How EdgeIQ’s picks work
A look inside the EdgeIQ pipeline — the six data providers that feed every pick, the per-sport cadence that decides when those picks refresh, and the FREE, Starter, and Elite tiers that decide how much of the model you actually see.
EdgeIQ runs on a six-provider feed stack — every input flows through one of these before a pick ever lands on the board. SportRadar carries the canonical game schedule, league transactions, and box scores across NFL, NHL, CFB, NBA, MLB, and WNBA. OpticOdds consolidates the sportsbook market layer in one normalized stream — full-game, derivative, and player-prop markets — so the model sees a single shape of market rather than wiring up each book individually. The Odds API is the second-look at odds, layered in as a reconciliation source against OpticOdds. balldontlie drives NBA, WNBA, and CFB schedule and roster coverage. SportsDataIO rounds out NHL injury reports and starting-goalie confirmation. API-Sports delivers the NFL and MLB confirmation flows — pitcher locks, lineup declaration — that gate what the model is willing to emit at all.
The cadence differs by sport, because the inputs do. NFL picks run once each morning of gameday ahead of the noon kickoff window, with a Thursday-night advance run the night before. NHL generates at 13:30 UTC once starting goalies are publicly confirmed — typically noon to 2 PM ET for a 7 PM puck drop. MLB runs once the morning of after the starting pitcher is confirmed, then re-runs six hours before first pitch to catch any bullpen-usage shift. NBA generates once each gameday afternoon ahead of tipoff; CFB runs in the morning with a wider damper pre-Week 1 so the post-portal reset is priced in.
Confidence sits clamped between 50% and 78% on every sport — no pick is ever a lock, and a calibrated 78% pick misses roughly 22% of the time by construction. Each pick carries a component-agreement score alongside the raw confidence, so a model that hits 71% with all three sub-models in agreement reads differently than 71% with two out of three.
What each tier unlocks
Three plans, one model. The gate is depth of disclosure — not the engine itself.
FREE shows one headline pick per sport. No player props, no per-pick component breakdown, no statistical edge. It exists so visitors can read the output and judge whether the modeling looks honest before paying anything.
STARTER is the full slate. Every sport, every market the model emits, on the day the slate publishes. Each per-pick breakdown shows the three sub-models side by side, the component-agreement score, and a line-movement panel across the pre-game window at DraftKings, FanDuel, and BetMGM. The internal model is the same — STARTER just surfaces more of the math that the FREE headline obscures.
ELITE layers on player props, per-pick statistical edge breaks, and priority email delivery. Subscribers see what the model thinks of individual player lines — yards, receptions, shots on goal, strikeouts — alongside the slate-level picks, and the email digest lands in their inbox before the public-facing board refreshes. ELITE is for readers who want to grade the model’s reasoning on individual legs, not just the headline market.
Where to start
Three sentences. Read the methodology page — full architecture, components, and the limitations list — so you can see what the model is and isn’t confident about. Pick one sport to follow on the public board for a week; the FREE headline on that sport is enough to judge whether the model’s structure tracks with how you read the game. Decide on Starter vs Elite based on whether player props matter to your book — if you’re a totals-and-sides reader, Starter is the right tier; if you size individual player lines, ELITE pays for itself in the first week.