Challenge the Model

Challenge the Model is NovaPredict's answer to "I could beat your algorithm." Sign in, and for Nova's calls you log your own position: Agree, Override, or Skip. Over the season, your calls are scored against the model's. It is the same accountability NovaPredict applies to itself, extended to you. The model's projections are locked before kickoff and graded in public; with Challenge the Model, your overrides are held to the same standard. Accounts are live now, so you can start logging calls today.

What It Does

• Lets you log one of three calls against any Nova position: Agree, Override, or Skip.
• Scores your calls against the model over the season, so the you-versus-Nova question gets a real answer instead of a hunch.
• Requires an account. Sign in to start; user accounts are live.
• Extends NovaPredict's core philosophy of accountability by default to the user. If Nova has to prove itself in public, so does your gut.

What It Does Not Do

• It does not change Nova's projections. Your Override records your disagreement and scores it; the model's call stands as its own locked, graded projection.
• It does not carry any pre-season track record, for you or for the model. Scoring accumulates over the live season, and NovaPredict publishes no backtested performance claims.
• It does not penalize you for skipping. Skip is a first-class option for calls where you have no opinion.

Status

Live now: user accounts and the full Challenge loop, meaning you can sign in and log Agree, Override, and Skip calls today. The season-long scoring of those calls fills in as the 2026 season plays out, alongside the model's own public record, which starts Week 1.

See how the model itself is graded at /accountability. Related: The Accountability Record, the standard your calls are measured by.