NuroPicks Data Stack

Last updated: 2026-04-19. Status: reference doc. Pulls from internal 2026-04-16 research.

NuroPicks runs on a layered odds + settlement stack. We start cheap and graduate to enterprise feeds as volume justifies the spend. This doc explains what each provider does, what it costs, and when to graduate.

Layer 1 - MVP odds (what we use today)

The Odds API. $30 to $249/mo. 40 sportsbooks, all major sports, historical data back to 2020. Live odds at 30-60 second cadence on the paid tiers. Good enough to launch with; cheap enough to eat at scale.

Why we picked it: documented, SLA, historical archive for backtests, pricing that starts at hobbyist and scales with us. Every pick surfaced on the bot and the web right now routes through this feed.

Layer 2 - cheap breadth

API-Sports. $10 to $19/mo. 2000+ competitions across niche leagues. We use it for secondary leagues that The Odds API underweights (cricket, some lower-tier soccer, esports coverage beyond headline events). Not the primary source.

BALLDONTLIE. Free. NBA, NFL, MLB, NHL, EPL stats. We use it for roster and basic stat lookups, not odds.

Layer 3 - production settlement

SportsDataIO. Custom enterprise pricing. 15 sports, 15 years of history, injury feeds, settlement-grade scores. Graduation target when the bankroll needs settlement verification (~10 to 15 minute post-game delay is normal for production-grade settlement). We will add this before we let users place real-money-tracked parlays at scale.

Layer 4 - enterprise real-time (later)

OpticOdds. Custom pricing, 200+ books, 1M+ odds per second. Graduation target when we run millisecond-sensitive arbitrage or steam-tracking features. Not needed pre-Series-A. Noted here so we do not rebuild the feed layer when we need it.

LSports TRADE module. Sub-second settlement for enterprise use cases. Same layer 4.

Optional - oracle layer

Chainlink sports oracle. For any tamper-proof public record of pick outcomes (track record transparency, verifiable CLV claims). Not used today. Flagged as an option because our 50/50 capper revenue-share model benefits from auditable pick history.

APIs we deliberately AVOID

• ESPN unofficial API. No SLA, undocumented, can break without notice. Using it is a production incident waiting to happen.
• Direct DraftKings or FanDuel scraping. ToS violation. Gets our IPs banned and our users limited faster.
• Pinnacle API. Closed to public access since July 2025.

Settlement approach

1. Subscribe to a settlement-grade feed (SportsDataIO or LSports TRADE at layer 3+).
2. Cross-reference 2 to 3 official sources before finalizing.
3. Flag mismatches for manual review; do not auto-settle a contested result.
4. Use a 5 to 15 minute post-game delay for high-value bets so corrections land before we credit.
5. Event-process via webhooks, not polling, on any feed that supports it.

How this maps to features

• /picks, /odds, /predict - layer 1 (The Odds API).
• /settle, /mybets auto-settle - layer 1 today, layer 3 at production scale.
• /bet, /parlay tracked outcomes - layer 3 once real-money-tracked parlays roll out.
• Capper leaderboard verifiable CLV - layer 3 with optional oracle layer for public proofs.

Cost ceiling

Current combined feed spend: under $250/mo. Target under $500/mo through Q3 2026. Layer 4 graduation comes with a minimum $3k to $10k/mo line item, so we plan for it, we do not accidentally trigger it.

Who owns this

Infra feeds: owner David. Any change to the feed layer or the settlement cross-reference logic goes through him. Issues in settlement should get flagged in #dev-chat with the affected bet_id and the conflicting feed sources.