Connection
Your API-Sports key lives in
.env on the server — never exposed to the browser. Run npm start then click Test.Season & week
Stats model weights
Offensive PPG25%
Def PPG allowed20%
Turnover diff20%
Yards per play15%
Home field10%
3rd down %10%
Elo settings
Optimised defaults — grid search across 2024 & 2025 seasons found
K=10, minElo edge=6 maximises win rate (61.3% in 2024, 56.6% in 2025, both >55%).
K=10 (slow/stable) outperforms K=20. Min edge=6 filters to only high-conviction picks. HFA has minimal effect at this edge threshold.
Load games from the Setup tab.
Elo power ratings
Build ratings above to see the Elo standings.
Stats model
Builds a composite power score from season averages. Best for mid-season when stats have stabilized.
GET /gamesSchedule, scores
GET /teams/statisticsSeason stats per team
GET /oddsSpread, moneyline, O/U
Elo model
Builds dynamic power ratings from every completed game result. Accounts for strength of schedule and recency — more reliable early in the season before stats accumulate.
Starting rating1500 for all teams
K-factor20 (adjustable)
Home field advantage55 Elo pts ≈ 2.2 spread pts
MOV multiplierlog(margin+1) × autocorr. correction
Season regression33% toward 1500 each new season
Spread conversion25 Elo pts = 1 spread point
Confidence tiers
Strong
Edge > 4 pts. Model diverges meaningfully from the line.
Lean
Edge 2–4 pts. Proceed with caution.
Avoid
Edge < 2 pts. Too close to the line.
Used today
—
Daily limit
—
Remaining
—
API call log
No calls yet.