OddsBeacon is a small research team — an AI/LLM engineer, business and marketing, plus friends who beta-test every release. Every match we've ever called is on the public record. The hits and the misses.
PRINCIPLES
What we publish, and why
01
Probabilities, not picks
A 64% is not a "lock". It's 64 — roughly two-thirds of the time it lands, one-third it doesn't. Honest about the variance.
02
Every miss on the record
No silent edits. No retroactive confidence changes. Every call — hit or miss — stays on the backtesting ledger exactly as published.
03
Calibration over confidence
A "HIGH" tag has to keep earning its historical hit rate. If it doesn't, the tier gets retrained. The calibration chart on Methodology and the full ledger on Backtesting are the receipts.
04
No betting advice
We publish numbers. You decide. We don't take a cut of any sportsbook, ever. Affiliate-free since launch.
THE BYLINE
Two builders, four testers
MC
Michael Copeland
FOUNDER & ENGINEER
AI / LLM · prediction models, RAG chatbot · built the platform and data pipeline
VS
Viktoria Simeonova
BUSINESS & MARKETING
Growth, brand, and go-to-market
A
Alex
BETA TESTER
Friend · club player · stress-tests picks before publish
S
Sam
BETA TESTER
Friend · sanity-checks confidence tiers on live matches
J
Jordan
BETA TESTER
Friend · flags confusing UI and bad calibration early
R
Riley
BETA TESTER
Friend · runs passes on tournament weekends
FROM THE DESK
"Every prediction we publish is the start of an argument with the future. We just want to be a little less wrong than the next person."