Some background

This research abstract was submitted to MIT’s Sloan Sports Conference aka SSAC in 2011. It wasn’t accepted, but represents my first thoughts and foray into sports analytics, statistics, and gaming.

Key points of the research:

• Explored whether crowd-sourced fan predictions could rival statistical models in sports forecasting
• Developed a “participation engine” that gamified the prediction process
• Used game dynamics (rewards, leaderboards, challenges) to incentivize accurate predictions
• Focused on NHL player statistics as the test case
• Challenged traditional sabermetrics by suggesting collective fan wisdom could match or exceed pure statistical analysis

It lead to the development of Stattleship and Spogo, and also some work in Sports Innovation Lab.

Sabremetrics, as defined by Bill James, is “the search for objective knowledge about baseball” that does not intend to deal with subjective judgements such as a fan’s favorite player or enjoyment of a team rivalry.[1] It contends that analysis of a player’s career statistics will likely outperform a small sample of fans’ observations when predicting a player’s future performance due to bias or illogicality. The objective of this study is to examine whether or not a large collection of subjective judgements can answer objective questions about hockey, such as “How many goals will Corey Perry score next year?” or “What will Tim Thomas’s Save Percentage be at the All-Star Break?”

Through a participation engine and the use of game dynamics, fans are driven and incentivized to collectively predict a player’s statistical performance thorough competitive events. By leveraging their desire for reward, status, achievement, self-expression, competition, and altruism and combined with leaderboards, challenges, and levels fans are motivated to make the best subjective prediction possible. Results show whether or not these predictions are accurate and categorize findings to determine any correlation with bias or fan-preference.

This work offers a counter argument to sabremetric-based forecasts in that subjective observations, taken on a vastly social scale, can be as (or more) valuable a predictive tool as mathematical models based on empirical statistics.

1. David Grabiner, Sabermetric Manifesto (The Baseball Archive), http://www.baseball1.com/baseball-archive/sabermetrics/sabermetric-manifesto (1994)