Scoring & Simulation — Cap 100

How projected wins, Monte Carlo odds, ranked score, and playoffs fit together after you finish a draft.

Projected wins. After five picks, Cap 100 builds a feature vector for your lineup and runs it through the promoted ridge model (with residual calibration when available). The result includes base wins plus bounded gameplay modifiers for role balance and positioning. Those modifiers are gameplay, not a second trained model.

Monte Carlo seasons. The engine simulates many full 82-game seasons with a seeded RNG so results are reproducible for a given draft. You see a win distribution, percentile-style intervals, and an explicit perfect-season probability rather than a single locked record.

Optimal lineup comparison. For the offers you were shown, Cap 100 also computes the best legal five it could have built under the same cap and roles. Comparing your five to that optimal set is how draft efficiency is judged for ranked craft.

Ranked score recipe. Public ranked scoring (balanced formula versions such as v2.5.x-balanced) emphasizes closeness to 82–0, then mixes underdog and efficiency terms with a pool multiplier so harder offer boards can boost craft. Exact weights are versioned and shown on Results when you finish a ranked run.

Playoffs path. When a lineup reaches the perfect-season threshold in Cap 100’s scoring path, playoff series can open. Wins and series progress can add to ranked score beyond the regular-season projection, rewarding teams that keep winning after 82–0.

What the meters are not. Category bars and strength meters on Results are a quick visual read. They are not the win model. Projected wins, the win range, and perfect-season odds come from the ridge projection and Monte Carlo run described above. Scout, Blackout, Aliases, and Jersey only hide or relabel what you see during the draft. Once a pick is locked, simulation uses the real player-season stats either way. Share links open the same five picks on Results for anyone with the URL. Ranked saves still use the server’s offer history for that draft, so a better board cannot be invented after the fact on the client alone.

How to Play · Strategy Guide · Game Modes · Scoring · About · FAQ · Leaderboard · llms.txt