Algorithmic sports analytics
Every pick logged.
Every loss shown.
A model produces our picks across NBA, NHL, MLB, EPL, and cricket. Edges, odds, and units are published before the game starts — wins and losses go in the same public sheet.
Public track record
Every pick we publish lands in this sheet at the same time it goes to Instagram and Telegram — with the odds we got, the model's fair price, and the edge in percent. Closing line value (CLV) is filled in the next morning.
The full sheet is public and view-only. Filter by sport, by week, by edge tier — it's all there.
Open the public sheet →How a pick gets made
-
01
Scrape today's games + market odds
One scraper per sport. Odds from The Odds API across multiple regulated books.
-
02
Run the sport's model
NBA & NHL: Elo with rest, pace, and home-court adjustments. MLB: pitcher-adjusted win prob. EPL: Dixon-Coles Poisson. Cricket: format-specific.
-
03
Filter by edge ≥ 3%
If the model's fair probability beats the market's implied probability by less than 3 points, no pick. Most days have 0–3 picks total. That's the point.
-
04
Quarter-Kelly sizing, capped at 2u
Full Kelly is too volatile for a public record. We size at quarter-Kelly, hard-cap at 2 units, and never use "max play" or "lock" language.
-
05
Log first, post second
The pick lands in the public sheet before it's posted to Instagram or Telegram. The timestamp is the proof.
Things we don't do
- ✕ Charge $300/week for "guaranteed lock VIP picks"
- ✕ Hide losses or quietly delete posted picks
- ✕ Post fake DM screenshots or fake payout receipts
- ✕ Use "lock", "guaranteed", "fixed", "inside info", "rigged"
- ✕ Sell locked picks behind a paywall (legal distinction we take seriously)
- ✕ Use sportsbook affiliates outside Ontario's AGCO-licensed list