Monte Carlo in a Nutshell
For each potential game, we use advanced analytics toestimate a win probability for both teams. We then simulate full tournaments by drawing outcomes from these probabilities. Repeating this process at massive scale yields a distribution of brackets that reflects the underlying model.
- Inputs: pre‑game strength metrics and matchup adjustments.
- Per game: sample a Bernoulli outcome with probability \( p(\text{team A wins}) \).
- Per bracket: propagate winners round‑by‑round for all 63 games.
Scale and Indexing
We compute and store bracket outcomes a trillion times before the tournament begins. As games finish, we filter the index to count how many simulated brackets remain consistent with reality.
Assumptions and Caveats
- Model error exists; simulation is only as good as its inputs.
- We report counts and frequencies; not betting advice.
Verify and Learn More
You can verify the dataset’s integrity and inclusion using our public commitments. See the Verify My Data page.
About Me
I am a guy that likes computers, math and sports. Reach out to me in the Contact page if you have any questions.