The Orioles & Run Differential

The Orioles’ poor run differential has become quite the topic recently.  I have given the issue quite a bit of thought over the past two months, even to the point of starting and stopping writing blog entries on the subject at least a dozen times.  Over the past couple of weeks, the topic has been heavily debated in several different venues.  I’ve read some very good discussions at Orioles Hangout and other Orioles message boards, ESPN.com and Sports Illustrated have ran features on the phenomenon, and the usual talking heads on the cable baseball shows have weighed in on it as well.  In other words, almost everything that can be said or written on the subject already has been.  That’s not going to stop be from chiming in, but I just want to hit on a couple of high level points rather than wading through the details again. Can a team operate in such a way as to overcome a poor run differential or is any variance between a team’s Pythagorean (expected) record and actual record strictly a function of luck?

This is really the starting point for any Pythagorean record discussion.  I obviously don’t have an answer to the question.  The sabremetricians would point out that allowing for a standard deviation of approximately 3-games, the Pythagorean record matches up very close with actual records over a 162 game season.  I have never seen an actual figure cited, but I get the impression that we are talking in the neighborhood of a 90%+ success rate when allowing for the 3-game standard deviation.   This side also points to the fact that only four teams have made the playoffs with a negative run differential as further proof that run differential is a good gauge of what a team’s actual record should be.  That’s a bit skewed since the number of playoff teams have changed over time, although only 14 teams have finished with a negative run differential and won more than 85 games in a season since 1969 which demonstrates the same point.  Analysis of Pythagorean record and actual records over the years illustrates a similar situation – the Orioles are currently outpacing their expected record by a near historic pace.  With such strong historic correlation between run differential and actual records, this side logically concludes that is very difficult for any team to make it through an entire season with a large variance and if they do, it is purely a fluky and lucky outcome.

The other side argues that that the 2012 O’s are built in such a manner as to create a perfect storm situation for a low run differential team that still wins games.  The Orioles’ strong bullpen is most often cited in this argument.  A good bullpen will keep games’ close – they can hold one run leads and keep games from getting out of hand so that the offense can score just enough to win.  The idea is that as long as the O’s starters keep the game close enough to get to the bullpen, Buck Showalter will call on his relievers to do their job, and that usually results in some close games.  Another argument is that those starters – specifically Jake Arrieta, Brian Matusz, Zach Britton, and Tommy Hunter – have been wildly inconsistent this season.  When they’ve been decent, they’ve been decent.  When they’ve been bad, they have been really bad.  Throw in a mediocre run producing offense and there does appear to be a strong foundation for a team that that outperforms their Pythagorean record but does so in a perfectly justifiable and explainable manner.

I sit somewhere in the middle.  There is no doubt that as the historic evidence shows, the Orioles would be in nearly-unprecedented territory if they were to reach the postseason with a near -50 run differential.  It is impossible to ignore that evidence and we shouldn’t want to ignore it.  When something happens that infrequently, it is perfectly logical to view its occurrence as random or luck driven.  At the same time, it does make sense to me that there are ways a club can be constructed to outperform their expected record – a strong bullpen being the main way.  Why don’t more teams outperform their Pythagorean records then?  I don’t know the answer to that, although I would begin any potential research with the hypothesis that consistently good bullpens that throw a ton of inning (O’s relievers have already logged 370+ innings in 2012) are an extreme rarity, therefore very few teams are ever really in a position to be able to outperform their Pythagorean record the way the Orioles have.

There is no easy answer.  My main takeaway from this debate is that I hope if the Orioles do continue to defy conventional run differential wisdom, that the bright sabremetric minds out there really dig deep to see if there is more to the team’s success than pure luck or good fortune.  Pythagorean records are a useful tool, but like any metric, they should not be viewed as a be-all-end all.  We should not lose track of the reason the sabremetric revolution happened in the first place – which was to dig deeper into statistics and challenge conventional wisdom.  Now that run differential and actual record variances have become conventional wisdom, it would be a disservice not to attempt to dig a bit deeper, test the theory, and make sure that some other explanation is not being missed.

Are the 2012 Orioles capable of playing better – in regards to run differential – over the next 51 games in order to make the playoffs?  How much better does the team need to play for that to happen?

The consensus opinion – one that I also agree with – is that the O’s do need to perform better over the last 30% of the season than they have over the first 70% in order to best position themselves for a playoff position.  There is just too much evidence suggesting otherwise to believe that the Orioles can continue to outperform their Pythagorean record to the extent they have thus far.  They might be able to,  but it would be foolish to assume that they definitely can.

The fact of the matter is, even if the Orioles were historically fortunate during the first four months of the season (and I think there was more to their success than just luck), those games are in the past – those wins have been banked.  As easy as it would make things, the universe does not operate in such a manner where variances and oddities even out over time.  It just doesn’t.  If you are flipping a coin 100 times and after 90 flips, it has landed on heads 50 times and tails 40 times, the odds of landing heads on the 91st flip are still 50%.  We would expect to land 50/50 heads and tails, but that doesn’t always occur.  Likewise, just because the O’s have outperformed their Pythagorean record through the first 110 games does not necessarily mean they will underperform going forward to even out their final record.  The results of the first 110 games have no bearing on the results of the last 52 in terms of over or under performing run differential.

Regardless of whether or not the O’s are lucky to be 60-51 on August 1st that is their record, which means the O’s need to go somewhere around 28-23 or 30-21 over their final 51 games to rack up between 88 and 90 wins.  88-90 wins is most likely the minimum amount needed to reach the playoffs.  Using the Pythagorean record and making assumptions on runs scored and runs allowed based on the O’s production up to this point in the season, I calculate that a +25 run differential (237 runs scored to 212 runs allowed) would result in 28 wins over the final 51 games.  The O’s run differential at the end of the 162-game schedule would STILL be -22.  That is an important point – at this point the O’s do not need to post a positive run differential at the end of the season in order to reach the playoffs.  They just (assuming they do not continue to outperform their expected record) need to post a positive run differential from here on out.

Can they do it?  The general consensus is that it isn’t likely.  Teams often do not have a +75 swing in run differential from the first four months of the season to the last two months.  Of course, most teams aren’t able to replace two very poor performing pitchers (Arrieta and Matusz) with two above average performing pitchers (Chris Tillman and Miguel Gonzales).  There are no guarantees that Tillman and Gonzales will continue pitching so well, but the possibility is there.  If they do, I do believe it is possible for the O’s to throw up a positive run differential down the stretch.  Add in the bump expanded September rosters can provide and Jason Hammel’s impending return from the disabled list and I have a difficult time buying into the idea that it is impossible – or even near-impossible – for the Orioles to show marked improvement of the final two months in regards to run differential.  At any rate, I think the possibility for them to do so and the foundation to do so is certainly there.  Putting up a +25 run differential over a 51 game stretch is not really that farfetched if the starting rotation performs well.

That’s where I am with the great run differential debate.  The O’s may or may not have been lucky to get themselves into the position they are currently in.  At this stage, that is largely irrelevant.  If the Orioles can play at a higher level (+25 run differential) over the final 50 games of the season, they have a good shot at making the playoffs.  We just have to wait and see if they are up to the challenge.