Marvel Rivals Win Predictor

How does the Marvel Rivals 6v6 win predictor work?

Pick up to 6 heroes for Team 1 and Team 2 and the neural model returns a calibrated draft win probability. It is trained on 3,472,858 real ranked Marvel Rivals matches across 57 heroes — composition-only, so individual player mechanics are deliberately out of scope.

How accurate is the Marvel Rivals win predictor?

We publish probability calibration rather than a headline hit rate: across 65,407 held-out matches, the model's Expected Calibration Error (ECE) is 0.005 — when it predicts 60%, that lineup wins about 60% of the time.

Does the predictor account for Vanguard, Duelist, and Strategist roles and Team-Ups?

Yes. Because the model learns multi-hero interactions directly from ranked match outcomes, role balance (such as 2-2-2 vs triple-Strategist or solo-Vanguard) and hero Team-Up synergies are automatically reflected in the predicted win rate.

How does the Marvel Rivals pick assistant rank hero recommendations?

For any open slot on your selected team, the assistant scores every remaining hero with all other picks held constant and ranks recommendations by Δ percentage-point win-rate gain over the slot average.

Can both teams pick the same hero (mirror matchups) in the simulator?

Yes. Marvel Rivals allows cross-team mirror picks while forbidding duplicate heroes within the same 6-player team, and the draft board enforces those exact rules.

Is the Marvel Rivals win predictor free and open to AI agents?

Yes — 100% free in the browser with no account or API key required. AI assistants can also query the same calibrated model via our Model Context Protocol (MCP) server at batru.gg/mcp.

More detail: how the model is built and evaluated · hero tier list · MCP server for AI assistants