← Back to demo hub · Dashboard (demo data) SportAI preview
Sport AI
NCAA Basketball Analytics Platform

Smart Spread Filter Active

Based on 21,789 backtested games, spread picks are now filtered by edge size. Only games with 1.5+ point edge are recommended — turning a break-even model into a profitable one.

56.5%
2-3pt Edge
61.4%
3-5pt Edge
88.4%
5+ pt Edge
Totals Win Rate
57.5%
16,219 games backtested
Totals ROI
+10.6%
All picks, standard juice
Spread ATS (Filtered)
61.4%
Edge 3+ pts, 1,925 games
Today's Games
8
Vetting cards generated
Model Features
68
Including 2 ghost variables

Today's Predictions

Live — Auto-refreshed

Your Profile

SAVED PROFILES

Ghost Variable Sandbox

Toggle any of the 68 model features on or off to see how the prediction changes. You can also add your own custom "ghost variables" — these are extra factors you think matter that the model doesn't track yet. Your sandbox is completely private and doesn't affect the master model.

Add Your Own Ghost Variables

Think something matters that we don't track? Add it here. Choose a category and set the impact level — the system will calculate the point value for you.

Model Performance

2023-24 + 2024-25 Seasons
Totals Overall
57.5%
9,318 W / 6,901 L (16,219 games)
Totals ROI
+10.6%
+$1,060 per $10K wagered
Spread ATS (Filtered)
61.4%
Edge 3+ pts (1,925 games)
Spread MAE
8.9 pts
Market MAE: 19.3 pts (2x better)

Totals Win Rate by Edge Size

Cumulative ROI Over Time

Probability Calibration

Spread ATS by Edge Tier (21,789 games)

Edge TierGamesATS Win %Model MAEMarket MAEVerdict
0 - 1 pts76150.7%9.319.32NO PLAY
1 - 2 pts82652.8%9.239.23LEAN
2 - 3 pts95056.5%8.929.06PLAY
3 - 5 pts1,92561.4%8.9519.30STRONG PLAY
5+ pts17,32788.4%8.9519.30LOCK
Totals by edge: 0-1pt 50.4% | 1-2pt 56.6% | 2-3pt 59.3% | 3-5pt 66.4% | 5-7pt 72.5% | 7+pt 81.8%

Bottom Line Assessment

Totals model is production-ready. 57.5% hit rate across 16,219 games with +10.6% ROI is strong and consistent. Win rate scales cleanly with edge: 50.4% at tiny edges up to 81.8% at 7+ point edges. This is exactly the pattern you want — higher confidence = higher accuracy.

Spread model is now profitable with smart filtering. Raw ATS across all games was break-even, but our new edge-based filter changes the picture completely. At 2-3 point edges: 56.5% ATS. At 3-5 point edges: 61.4% ATS. The model's MAE is 2x better than the market — the key was only recommending bets when the edge is meaningful (1.5+ points).

Content Hub

Social Media & Podcasts
Posts This Week
12
Across all platforms
Published
8
Approved and live
Pending Review
3
Waiting for approval
Podcast Episodes
4
Season 1

Create New Post

Record Podcast

Developer API

Sell Your Data
API Keys Active
0
Across all customers
Requests Today
0
All endpoints
Data Points Available
117K+
Spreads + Totals + Draft
Pricing
3 Tiers
Free / Pro / Enterprise
Free
$0
Perfect for exploring the data
30 requests / minute
1,000 requests / day
Historical spreads & totals
Team efficiency ratings
Draft prospect rankings
Pro
$49/mo
For analysts and DFS players
120 requests / minute
10,000 requests / day
Everything in Free +
Live predictions (today)
Full vetting card details
SHAP feature breakdowns
Enterprise
$199/mo
For apps and analytics shops
600 requests / minute
100,000 requests / day
Everything in Pro +
Bulk CSV exports
Webhook notifications
Priority support

API Endpoints

Base URL: https://api.sportai.com/api/public/v1

Key Management
POST/keysAUTH
GET/keysAUTH
DEL/keys/{key_id}AUTH
Historical Data
GET/data/teams?season=2025FREE
GET/data/historical-spreads?season=2024-25&limit=100FREE
GET/data/historical-totals?season=2024-25&limit=100FREE
GET/data/draft-board/2025?limit=50FREE
Live Predictions
GET/predictions/todayPRO
GET/predictions/{game_id}PRO
Bulk Export
GET/data/export/spreads?season=2024-25ENTERPRISE

Quick Start

# Get your API key first, then:
curl -H "X-API-Key: sai_live_your_key_here" \
  https://api.sportai.com/api/public/v1/data/teams?season=2025

# Python example:
import requests
headers = {"X-API-Key": "sai_live_your_key_here"}
r = requests.get("https://api.sportai.com/api/public/v1/data/draft-board/2025", headers=headers)
prospects = r.json()["prospects"]

NBA Draft Prospect Evaluator

68,782 players scored
Total Prospects Scored
68,782
All NCAA seasons in database
Verified Draft Picks
1,060
Actual NBA draft selections
Model Features
19
Stats + team metrics + position
Current Season
2025
Latest scored prospects

Scouting Notes & Watchlist

0 notes

My Watchlist

My Notes