Pittsburgh Pirates Research Room: is the team actually overachieving?

The common thought is that the Pittsburgh Pirates are underachieving, but a closer look at run differential shows the opposite is true

I am by no means a Pittsburgh Pirates pessimist, but I am also a big believer in statistics and probabilities.

As I write this article, we are just a day away from the July 31st trade deadline, the Pirates are four games under .500 and sit 5.5 games out of first place in the division.

Statistically, that’s not a great look for a team hoping to get a playoff berth.

However, if there were some statistic that had a high predictive power on team record, that showed that the Pirates were truly under-achieving, we could use that statistic to project where the Pirates may be headed. In other words, if this hypothetical statistic showed that the Pirates were due for some positive regression, we might feel more comfortable with being a buyer in trades etc.

Fortunately we do have such a statistic; it’s a derivation of the Pythagorean Expectation, which predicts a team’s winning percentage based on the number of runs they’ve scored and the number of runs they’ve given up. The derivation we will be using is much simpler than that, we will be only be using Run Differential, which is just Runs Scored minus Runs Given Up.

Run Differential is quite a useful statistic for calculating how many games above .500 we expect a team to be. It ends up that for every 10 additional runs above 0 a team is in Run Differential we expect a team to be about one game over .500, and for every 10 runs below 0 in RD, we expect a team to be 1 more game under .500. This correlates at about 93 percent.

For those interested, here’s what the statistical analysis looks like over the last 7 years in baseball;

Wins = 81 + 0.1 * Run Differential

Here we have our constant at about 81 wins and the coefficient to Run Differential is about 0.1. This means that a Run Differential of 0 should mean about 81 wins, and that each additional 10 runs above that 0 mark, we would expect to be worth 1 win over .500 (81+0.1*10=82).

While this 10 runs/win mark may seem excessive at first glance it makes more sense over the course of a season. A good team would tend to win games by some larger margins of runs than they lose games by, and vice versa for bad teams. To give an example, a team that wins 4 games by an average of 3 runs, and loses 2 games by an average of 1 run would only need to play 6 games to have a +10 Run Differential and would be exactly 1 win over a 3-3 record.

Now we have a way of determining what team records should look like, including the Pittsburgh Pirates, and can analyze that against what they do look like, to determine over and under performance. To do this we just have to take a team’s actual games over .500 and then subtract their expected games over .500 by using Run Differential/10.

This statistic will be denoted “GO.5-RD/10” and from it we can tell if a team is under or over performing. A team with a positive GO.5-RD/10 means they have more wins than expected, and thus are over performing, and a team with a negative GO.5-RD/10 would have fewer wins than expected and would be underperforming.

Here is a graph of the 2017 Pittsburgh Pirates game by game RD/10 (Blue), Games over .500(Red), and GO.5-RD/10 (Green);

You can see how closely RD/10 and GO.5 match each other, when one goes down or up so too does the other, that is until about 7/15. At 7/15, the last game the Pirates lost before going on a 6 game winning streak, there is a sudden divergence of the two lines. The high water mark came on 7/21 when the Pittsburgh Pirates beat Colorado 13-5 and reached 1 game over .500, the problem was that their Run Differential only peaked at -11, or an expected 1 game under .500.

What does it all mean?

What does this mean? If you look at the green line you can see the Pirates have been over performing their run differential by somewhere between .5 and 1 game most of the season, until their hot streak after the All-Star Break when they over performed by nearly two games. This is a sustainable level of play; a standard deviation from the mean is about 4.4 games league wide, so if the Pirates were over performing by closer to 4.4 games we would expect some regression, but playing at around .5-1.5 games better than expected is sustainable.

The McCutchen Effect

Another thing to note is the impact of Andrew McCutchen on this team. If we take team Run Differential up until 5/25, the last day before McCutchen began heating up, and compare it to their Run Differential after 5/25 the contrast could not be more clear. Here is the Pirates Run Differential prior to 5/25 (Blue) and post 5/25 (Red);

The light blue and light red lines are the trend lines of RD/10 pre 5/25 and post 5/25 respectively, or where we project the Run Differential to head. The Pirates were projecting to be a team with a nice draft spot, likely ending up more than 10 games under .500 by the end of the season prior to 5/25. In late May, however, there was a sudden turnaround sparked by McCutchen’s hot bad, stabilized and actually pushed the trend line slightly positive. If this trend holds, which I think is likely, the Pirates should sustain a push for a winning season and potentially a playoff berth. This sustained upwards trend is likely because, with the addition of Starling Marte, and a hopefully healthy Gregory Polanco coming off the DL, their Run Differential should to continue to trending upward and perhaps at an even faster pace.

One final thing to look at is how the Pittsburgh Pirates are doing win respect to the Division. Here’s a table of the team Records, Run Differentials, and GO.5-RD/10, sorted by Run Differential;

So we would expect the division to break down with the Cubs in 1st, Milwaukee and St. Louis to switch 2nd and 3rd places, and the Pirates to be in 4th. Also it would seem as though the Cardinals are due for a bit of regression towards a better record underperforming by more than 3 games.

What’s comforting about this information is that the division is not out of reach; no team is primed to start winning like crazy and run away with the division, because every team is largely playing at their expected level. What is disconcerting is that the Pirates are 5.5 games back and have a lot of ground to make up in a little more than 2 months of the season left.

All in all the Pirates are a team with the right trajectory, they’re record reflects their play pretty accurately. Furthermore the Pirates are a team on the right trajectory, with a Run Differential that is steadily getting better; they project to be a team around .500 come October 1st. This means that they likely won’t be a playoff contender, unless they get productive offensive play from Marte and Polanco down the stretch, in addition to continued production from McCutchen and Bell.

To answer the question of if the Pittsburgh Pirates are underachieving, the answer by the statistics is no, in fact they’re actually overachieving a bit. A less mathematical argument could be made that this team is underachieving relative to where they would have been with a hot — or at least capable — McCutchen from the start, an unsuspended Marte, and an available Kang, but worrying about those things is for the off season. Right now the Pirates should be focused on winning games and continuing to shrink their Run Differential, in their hunt for the division.