About Sportslyx

Sportslyx takes the public schedule for NFL, NBA, MLB, NHL, soccer, UFC and golf and turns every game into a model-backed breakdown. The market line is de-vigged into a fair win probability; a per-game package of statistics, form, head-to-head history, injuries and availability is assembled from public feeds; a language model reads that package and the line and writes a structured matchup analysis; and after the final result the read is graded and counted in a calendar-month record that anyone can inspect on the Pick Tracker.

Around that core sit the tools a working analyst wants: a per-sport matches board with breakdowns, a company Board with a daily ledger, an AI Assistant for follow-up questions, same-game Combos priced from one score distribution, a Studio for custom stat-weighted models, and a pick tracker that grades your own reads automatically. An esports section exists behind a feature flag. Everything is built for adults who want the numbers before the noise.

Market first. Sports markets are efficient enough that a model which wanders far from them is usually wrong rather than clever. Sportslyx therefore starts from the line and asks the model to explain it — where the price looks fair, what it may be underweighting, what would have to be true for the other side to win — instead of asking it to out-guess a number set by informed money. The percentage you see is the market’s, de-vigged, everywhere in the product.

Transparency. The methodology describes how a line becomes a probability, what the language model reads and how its output is validated and cached. The data sources page names every feed and its known gaps. The grading rules — final score, box score for MLB props, draw as push, voids removed, golf previews outside the record, 15-minute cut-off, one calendar month in UTC — are the same rules the subscriber sees inside the product. There is no marketing version of the numbers.

Grade everything. The company Board, your tracked picks, Studio runs and Combos are all settled by the same jobs from official results. A record that is measured is the only honest claim an analytics product can make about itself, and the only way to see whether the estimates are calibrated.

Models are wrong often. A high percentage means the market prices one side as more likely; it never means the outcome is certain, and past performance shown anywhere in the product does not predict future results. If play stops being fun — betting more than planned, chasing losses, hiding it from people close to you — the Responsible Play page lists free, confidential help services and explains how to close your account in one step.