Balanced discovery uses local Elo hero ratings (starting at 1000, K = 24), plus a penalty for repeated one-sided head-to-head results. Weighted randomness keeps variety. Fresh rivalries favors your least-played pairings; pure random gives every eligible pairing equal weight. Excluded pairings are never selected. Recent pairings are avoided when alternatives exist. Sides are randomized.

No verified global matchup dataset is included. With no results, balance is unknown and selection is random. Hero complexity is not used as a substitute for strength. Keep both players on the same current rules and hero updates. For a fairer comparison, play a rematch with sides swapped.