Pascal Institute
Research articles, case studies, and educational content on forecasting, uncertainty, model evaluation, market efficiency, and risk, alongside team ratings and weekly power rankings scored in public and The Decision Lab podcast.
/PASCAL
The Pascal Initiative
BETTER DECISIONS. BETTER OUTCOMES.
Founder, The Pascal Initiative
Independent research, education, and software initiative.
A simple question
WHY PASCAL?How can we make better decisions when the outcome is uncertain?
The Pascal Initiative brings together decision science, forecasting, data engineering, education, and software development to build practical tools for evaluating evidence, measuring uncertainty, and improving the decision-making process. Sports provide the initial proving ground: frequent decisions, measurable outcomes, rich data, and markets against which judgment can be tested.
In practice, that means publishing work people can check. Pascal rates every team in MLB, the NFL, college football, the NBA, and the NHL, turns those ratings into weekly Power Rankings, and for the NFL and college football records a win probability for each game before kickoff and scores it afterward against the betting market's closing line. Research articles and case studies, from a five-part study of MLB's ABS challenge system to fourth-down and pitching-change decisions, apply the same discipline. Pick of the Day invites visitors to make one pick per sport each day, recorded at the market price and scored on public leaderboards. The Decision Lab podcast talks through the cases.
After decades of building data systems, years working with professional baseball data, and completing a master's degree in data science, I became increasingly interested in what happens after we have the data.
More information doesn't automatically produce better decisions. Models can be wrong. Good decisions can produce bad outcomes. Bad decisions can occasionally be rewarded. Pascal grew out of my desire to understand that problem better—and build tools that help people navigate it.
Four working areas
PASCAL/*AnalytIQ
The model-development and evaluation environment behind Pascal's ratings and forecasts: building models, recording every prediction before the outcome, measuring calibration and accuracy, and comparing models with each other and with the market over time.
The Pascal Book
An upcoming book on decision science: how to turn evidence and forecasts into action, set the threshold for acting before the outcome is known, and keep honest accounts of decisions so that skill isn't confused with luck. It draws on Pascal's own research and published track record.
Pick of the Day & Leaderboards
A public forecasting game: make one pick per sport each day, recorded at the betting market's price, and see how your judgment holds up on the Pick Leaderboard, in Pick Challenges, and against the Pascal Rating itself.
FUTURE NOTE
You don't have to outrun the bear.
Forecasting isn't necessarily about predicting reality perfectly. In many decision environments, you only need to estimate reality better than the alternative you're being offered. The hard part is knowing whether you're actually better—or randomness has simply made you look better.