Pages

Wednesday, June 12, 2013

Predicting the Pirates - A Fan with Statistics

I've been a Pirates fan since 1991 (I was seven years old then), and for the first two seasons, I thought the Pirates always made the playoffs. The next two decades beat that assumption out of me to say the least. For 20 of the previous 22 seasons, the Pirates have broken my heart over and over. During the 2000's, I stopped caring for a while. They were so bad that I had no glimmer of expectation that my Bucs would ever break the .500 line again.

...Then this decade happened. The Pirates have improved each of the last three seasons since 2010, which wasn't that hard considering how bad they had been. In some ways though, it's been worse to be a fan these last couple of years. Each of the last two seasons, we've been wooed by the promise that the Pirates might actually finish the season with a winning record, or better yet, maybe even make the playoffs. Both times, the Bucs came back to earth and broke our hearts all over again. Last year was the worst, when the Pirates entered were 18 games above .500 in August and still found a way to finish with a losing record.

This year, it seems like we might actually have a winning season. We've been consistently above .500 since April 19th, and have what "looks" like an outstanding pitching staff. I say that it "looks" like it is outstanding because it looked great last year too, before falling apart late in the season.

I wanted to get a good sense of where the Pirates should be at the end of the season based on the way they've been performing. Being a stats "guy", I went over to http://www.baseball-reference.com/, which is a sports statistician's goldmine. I pulled the team pitching and batting statistics for the last four seasons (2009-2012) to figure out the best predictors for end-of-season win totals. I looked at, literally, dozens of different stats and ran all sorts of models (nothing too fancy, but too boring to detail here). I found out that the two team stats that really seem to matter are ballpark adjusted ERA (ERA+) and ballpark adjusted OPS (OPS+). Together, these two team statistics account for 85% of the variability in teams' win totals at the end of the season. That's pretty impressive for two statistics.

ERA+ and OPS+ are essentially measures that attempt to normalize pitcher and batter performances in terms of both league averages and ballpark conditions. If a pitcher has an ERA+ of over 100 it means, accounting for the ballparks he pitched in, he allowed fewer runs than the average pitcher in the league. OPS+ does the same thing for batters, but with the more complicated OPS statistic (which accounts for both slugging percentage and on base percentage). If a batter has an OPS+ above 100, then that batter reached base, hit for extra bases, or did both more often than the average batter in the league.

Using the ERA+/OPS+ model to predict the number of wins a team should have at the end of the season, I compared the predicted number of wins to the actual number of wins that the Pirates had from 2009-2012. The model has a standard error estimate of 4.56, meaning that it should predict, within 4 1/2 games, the number of wins the Pirates had during those seasons. It did. Here are the results of my analysis -



The quality of the table isn't great, but it's easy to see that the predicted wins are very close to the actual wins. 2012 is the only year where the "Real Wins" were at the upper limit of the margin of error, and it actually looks like the Pirates should have finished worse that year.

2013 looks pretty murky for the Pirates to me. Based on the model, the Pirates should finish with around 86 wins, which is above .500 (yay!) but isn't likely to get them into the playoffs (boo!). It looks like the pitching staff is, indeed, performing better than the average staff in the league, but not by a whole lot (Texas and Atlanta have team ERA+'s of 123 and 122, respectively). The batting is below average, but no where near the league cellar (The White Sox and Marlins have teams OPS+'s of 77 and 70, respectively).

To sum it up, the Pirates are the definition of an average team this year. They are a little above average in pitching, a little below average in batting, and should finish somewhere in the neighborhood of .500 this year if they stay that way (a big if). The lowest estimated win total for the year is just below 82, which is the magic number for a non-losing season, so I would expect things to get pretty tight around August-September. Until they get their OPS+ at or above 100, I don't expect the Pirates to contend for the division or the Wild Card. As a fan, I'd love for my Bucs to wind up in the playoffs, but right now, I'll settle for a realistic expectation that they will be just a little better than they were last year.