Manager John Farrell to the Red Sox: An Analysis

New Red Sox Manager John Farrell

The Boston Red Sox finally landed their man over the weekend, acquiring Toronto Blue Jays’ skipper John Farrell for infielder Mike Aviles. The Red Sox, embarrassed by the team’s worst performance in decades under the chaotic leadership of Bobby Valentine, took the unusual action of trading a player for a manager. Historically, these types of deals have been rare, with very mixed results for the team acquiring the manager. For this trade to make sense for Boston, they must have felt that John Farrell is significantly more valuable than any other managerial candidate they considered. But how much more productive will Farrell have to be to justify the trade?


The first part of the question is relatively easy, as the Red Sox have already answered it. Farrell simply has to be one “Mike Aviles” more productive. Here, we will express Aviles’s production as Wins Above Replacement (WAR), a metric which assigns a value in terms of wins above what a replacement level player (veteran AAA or AAAA) would contribute. According to Baseball Reference, in 2012, Aviles had a WAR of 2.0, a solid total for a fringe everyday player. Aviles was likely going to be a utility infielder for the Red Sox in 2013, but he still had significant value due to his defensive ability and versatility, and tolerable production at the plate. He’s first time arbitration eligible this offseason, and will receive a sizable raise, although he still will be an inexpensive option for Toronto at one of the middle infield positions. Overall, I think it’s reasonable to suggest the Red Sox gave up 2.0 WAR to get Farrell; or in other words, that Farrell will contribute at least the equivalent of 2.0 WAR next season.

But here is the problem.

There is no way to convert a manager’s performance into value. This is largely due to the fact there is no agreed upon or reliable method for evaluating managerial results. Classic statistics like win-loss record, pennants, and World Series victories may be relevant in Hall of Fame discussions, but they aren’t particularly useful when attempting to predict future performance. It doesn’t matter a whole lot to the Red Sox that the Blue Jays finished 73-89 last season. They’re far more interested in how Farrell performed in the context of that team. To objectively measure a manager’s value, it can be useful to separate his personal performance from that of his players.

Two such methods are discussed by Adam Darowski in his post on Beyond The Box Score. Both attempt to show “Wins Above Expectancy”, or WAE. The first compares a team’s actual win-loss record to their Pythagorean record, or WAE/pyth. The Pythagorean calculation was developed by Bill James to determine a team’s expected win-loss based on runs scored and runs allowed. Most teams end up within 1-2 wins of their Pythagorean record. The theory on how this relates to a manager performance is rooted in the idea that teams that overachieve compared to their run differential do “little things” or “intangibles” at a higher rate. These teams do more with less, Here, we are attributing any difference between the actual win-loss and the Pythagorean win-loss to the manager. It’s not unreasonable to think that some of the in-game impact that managers have may show up in this calculation. According to the Pythagorean method, the 2011 Blue Jays should have won 79.2 games. In reality, they finished .500 going 81-81. So for 2011, John Farrell gets 1.8 WAE/pyth. In 2012, the Jays were expected to win 74.3 games by this model, but instead won only 73 games, putting Farrell at -1.3 WAE/pyth for the year, and .5 WAE/pyth over his short career. But there are certainly limitations to this approach. Attributing all, or even some of, the discrepancy between expected and actual WAR to the manager is some quite a leap. An even bigger assumption, which is frankly untrue, is that the manager is not influencing the amount of runs scored or allowed. Obviously managers are expected to improve the ability of hitters to score runs and of pitchers to prevent runs, through both instruction and strategy.

Our second method for determining managerial value goes a little bit farther than the Pythagorean record. This calculation will again give us a form of Wins Above Expectancy, but uses WAR to develop the expected total of wins. WAE/war is constructed in a very different way from its Pythagorean counterpart. It starts by assuming a replacement level team (An entire roster of replacement level players) would finish with a record of 52-110. This replacement level team was originally modeled from expansion teams but has been linked to a more precise mathematical formula. The individual WAR values for the players will contribute toward making the team better than this level. So if all of the players’ WAR values on a given team are summed, and added to the base line of 52 wins, it should give a pretty clear indication of how the team finished. But there is normally some variation, and this variation will is the WAE/war. So in 2011, Toronto players collectively produced a 30.4 WAR; from which we would expect them to win 82.4 games. They really only won 81 games, so we would attribute a -1.4 WAE/war to Farrell for the year. In 2012, they would produce only 22.4 WAR, meaning we would expect them to have won 74.8 games, when in reality they won only 73, leaving Farrell at -1.8 WAE/war for the year and -3.2 WAE/war for his career. Here’s all of the results put into a table (Darowski has related data for any manager you’d be interested in).

Year	Actual Wins	ExpW/     pyth	WAE/     pyth	ExpW/      war	WAE/   war
2011	81	79.2	1.8	82.4	-1.4
2012	73	74.3	-1.3	74.8	-1.8
Total (2)	154	153.5	0.5	157.2	-3.2
Avg.	77	76.75	0.25	78.6	-1.6

This measurement really gets at how a manager utilizes his players’ production. While WAR is modeled to reflect players’ contribution to runs, it does not fully correlate. This variation could at least partially be explained by how a manager behaves relative to all of the events that place on the field over the course of a season. But again, the value of this stat is severely limited by the fact that it doesn’t consider the manager’s ability to impact his player’s performance, as if they all prayed in a vacuum. But in general, “good” managers do well in these metrics, while “bad” managers do poorly. So while there isn’t much data to judge Farrell by here, we can be confident that he’s not some drastic overachiever, guiding terrible Blue Jay rosters to mediocrity. At best, he’s been a mixed bag, taking average to below average Toronto teams to finishes maybe just a touch under what we might expect.

As you can see, it’s very difficult to value managerial performance, especially relative to player production. But the Red Sox attempted to do just that. If you accept the data above, it seems as though Boston made a mistake. If you don’t except the data, you probably feel that selecting a manager is a gut decision. With that philosophy, any candidate they hired would have been a judgment call. But getting Farrell was the only move that had a quantifiable cost in Mike Aviles’s production. For this move to be reasonable, the Red Sox must see something very special in Farrell. They really raised the stakes on an essentially gut decision.