How it works

Each step has one job. Inputs remain visible, progress is explicit, unavailable data is named, and the result can be reopened exactly as produced.

06

clear stages

01

Search the canonical player index so identity, team, and game context stay attached to the same record.

02

Use the schedule-backed matchup, venue, and game date instead of typing opponent context by hand.

03

Pick the statistic, line, side, and display window. The line remains your input, never a value the model silently changes.

OKC · canonical id 1628983

Schedule-driven: opponent, home/away and tipoff fill from the game

Validated to the market's line step; never auto-changed to the projection

Side
Display window

Preview of the composer controls, prefilled from the stored example. The window changes what is displayed, not what the model trained on. Create an account to run an analysis.

04

A visible progress state moves through validation, data loading, computation, and result assembly with explicit retry states.

If validation or data loading fails, the interface keeps the input and exposes a direct retry instead of showing an empty result.

05

Read projection, recent games, historical rates, matchup context, reasons, risks, freshness, and limitations separately.

06

Reopen an immutable saved result later, or carry its exact selection and probability into an eligible parlay draft.