Background

The databurst MLB power rankings are generated from a quantitative mathematical model constructed from the current season series win-loss records for each team. The model generates a ranking value for each team proportional to the strength of the team. The mathematical analysis is an iterative model which determines a best-fit prioritized ranking for each team that most accurately matches the overall results from the MLB season. The model determines the strength of each team, as well as a home field advantage factor.

Since each team in MLB plays all teams within their own league, and a number of cross-over inter-league games, the relative ranking of the teams is generated through head-to-head and relative performance versus common opponents. There are natural strength differences between the AL and NL, as well as within the different divisions, and these relative strength of schedule effects are implicitly included in the model.

Model Details

Base Model - all season series matchups and results are weighted equally in the model.

Momentum Model - increased weighting is place on recent series to capture short term trends.

Series Outcomes - series victories are weighted for differential in games won. Weighting factors = 1.3 for 3+ W differences, 1.2 for 2 W differences, and 1.0 for single W differences.

Home Field Advantage - determined as part of the overall model generation process, and currently has a value of 61 team rating points.

Predictive Accuracy - retrospectively the base model correctly predicts 71% of the series outcomes.

Rating

The rating for each team is a numerical value which can be used to determine the probability of a team winning a series matchup. Larger ranking numbers indicate stronger teams, and the difference in rankings between two teams is a measure of their strength disparity. The generic win probability is the is the probability of a given team winning a series against the mean level MLB team.

The win probability for the home team for any matchup is calculated by WP = 1 / (1 + 10^(dR)/400), where dR = home team ranking - away team ranking + homefield edge. The win probability of the visiting team is 1 - WP.