Why a great NBA team can look inevitable while a great MLB team still loses 60+ games.
Top-team win percentages are not just about “how good the best team is.” They are a collision between scoring frequency, possession volume, superstar leverage, roster parity, draw rules, schedule length, and plain single-game variance.
Click a league. Watch the dominance machine rewire.
The chart uses recent completed regular seasons. For NHL and EPL, the “points share” view accounts for overtime-loss points and draws; the “win %” view is literal wins divided by scheduled games.
Teaches: real top records are the fastest way to see how each sport’s format changes dominance.
Team
Dominance is signal minus noise, then filtered through rules.
The radar is an explanatory scoring model, not an official league metric. It translates sport structure into comparable design forces.
Teaches: game signal, star leverage, season length, parity, variance, and draw rules pull records in different directions.
Structural fingerprint
Higher area means the sport’s format more reliably lets the best team convert superiority into wins.
What is driving NBA?
Build your own sport and simulate the ceiling.
Move the sliders to see how game signal, parity, draws, and season length change an elite team’s expected record. Presets load the five leagues’ approximate structural profiles.
Teaches: repeated seasons separate stable structure from noisy one-year records.
Controls
Think of this as a toy model: useful for intuition, not a betting model.
Distribution of simulated season records
The bars show how often each win percentage appears over repeated seasons.
One simulated season
Green = wins, red = losses, gold = draws/non-wins.
Design the league. Watch dynasties fight the format.
Build a sport’s rules, simulate an era, and see whether the best teams actually become champions. Settings can be shared as a GitHub Pages URL.
Teaches: dynasties depend on talent persistence, playoff design, league size, and how reliably favorites survive.
League controls
This is an explanatory toy model. It is built for intuition, not forecasting.
Pick a mission
Tune the league rules to chase a grade. Scores use the expected model values above, not the example timeline.
Run an era to score the active mission.
Headlines are expected values across deterministic scenario samples; the champion timeline is one seeded example era.
Choose settings, then run an era to see whether your rules manufacture dynasties or chaos.
Champion timeline
Run the simulator to generate an era.
Even elite teams are prisoners of the single game.
Run 1,000 top-team-versus-average games. The more random the sport’s game unit is, the more often the favorite gets clipped.
Teaches: even a strong favorite can look fragile when a sport’s single-game unit is noisy.
NBA: Top team vs average team
High possession volume makes one-game randomness less dominant than in low-scoring sports.
Flip the cards. Each force pushes the ceiling.
Click or hover a card. These are the conceptual levers behind the numbers, independent of a single season.
What if one sport borrowed another sport’s physics?
The model below is deliberately stylized. Its job is to make the causal mechanisms visible.
Top records are not pure greatness. They are greatness multiplied by format.
The NBA tends to produce very high top-team records because the sport has many possessions, repeated shot opportunities, and stars who can influence a large fraction of high-leverage moments. MLB tends to suppress top win percentages because a single baseball game contains enormous randomness: starting-pitcher matchups, batted-ball variance, sequencing, and the fact that even elite hitters fail most of the time. The NFL can show huge percentages because 17 games is a small sample, while the EPL and NHL are constrained by low scoring plus draws/overtime-loss mechanics.
Structural ranking for “win% ceiling”
The real records are sourced. The mechanics are modeled.
Records were selected from recent completed regular seasons. The explanatory sliders and radar values are approximate, transparent heuristics.