Tuesday, October 4, 2011

Lessons from Moneyball

Over the weekend Nicole and I got out of the house and watched Moneyball. I believe this was the first movie that we have watched in the theaters since our daughter, Keagan, was born about a year and a half ago. We aren't too much into watching movies in the theater in the first place, but we figured we should go out an support a good story about data and statistics trumping “conventional wisdom”.

The movie is the (mostly) true story about how Billy Beane (Brad Pitt), the Oakland Athletics general manager, turned their very small budget into a wining team by rethinking what makes a winning team. The old timers in the organization wanted to replace key players from the team that were lost, where Billy questioned if there was a better way to build the team from the ground up. Billy meets up with a Yale graduate who looks at baseball through a statistical lens. And together they go about rebuilding the A's.

This movie should be seen by all college students about the power (and more importantly the pitfalls) of applying statistics to their profession. In the movie, Billy assembles his team of undervalued players which were gathered to stay under budget while maximizing the probability of winning each game, from a statistical standpoint.

So they used statistics to overcome their problem, happy ending right? Heh, we aren't even half way through the movie yet. The problem is now an unconventional baseball team is taking the field and the A's manager, Art Howe, is trying to play them like a conventional team -with disastrous results. This is the first great lesson that this movie teaches about statistics. If some decisions are made due to analysis, but then other decisions are made that do not follow the results, the resulting situation could be worse than if no analysis had been done, this is especially true if the analysis suggests results that are not conventional.

Luckily, Billy stood by his team and pulled some trades so that the correct players had to be played. So the A's took the lead in the division, right? After all, decisions were made based on statistical analysis. Not yet, see there is no magic bullet, and even though a good team was assembled and it was being used correctly there still was a lot of ground that the A's had to make up to catch the raw talent that the Yankee's have. Billy then embraces other data driven results about how to play baseball; bunting is abandoned, walking is welcomed, and they strive for easy outs. The movie doesn't spend too much time on this part, but does make note of it. The first important idea here is that there are many areas for statistical improvement in every process, an analysis leads to answers, but also leads to other areas for further improvement. Secondly, Billy took his findings and got the players involved in why the decisions were being made, he didn't just proclaim the decisions made by the analysis, but instead talked with the players about what he found and how the changes can make an improvement. It doesn't matter if something is highly significant, if the people controlling the process do not implement the results of the analysis it is a waste.

So the A's started to win and the great Moneyball experiment was paying off, they broke the record for wins in a row and things were going to plan. So they won the World Series right? Well, no, didn't even make it there. This might be the greatest lesson of the movie. Even though everything was correctly being followed they still lost. The problem is that the regular season is a large sample of games, and over the long haul the statistical results will hold up, but when we look at a playoff series there is only 7 games to play, and there is random chance -even the best teams can lose 4 out of 7 games. So the A's not making it to the World Series is just a case of random chance in a small sample, not a case of a bad analysis. And this is important to remember, generally in statistical analysis we are concerned about the long term results. We shouldn't get discouraged if the results don't come back good right away after implementation, give it time to work towards the long-term means.

It turns out that the ideas that the A's introduced to the league soon became a common way to view baseball and the Red Sox won their recent Championship using the principles.

Statistical analysis is like having an ace relief pitcher or a sure-thing pinch hitter -they could change the game, but only if used correctly.