD proven null; BetMGM/Bet365/BetOnline unscreened).",
+  "type": [
+    "object",
+    "null"
+  ]
+}

Previous value: -"Sharp-consensus fair: {probability, books, n_books, is_consensus}. MLB: Pinnacle+Circa; MLS/tennis: Pinnacle. Price-gap reference for edges_by_book/best — not EV."New value: +"Sharp-consensus fair: {probability, books, n_books, is_consensus}. MLB/NFL: Pinnacle+Circa; MLS/tennis: Pinnacle. Price-gap reference for edges_by_book/best — not EV."

Previous value: -"Published probability for this outcome, 0-1. Outcomes of one market sum to 1. Customer surface for MLB, soccer, tennis."New value: +"Published probability for this outcome, 0-1. Outcomes of one market sum to 1. Customer surface for MLB, soccer, tennis, NFL."

Removed value: -{
-  "description": "Short natural-language read on the event as a whole. Null when no narrative has been generated.",
-  "type": [
-    "string",
-    "null"
-  ]
-}

Previous value: -"Human-readable rationale bullets. Points-model sports only."New value: +"Per-bet signal-strength bullets. Points-model sports only. Distinct from top-level rationale (event-level context chips)."

Previous value: -"Longer natural-language matchup preview — form, context, what to watch. Null when no narrative has been generated."New value: +"Natural-language matchup preview — form, context, what to watch. On predictive MLB may be a Search-backed context overlay (not a recommendation). Null when no narrative has been generated."

Added value: +{
+  "description": "Event-level factual matchup chips (starting pitchers, injuries, form). Context overlay only — not a pick. Distinct from bets[].rationale on points-model sports. Predictive MLB may populate on first get_intelligence; later calls reuse the stored row.",
+  "items": {
+    "type": "string"
+  },
+  "type": [
+    "array",
+    "null"
+  ]
+}

Previous value: -"Qualitative-overlay agreement detail. Populated only when phase is 'full'; null for every 'quant' assessment. Probability-model sports only."New value: +"Internal — omitted from customer payloads (MLB/soccer/tennis)."

Previous value: -"Highest price-gap pick across edges_by_book: {book, price, edge, quote_age_seconds}. Same caveats as edges_by_book — price gap, not EV. On MLB treat as Tier C informational line-shopping (DK/FD ML and RL/totals are hard-null for Edge). Use best.book/best.price as the line-shopping target; do not read best.edge as EV. Stale quotes (>30 min) excluded. Populated for MLS and MLB; null otherwise. Probability-model sports only."New value: +"Highest price-gap pick across edges_by_book: {book, price, edge, quote_age_seconds}. Price gap ≠ EV; Tier C informational line-shopping."

Previous value: -"Weight on p_model when blending with p_market, 0-1. 0 means the published probability is purely the de-vigged market. Enabled per league and per bet token. Probability-model sports only."New value: +"Internal — omitted from customer payloads (MLB/soccer/tennis)."

Previous value: -"Signal-coverage fraction, 0-1. Points-model sports only — the probability model uses sufficiency instead."New value: +"Signal-coverage fraction, 0-1. Points-model sports only."

Previous value: -"Named, signed contributions to probability: {id, input, effect, direction, evidence} where effect is the probability shift, direction is up/down/neutral, and evidence is null (quant drivers) or a list of {fact_id, as_of} citations into the Fact Ledger (evidence-derived drivers) — never an embedded quote/URL. Normally empty when no model contributed (blend_w 0); two exceptions carry effect 0 and never move probability: Stage 6 Match Context drivers (soccer.match_context.*, evidence-backed) and soccer.cross_book_edge (market-anchored price gap vs. the sharp consensus; input is the same gap already published in edges_by_book — a line-shopping signal, not an expected-value claim). Probability-model sports only."New value: +"Internal — omitted from customer payloads (MLB/soccer/tennis)."

Previous value: -"Expected profit per 1 unit staked at market.price (probability x decimal_odds - 1). Null whenever blend_w is 0, since a market-derived probability has no honest edge against its own price. Probability-model sports only."New value: +"Expected profit per 1 unit staked at market.price. Null while no model-backed edge is published (Stage 1). Present on customer surface for recommend gating."

Previous value: -"Price gap versus the sharp-reference fair.probability, keyed by bookmaker (fair.probability × decimal_odds − 1). A price-gap / line-shopping metric, NOT expected value and NOT a claim-ladder Edge badge. Rank soft books by gap for line-shopping; do not treat a positive gap as EV or badge '+EV'. Populated for MLS (FanDuel/Hard Rock) and MLB (retail soft books — Price surface; Edge allowlist remains DK/FD). Null otherwise. Probability-model sports only."New value: +"Price gap versus fair.probability per soft book (fair.probability × decimal_odds − 1). Line-shopping only, NOT EV / NOT a recommend."

Previous value: -"Sharp-consensus fair price: {probability, books, n_books, is_consensus}. Distinct from p_market (the priced book's own de-vig) and fair_price (odds implied by probability) — this is the sharp-book reference a cross-book edge is measured against. is_consensus is false at n_books < 2 (true for every soccer row today — Pinnacle is the only soccer sharp book). Populated for MLS as of Stage 5 v1 (2026-07-29); null for every other league until its own multi-book fetch is wired. Probability-model sports only."New value: +"Sharp-consensus fair: {probability, books, n_books, is_consensus}. MLB: Pinnacle+Circa; MLS/tennis: Pinnacle. Price-gap reference for edges_by_book/best — not EV."

Previous value: -"American-odds fair price implied by probability — the vig-free line. Compare to market.price for the book's margin on this side. Probability-model sports only."New value: +"American-odds fair price implied by probability — the vig-free line. Compare to market.price for the book's margin on this side."

Previous value: -"[lo, hi] band around probability driven by sufficiency — how much evidence backs the number, not a statistical confidence interval. When blend_w is 0 it reflects how mature and well-traded the quoted line is. Probability-model sports only."New value: +"[lo, hi] band around probability — how much evidence backs the number, not a statistical confidence interval."

Previous value: -"Parameter-set identifier so a number can be traced to the version that produced it. 'market_anchor' means no fitted model contributed. Per bet. Probability-model sports only."New value: +"Internal — omitted from customer payloads (MLB/soccer/tennis)."

Previous value: -"De-vigged market probability, pre-blend. Vig is removed across the whole market so outcomes sum to 1 — not recoverable from one price by implied-probability math. Probability-model sports only."New value: +"Internal — omitted from customer payloads (MLB/soccer/tennis). Use probability/fair_price."

Previous value: -"The fitted model's own probability, pre-blend. Null for leagues with no model cleared for publication — currently every soccer league. Probability-model sports only."New value: +"Internal — omitted from customer payloads (MLB/soccer/tennis)."

Previous value: -"'quant' when purely deterministic; 'full' once a qualitative overlay (lineups, injuries, research) is attached, which is when alignment is populated. Probability-model sports only."New value: +"Internal — omitted from customer payloads (MLB/soccer/tennis)."

Previous value: -"Published probability for this outcome, 0-1. Equals p_market when blend_w is 0; otherwise the blend of p_model and p_market. Outcomes of one market sum to 1. Probability-model sports only."New value: +"Published probability for this outcome, 0-1. Outcomes of one market sum to 1. Customer surface for MLB, soccer, tennis."

Previous value: -"Per-signal point breakdown. Points-model sports only — the probability model uses drivers instead."New value: +"Per-signal point breakdown. Points-model sports only."

Previous value: -"How much evidence backs this assessment, 0-1; sets interval width and caps tier. While blend_w is 0 it measures the quoted line's maturity (movement count and hours since first quote); once a model carries weight it reflects model sample depth. Probability-model sports only."New value: +"Internal — omitted from customer payloads (MLB/soccer/tennis)."

Previous value: -"Confidence tier: 'very_high', 'strong', 'moderate', or 'avoid'. Present for both shapes. For probability-model sports it is null whenever edge is null — a tier ranks a bet against its price."New value: +"Confidence tier: 'very_high', 'strong', 'moderate', or 'avoid'. Null whenever edge is null. Present on MLB customer surface and all other shapes."

Previous value: -"True when at least one bet meets the recommendation threshold; null when intelligence has not been computed. False for any probability-model event whose bets are all market-anchored (blend_w 0), since a recommendation requires an edge."New value: +"True when at least one bet meets the recommendation threshold; null when intelligence has not been computed. False while Edge is not published (current MLB/soccer/tennis Stage 1) — a recommendation requires a non-null tier derived from edge."

Previous value: -"MLB-only probable-starter context (home_starter/away_starter with name/hand/era/confirmed). Absent for every other sport."New value: +"Legacy points-engine probable-starter context. Not returned on the MLB predictive customer surface — use get_stats for pitcher/lineup Data."

Previous value: -"Price gap versus the sharp-reference fair.probability, keyed by bookmaker (fair.probability × decimal_odds − 1). A price-gap / line-shopping metric, NOT expected value and NOT a claim-ladder Edge badge. Rank soft books by gap for line-shopping; do not treat a positive gap as EV or badge '+EV'. Populated for MLS (FanDuel/Hard Rock) and MLB (Owls retail soft books — Price surface; Edge allowlist remains DK/FD). Null otherwise. Probability-model sports only."New value: +"Price gap versus the sharp-reference fair.probability, keyed by bookmaker (fair.probability × decimal_odds − 1). A price-gap / line-shopping metric, NOT expected value and NOT a claim-ladder Edge badge. Rank soft books by gap for line-shopping; do not treat a positive gap as EV or badge '+EV'. Populated for MLS (FanDuel/Hard Rock) and MLB (retail soft books — Price surface; Edge allowlist remains DK/FD). Null otherwise. Probability-model sports only."

Previous value: -"Bookmaker slug for market prices on recommended bets. Defaults to pinnacle. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline. No effect for probability-model sports (currently soccer/MLS), which report the book their assessment was priced against."New value: +"Bookmaker slug for market prices on recommended bets. Defaults to pinnacle. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline. No effect for probability-model sports (MLS soccer and MLB), which report the book their assessment was priced against."

Previous value: -"Highest price-gap pick across edges_by_book: {book, price, edge, quote_age_seconds}. edge is the raw max (shrinkage=0.0 — won't-fit at N=2). Same caveats as edges_by_book: a price gap versus a de-vigged sharp reference, NOT an expected-value claim; at N=2 winner's-curse ≈171% of mean positive gap and closing-time N=2 backfill realized ≈−2.5% on positive-gap picks. Use best.book/best.price as the line-shopping target for this outcome; do not read best.edge as EV or as a bet recommendation. quote_age_seconds is best.book's posted-price age; stale quotes (>30 min) are excluded before selection. Populated for MLS as of Stage 5 v1; null otherwise. Probability-model sports only."New value: +"Highest price-gap pick across edges_by_book: {book, price, edge, quote_age_seconds}. Same caveats as edges_by_book — price gap, not EV. On MLB treat as Tier C informational line-shopping (DK/FD ML and RL/totals are hard-null for Edge). Use best.book/best.price as the line-shopping target; do not read best.edge as EV. Stale quotes (>30 min) excluded. Populated for MLS and MLB; null otherwise. Probability-model sports only."

Previous value: -"Price gap versus the sharp-reference fair.probability, keyed by bookmaker (fair.probability × decimal_odds − 1 at each book's posted price). A price-gap / line-shopping metric, NOT an expected-value recommendation: at N=2 soft-book coverage the measured winner's-curse component of a max-of-N pick is ~171% of the mean positive gap, and a closing-time N=2 backfill simulation realized ≈−2.5% on the positive-gap population (SOCCER_PROGRAM_PLAN.md §5.0 E/tenth pass). Rank soft books by these gaps for line-shopping (largest gap = furthest above the sharp reference); do not treat a positive gap as EV or as a bet recommendation, and do not badge picks as '+EV'. Populated for MLS as of Stage 5 v1 (FanDuel/Hard Rock); null otherwise. Probability-model sports only."New value: +"Price gap versus the sharp-reference fair.probability, keyed by bookmaker (fair.probability × decimal_odds − 1). A price-gap / line-shopping metric, NOT expected value and NOT a claim-ladder Edge badge. Rank soft books by gap for line-shopping; do not treat a positive gap as EV or badge '+EV'. Populated for MLS (FanDuel/Hard Rock) and MLB (Owls retail soft books — Price surface; Edge allowlist remains DK/FD). Null otherwise. Probability-model sports only."

Previous value: -"Named, signed contributions to probability: {id, input, effect, direction, evidence} where effect is the probability shift, direction is up/down/neutral, and evidence is null (quant drivers) or a list of {fact_id, as_of} citations into the Fact Ledger (evidence-derived drivers) — never an embedded quote/URL. Normally empty when no model contributed (blend_w 0); exception: Stage 6 Match Context drivers (soccer.match_context.*) cite the Fact Ledger with effect 0 and never move probability. Probability-model sports only."New value: +"Named, signed contributions to probability: {id, input, effect, direction, evidence} where effect is the probability shift, direction is up/down/neutral, and evidence is null (quant drivers) or a list of {fact_id, as_of} citations into the Fact Ledger (evidence-derived drivers) — never an embedded quote/URL. Normally empty when no model contributed (blend_w 0); two exceptions carry effect 0 and never move probability: Stage 6 Match Context drivers (soccer.match_context.*, evidence-backed) and soccer.cross_book_edge (market-anchored price gap vs. the sharp consensus; input is the same gap already published in edges_by_book — a line-shopping signal, not an expected-value claim). Probability-model sports only."

Previous value: -"Named, signed contributions to probability: {id, input, effect, direction, evidence} where effect is the probability shift, direction is up/down/neutral, and evidence is null (quant drivers) or a list of {fact_id, as_of} citations into the Fact Ledger (evidence-derived drivers) — never an embedded quote/URL. Empty whenever no model contributed (blend_w 0). Probability-model sports only."New value: +"Named, signed contributions to probability: {id, input, effect, direction, evidence} where effect is the probability shift, direction is up/down/neutral, and evidence is null (quant drivers) or a list of {fact_id, as_of} citations into the Fact Ledger (evidence-derived drivers) — never an embedded quote/URL. Normally empty when no model contributed (blend_w 0); exception: Stage 6 Match Context drivers (soccer.match_context.*) cite the Fact Ledger with effect 0 and never move probability. Probability-model sports only."

Previous value: -"Highest price-gap pick across edges_by_book: {book, price, edge, quote_age_seconds}. edge is the raw max (shrinkage=0.0 — won't-fit at N=2). Same caveats as edges_by_book: a price gap versus a de-vigged sharp reference, NOT an expected-value claim; at N=2 winner's-curse ≈171% of mean positive gap and closing-time N=2 backfill realized ≈−2.5% on positive-gap picks. Do not sort, rank, badge, or recommend on it. quote_age_seconds is best.book's posted-price age; stale quotes (>30 min) are excluded before selection. Populated for MLS as of Stage 5 v1; null otherwise. Probability-model sports only."New value: +"Highest price-gap pick across edges_by_book: {book, price, edge, quote_age_seconds}. edge is the raw max (shrinkage=0.0 — won't-fit at N=2). Same caveats as edges_by_book: a price gap versus a de-vigged sharp reference, NOT an expected-value claim; at N=2 winner's-curse ≈171% of mean positive gap and closing-time N=2 backfill realized ≈−2.5% on positive-gap picks. Use best.book/best.price as the line-shopping target for this outcome; do not read best.edge as EV or as a bet recommendation. quote_age_seconds is best.book's posted-price age; stale quotes (>30 min) are excluded before selection. Populated for MLS as of Stage 5 v1; null otherwise. Probability-model sports only."

Previous value: -"Price gap versus the sharp-reference fair.probability, keyed by bookmaker (fair.probability × decimal_odds − 1 at each book's posted price). A price-gap / line-shopping metric, NOT an expected-value recommendation: at N=2 soft-book coverage the measured winner's-curse component of a max-of-N pick is ~171% of the mean positive gap, and a closing-time N=2 backfill simulation realized ≈−2.5% on the positive-gap population (SOCCER_PROGRAM_PLAN.md §5.0 E/tenth pass). Do not sort, rank, badge, or recommend on these values. Populated for MLS as of Stage 5 v1 (FanDuel/Hard Rock); null otherwise. Probability-model sports only."New value: +"Price gap versus the sharp-reference fair.probability, keyed by bookmaker (fair.probability × decimal_odds − 1 at each book's posted price). A price-gap / line-shopping metric, NOT an expected-value recommendation: at N=2 soft-book coverage the measured winner's-curse component of a max-of-N pick is ~171% of the mean positive gap, and a closing-time N=2 backfill simulation realized ≈−2.5% on the positive-gap population (SOCCER_PROGRAM_PLAN.md §5.0 E/tenth pass). Rank soft books by these gaps for line-shopping (largest gap = furthest above the sharp reference); do not treat a positive gap as EV or as a bet recommendation, and do not badge picks as '+EV'. Populated for MLS as of Stage 5 v1 (FanDuel/Hard Rock); null otherwise. Probability-model sports only."

Previous value: -"Best pick across edges_by_book: {book, price, edge, quote_age_seconds}. edge is the raw max — shrinkage=0.0 (no winner's-curse correction) until fitted against multi-book historical odds, so treat as v1. quote_age_seconds is best.book's posted-price age at publish time; a stale soft-book price (>30 min) is excluded before selection. Populated for MLS as of Stage 5 v1 (2026-07-29); null otherwise. Probability-model sports only."New value: +"Highest price-gap pick across edges_by_book: {book, price, edge, quote_age_seconds}. edge is the raw max (shrinkage=0.0 — won't-fit at N=2). Same caveats as edges_by_book: a price gap versus a de-vigged sharp reference, NOT an expected-value claim; at N=2 winner's-curse ≈171% of mean positive gap and closing-time N=2 backfill realized ≈−2.5% on positive-gap picks. Do not sort, rank, badge, or recommend on it. quote_age_seconds is best.book's posted-price age; stale quotes (>30 min) are excluded before selection. Populated for MLS as of Stage 5 v1; null otherwise. Probability-model sports only."

Previous value: -"Expected value per 1 unit staked against fair.probability, keyed by bookmaker, for every independently-eligible soft book — the cross-book +EV surface. Populated for MLS as of Stage 5 v1 (FanDuel/Hard Rock, both cleared by the book-independence screen); null for every other league, and null for any market with no eligible soft book quoting it. Probability-model sports only."New value: +"Price gap versus the sharp-reference fair.probability, keyed by bookmaker (fair.probability × decimal_odds − 1 at each book's posted price). A price-gap / line-shopping metric, NOT an expected-value recommendation: at N=2 soft-book coverage the measured winner's-curse component of a max-of-N pick is ~171% of the mean positive gap, and a closing-time N=2 backfill simulation realized ≈−2.5% on the positive-gap population (SOCCER_PROGRAM_PLAN.md §5.0 E/tenth pass). Do not sort, rank, badge, or recommend on these values. Populated for MLS as of Stage 5 v1 (FanDuel/Hard Rock); null otherwise. Probability-model sports only."

Previous value: -"Winner's-curse-corrected best pick across edges_by_book: {book, price, edge}. edge is shrunk toward the field median, never the raw max. Null until shrinkage is fitted against multi-book historical odds. Probability-model sports only."New value: +"Best pick across edges_by_book: {book, price, edge, quote_age_seconds}. edge is the raw max — shrinkage=0.0 (no winner's-curse correction) until fitted against multi-book historical odds, so treat as v1. quote_age_seconds is best.book's posted-price age at publish time; a stale soft-book price (>30 min) is excluded before selection. Populated for MLS as of Stage 5 v1 (2026-07-29); null otherwise. Probability-model sports only."

Previous value: -"Expected value per 1 unit staked against fair.probability, keyed by bookmaker, for every independently-eligible soft book — the cross-book +EV surface. Null until the book-independence gate and multi-book fetch are wired. Probability-model sports only."New value: +"Expected value per 1 unit staked against fair.probability, keyed by bookmaker, for every independently-eligible soft book — the cross-book +EV surface. Populated for MLS as of Stage 5 v1 (FanDuel/Hard Rock, both cleared by the book-independence screen); null for every other league, and null for any market with no eligible soft book quoting it. Probability-model sports only."

Previous value: -"Sharp-consensus fair price: {probability, books, n_books, is_consensus}. Distinct from p_market (the priced book's own de-vig) and fair_price (odds implied by probability) — this is the sharp-book reference a cross-book edge is measured against. is_consensus is false at n_books < 2. Null until Stage 5's multi-book fetch is wired at publish time. Probability-model sports only."New value: +"Sharp-consensus fair price: {probability, books, n_books, is_consensus}. Distinct from p_market (the priced book's own de-vig) and fair_price (odds implied by probability) — this is the sharp-book reference a cross-book edge is measured against. is_consensus is false at n_books < 2 (true for every soccer row today — Pinnacle is the only soccer sharp book). Populated for MLS as of Stage 5 v1 (2026-07-29); null for every other league until its own multi-book fetch is wired. Probability-model sports only."

Added value: +{
+  "description": "Winner's-curse-corrected best pick across edges_by_book: {book, price, edge}. edge is shrunk toward the field median, never the raw max. Null until shrinkage is fitted against multi-book historical odds. Probability-model sports only.",
+  "type": [
+    "object",
+    "null"
+  ]
+}

Previous value: -"Named, signed contributions to probability: {id, input, effect, direction} where effect is the probability shift and direction is up/down/neutral. Empty whenever no model contributed (blend_w 0). Probability-model sports only."New value: +"Named, signed contributions to probability: {id, input, effect, direction, evidence} where effect is the probability shift, direction is up/down/neutral, and evidence is null (quant drivers) or a list of {fact_id, as_of} citations into the Fact Ledger (evidence-derived drivers) — never an embedded quote/URL. Empty whenever no model contributed (blend_w 0). Probability-model sports only."

Added value: +{
+  "description": "Expected value per 1 unit staked against fair.probability, keyed by bookmaker, for every independently-eligible soft book — the cross-book +EV surface. Null until the book-independence gate and multi-book fetch are wired. Probability-model sports only.",
+  "type": [
+    "object",
+    "null"
+  ]
+}

Added value: +{
+  "description": "Sharp-consensus fair price: {probability, books, n_books, is_consensus}. Distinct from p_market (the priced book's own de-vig) and fair_price (odds implied by probability) — this is the sharp-book reference a cross-book edge is measured against. is_consensus is false at n_books < 2. Null until Stage 5's multi-book fetch is wired at publish time. Probability-model sports only.",
+  "type": [
+    "object",
+    "null"
+  ]
+}

Previous value: -"Bookmaker slug for market prices on recommended bets. Defaults to pinnacle. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline. No effect for predictive-framework sports (currently soccer/MLS)."New value: +"Bookmaker slug for market prices on recommended bets. Defaults to pinnacle. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline. No effect for probability-model sports (currently soccer/MLS), which report the book their assessment was priced against."

Added value: +"Short natural-language read on the event as a whole. Null when no narrative has been generated."

Previous value: -"Qualitative-overlay agreement detail; populated only when phase is 'full' (a later stage). Predictive-framework sports only."New value: +"Qualitative-overlay agreement detail. Populated only when phase is 'full'; null for every 'quant' assessment. Probability-model sports only."

Previous value: -"Attribution keys backing rationale, same order. Legacy sports only."New value: +"Attribution keys backing rationale, same order. Points-model sports only."

Previous value: -"Weight on p_model in the blend, 0-1. Always 0 in Stage 1. Predictive-framework sports only."New value: +"Weight on p_model when blending with p_market, 0-1. 0 means the published probability is purely the de-vigged market. Enabled per league and per bet token. Probability-model sports only."

Previous value: -"When this row was last computed/published, ISO-8601."New value: +"ISO-8601 UTC time this bet's numbers last materially changed — not when last checked. Rows are only rewritten when price, line, or probability moves beyond a tolerance, so an older value means 'unchanged since', not 'stale'. Differs between bets on one event because markets move independently."

Previous value: -"Legacy points-engine score, 0-1, not a calibrated probability. Legacy sports only."New value: +"Points-engine score, 0-1, not a calibrated probability. Points-model sports only."

Previous value: -"Legacy signal-coverage fraction, 0-1. Legacy sports only — superseded by sufficiency below."New value: +"Signal-coverage fraction, 0-1. Points-model sports only — the probability model uses sufficiency instead."

Previous value: -"Named, signed contributions to probability: {id, input, effect, direction}. Replaces signals. Always empty in Stage 1. Predictive-framework sports only."New value: +"Named, signed contributions to probability: {id, input, effect, direction} where effect is the probability shift and direction is up/down/neutral. Empty whenever no model contributed (blend_w 0). Probability-model sports only."

Previous value: -"Expected value per unit staked at market.price. Null whenever the model carried no blend weight — always null in Stage 1. Predictive-framework sports only."New value: +"Expected profit per 1 unit staked at market.price (probability x decimal_odds - 1). Null whenever blend_w is 0, since a market-derived probability has no honest edge against its own price. Probability-model sports only."

Previous value: -"American-odds fair price implied by probability. Predictive-framework sports only."New value: +"American-odds fair price implied by probability — the vig-free line. Compare to market.price for the book's margin on this side. Probability-model sports only."

Previous value: -"[lo, hi] uncertainty band around probability. Predictive-framework sports only."New value: +"[lo, hi] band around probability driven by sufficiency — how much evidence backs the number, not a statistical confidence interval. When blend_w is 0 it reflects how mature and well-traded the quoted line is. Probability-model sports only."

Previous value: -"Market quote: {price: American odds, line: handicap/total line or null for moneyline, book: bookmaker slug when known (always set for predictive-framework sports)}."New value: +"Market quote: {price: American odds, line: handicap/total line or null for moneyline, book: bookmaker slug when known (always set for probability-model sports)}."

Previous value: -"Parameter-set identifier, e.g. 'market_anchor' (Stage 1: no fitted model). Predictive-framework sports only."New value: +"Parameter-set identifier so a number can be traced to the version that produced it. 'market_anchor' means no fitted model contributed. Per bet. Probability-model sports only."

Previous value: -"Legacy per-bet narrative sentence. Legacy sports only."New value: +"Per-bet narrative sentence. Points-model sports only."

Previous value: -"De-vigged market probability, pre-blend. Predictive-framework sports only."New value: +"De-vigged market probability, pre-blend. Vig is removed across the whole market so outcomes sum to 1 — not recoverable from one price by implied-probability math. Probability-model sports only."

Previous value: -"The fitted model's own probability, pre-blend. Null until a model exists (all of Stage 1). Predictive-framework sports only."New value: +"The fitted model's own probability, pre-blend. Null for leagues with no model cleared for publication — currently every soccer league. Probability-model sports only."

Previous value: -"'quant' (deterministic stack) or 'full' (qualitative overlay, a later stage). Predictive-framework sports only."New value: +"'quant' when purely deterministic; 'full' once a qualitative overlay (lineups, injuries, research) is attached, which is when alignment is populated. Probability-model sports only."

Previous value: -"Set when this bet is about an individual player."New value: +"Set when this bet is about an individual player. Null for team bets and match-level tokens."

Previous value: -"Display name of the player or team this bet is about."New value: +"Display name of the player or team this bet is about. Null for match-level tokens. Duplicates players[player_role].name."

Previous value: -"'home'/'away' or 'p1'/'p2'. Null for match-level tokens (OVER/UNDER)."New value: +"'home'/'away' or 'p1'/'p2'. Null for match-level tokens (OVER, UNDER, ML_DRAW)."

Previous value: -"Published probability, 0-1. Pure de-vigged market price in Stage 1. Predictive-framework sports only."New value: +"Published probability for this outcome, 0-1. Equals p_market when blend_w is 0; otherwise the blend of p_model and p_market. Outcomes of one market sum to 1. Probability-model sports only."

Previous value: -"Legacy human-readable rationale bullets. Legacy sports only."New value: +"Human-readable rationale bullets. Points-model sports only."

Previous value: -"Legacy per-signal point breakdown. Legacy sports only — superseded by drivers below."New value: +"Per-signal point breakdown. Points-model sports only — the probability model uses drivers instead."

Previous value: -"Evidence strength, 0-1. In Stage 1: max of per-market movement depth and hours-since-first-quote maturity — not team-form sample size. Predictive-framework sports only."New value: +"How much evidence backs this assessment, 0-1; sets interval width and caps tier. While blend_w is 0 it measures the quoted line's maturity (movement count and hours since first quote); once a model carries weight it reflects model sample depth. Probability-model sports only."

Previous value: -"Set when this bet is about a team."New value: +"Set when this bet is about a team. Null for match-level tokens (OVER, UNDER, ML_DRAW) — a draw is not a bet on either team, so summing exposure by team_id never double-counts it."

Previous value: -"Confidence tier: 'very_high', 'strong', 'moderate', or 'avoid'. Present for both shapes; always null in predictive-framework Stage 1 (no edge yet to rank)."New value: +"Confidence tier: 'very_high', 'strong', 'moderate', or 'avoid'. Present for both shapes. For probability-model sports it is null whenever edge is null — a tier ranks a bet against its price."

Previous value: -"Legacy qualitative-validator verdict: {stance, confidence, delta, validated_at}. Legacy sports only."New value: +"Qualitative-validator verdict: {stance, confidence, delta, validated_at}. Points-model sports only."

Added value: +"Lumify event ID this intelligence describes."

Previous value: -"Legacy: true when a bet is recommended. Predictive Stage 1 (MLS): always false when assessments exist; null when unavailable."New value: +"True when at least one bet meets the recommendation threshold; null when intelligence has not been computed. False for any probability-model event whose bets are all market-anchored (blend_w 0), since a recommendation requires an edge."

Previous value: -"UTC datetime of last intelligence update. Predictive sports: max(assessments.computed_at)."New value: +"ISO-8601 UTC timestamp of the most recent change anywhere in this payload (max of per-bet computed_at). Use per-bet computed_at to reason about a specific bet."

Added value: +"Longer natural-language matchup preview — form, context, what to watch. Null when no narrative has been generated."

Previous value: -"Bookmaker bets[].market prices were sourced from. Predictive sports: assessments.market_book (not a ?bookmaker= overlay)."New value: +"Bookmaker bets[].market prices were sourced from. For probability-model sports this is the book the assessment was priced against, not a bookmaker overlay; per-bet market.book is authoritative if they ever differ."

Previous value: -"Bookmaker slug for market prices on recommended bets. Defaults to pinnacle. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline."New value: +"Bookmaker slug for market prices on recommended bets. Defaults to pinnacle. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline. No effect for predictive-framework sports (currently soccer/MLS)."

Added value: +"False when no intelligence has been computed for this event yet — every other field is then null/empty and the request is not charged."

Added value: +{
+  "description": "Qualitative-overlay agreement detail; populated only when phase is 'full' (a later stage). Predictive-framework sports only.",
+  "type": [
+    "object",
+    "null"
+  ]
+}

Added value: +"Attribution keys backing rationale, same order. Legacy sports only."

Added value: +"Canonical bet token, e.g. ML_HOME/ML_AWAY/ML_DRAW (soccer), ML_P1/ML_P2 (head-to-head sports), SPREAD_HOME/SPREAD_AWAY, OVER, UNDER."

Added value: +{
+  "description": "Weight on p_model in the blend, 0-1. Always 0 in Stage 1. Predictive-framework sports only.",
+  "type": [
+    "number",
+    "null"
+  ]
+}

Added value: +{
+  "description": "When this row was last computed/published, ISO-8601.",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Added value: +"Legacy points-engine score, 0-1, not a calibrated probability. Legacy sports only."

Previous value: -"number"New value: +[
+  "number",
+  "null"
+]

Added value: +"Legacy signal-coverage fraction, 0-1. Legacy sports only — superseded by sufficiency below."

Previous value: -"number"New value: +[
+  "number",
+  "null"
+]

Added value: +{
+  "description": "Named, signed contributions to probability: {id, input, effect, direction}. Replaces signals. Always empty in Stage 1. Predictive-framework sports only.",
+  "items": {
+    "type": "object"
+  },
+  "type": "array"
+}

Added value: +{
+  "description": "Expected value per unit staked at market.price. Null whenever the model carried no blend weight — always null in Stage 1. Predictive-framework sports only.",
+  "type": [
+    "number",
+    "null"
+  ]
+}

Added value: +{
+  "description": "American-odds fair price implied by probability. Predictive-framework sports only.",
+  "type": [
+    "integer",
+    "null"
+  ]
+}

Added value: +{
+  "description": "[lo, hi] uncertainty band around probability. Predictive-framework sports only.",
+  "items": {
+    "type": "number"
+  },
+  "type": [
+    "array",
+    "null"
+  ]
+}

Added value: +"Market quote: {price: American odds, line: handicap/total line or null for moneyline, book: bookmaker slug when known (always set for predictive-framework sports)}."

Added value: +{
+  "description": "Parameter-set identifier, e.g. 'market_anchor' (Stage 1: no fitted model). Predictive-framework sports only.",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Added value: +"Legacy per-bet narrative sentence. Legacy sports only."

Added value: +{
+  "description": "De-vigged market probability, pre-blend. Predictive-framework sports only.",
+  "type": [
+    "number",
+    "null"
+  ]
+}

Added value: +{
+  "description": "The fitted model's own probability, pre-blend. Null until a model exists (all of Stage 1). Predictive-framework sports only.",
+  "type": [
+    "number",
+    "null"
+  ]
+}

Added value: +{
+  "description": "'quant' (deterministic stack) or 'full' (qualitative overlay, a later stage). Predictive-framework sports only.",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Added value: +{
+  "description": "Set when this bet is about an individual player.",
+  "type": [
+    "integer",
+    "null"
+  ]
+}

Added value: +{
+  "description": "Display name of the player or team this bet is about.",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Added value: +{
+  "description": "'home'/'away' or 'p1'/'p2'. Null for match-level tokens (OVER/UNDER).",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Added value: +{
+  "description": "Published probability, 0-1. Pure de-vigged market price in Stage 1. Predictive-framework sports only.",
+  "type": [
+    "number",
+    "null"
+  ]
+}

Added value: +"Legacy human-readable rationale bullets. Legacy sports only."

Added value: +"Legacy per-signal point breakdown. Legacy sports only — superseded by drivers below."

Added value: +{
+  "description": "Evidence strength, 0-1. In Stage 1: max of per-market movement depth and hours-since-first-quote maturity — not team-form sample size. Predictive-framework sports only.",
+  "type": [
+    "number",
+    "null"
+  ]
+}

Added value: +{
+  "description": "Set when this bet is about a team.",
+  "type": [
+    "integer",
+    "null"
+  ]
+}

Added value: +"Confidence tier: 'very_high', 'strong', 'moderate', or 'avoid'. Present for both shapes; always null in predictive-framework Stage 1 (no edge yet to rank)."

Previous value: -"string"New value: +[
+  "string",
+  "null"
+]

Added value: +"Legacy qualitative-validator verdict: {stance, confidence, delta, validated_at}. Legacy sports only."

Added value: +"Legacy: true when a bet is recommended. Predictive Stage 1 (MLS): always false when assessments exist; null when unavailable."

Added value: +"UTC datetime of last intelligence update. Predictive sports: max(assessments.computed_at)."

Added value: +{
+  "description": "League slug for this event, if any.",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Added value: +{
+  "description": "MLB-only probable-starter context (home_starter/away_starter with name/hand/era/confirmed). Absent for every other sport.",
+  "type": [
+    "object",
+    "null"
+  ]
+}

Added value: +{
+  "description": "Bookmaker bets[].market prices were sourced from. Predictive sports: assessments.market_book (not a ?bookmaker= overlay).",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Added value: +{
+  "description": "Home/away (or p1/p2) participant identification, keyed by role: {role: {name, player_id, team_id}}.",
+  "type": "object"
+}

Added value: +{
+  "description": "Sport slug for this event.",
+  "type": [
+    "string",
+    "null"
+  ]
+}

Previous value: -"Bookmaker slug; defaults to pinnacle."New value: +"Bookmaker slug for market prices on recommended bets. Defaults to pinnacle. Valid: pinnacle, fanduel, draftkings, betmgm, caesars, bet365, circa, hardrock, betonline."

Previous value: -"string"New value: +[
+  "string",
+  "null"
+]

Previous value: -"object"New value: +[
+  "object",
+  "null"
+]

Previous value: -"string"New value: +[
+  "string",
+  "null"
+]

Previous value: -"object"New value: +[
+  "object",
+  "null"
+]

Previous value: -"object"New value: +[
+  "object",
+  "null"
+]

Previous value: -"boolean"New value: +[
+  "boolean",
+  "null"
+]

Previous value: -"string"New value: +[
+  "string",
+  "null"
+]

Previous value: -"string"New value: +[
+  "string",
+  "null"
+]

Added value: +"Bookmaker slug; defaults to pinnacle."

Added value: +"Event id, from list_events, query_events, or search results."

Previous value: -nullNew value: +{
+  "properties": {
+    "analyst_take": {
+      "type": "string"
+    },
+    "available": {
+      "type": "boolean"
+    },
+    "bets": {
+      "items": {
+        "properties": {
+          "attribution": {
+            "items": {
+              "type": "string"
+            },
+            "type": "array"
+          },
+          "bet_type": {
+            "type": "string"
+          },
+          "confidence_score": {
+            "type": "number"
+          },
+          "coverage": {
+            "type": "number"
+          },
+          "market": {
+            "type": "object"
+          },
+          "narrative": {
+            "type": "string"
+          },
+          "rationale": {
+            "items": {
+              "type": "string"
+            },
+            "type": "array"
+          },
+          "signals": {
+            "type": "object"
+          },
+          "tier": {
+            "type": "string"
+          },
+          "validator": {
+            "type": "object"
+          }
+        },
+        "type": "object"
+      },
+      "type": "array"
+    },
+    "event_id": {
+      "type": "integer"
+    },
+    "has_recommend": {
+      "type": "boolean"
+    },
+    "intelligence_updated_at": {
+      "type": "string"
+    },
+    "match_overview": {
+      "type": "string"
+    }
+  },
+  "type": "object"
+}