An MLB team ranked prospects by optimizing scouts' conflicting opinions.

Preparing for the annual Major League Baseball draft is hard: with about 1,500 players selected each year, teams must evaluate and rank hundreds of prospects. Scouts file qualitative and quantitative reports, but their opinions often disagree sharply.

Georgia Tech researchers worked with a major league team to model the problem: suggest a consensus ranking of all players scouted by the team's representatives. The same methodology recommends in-season scout scheduling based on how much information each scout would add and the uncertainty in each player's true ranking.

The team used the optimization tool for two years, and a second major league team later asked the researchers to evaluate its ranking data.

Turn disagreeing scouts into one consensus draft ranking.

Consensus from many noisy expert opinions is exactly what optimization does well.

Scout scheduling follows naturally from the same model, targeting uncertainty.

Two teams adopting the tool shows it survived real draft-room use.

When experts disagree, don't average their lists by hand—model the aggregation explicitly, and you get a ranking plus insight into where the next observation would be most valuable.

The case was published in Interfaces in 2012; the tool stayed in use and a second team requested the evaluation.