Expected run differential per game: A primer for Blue Jays fans

Expected run differential per game? Keep reading to discover an interesting new statistic, and how it relates to your Toronto Blue Jays



Embed from Getty Images



Regular readers of Jays from the Couch might be familiar with a statistic I introduced earlier this season, expected run differential per game (xRD/G). I’ve long intended to give readers a more thorough explanation of how it works and why it is a useful stat for baseball fans. For a more technical explanation, I recommend checking out my FanGraphs Community post on the metric. For a less technical, more Jays-focused explanation, read on.


xRD/G tries to capture all of the things that a baseball team does on the field to win games. Here is the equation that I currently use:

xRD/G = 23.31*(Batting xwOBA) – 2.52*(BsR/600) + 8.34*(Batting xwOBA)*(BsR/600) – 13.16*(SP xwOBA) – 10.19*(RP xwOBA) + 0.004*(DRS/150)


Batting xwOBA captures how well a team’s batters create good contact, avoid strikeouts and accumulate walks. Starting pitcher xwOBA captures how well a team’s rotation prevents good contact, accumulates strikeouts and avoids walks. Relief pitcher xwOBA does the same, but for a team’s bullpen. Base running runs per 600 PA (BsR/600) captures how well a team does at stealing bases, avoiding double plays and making good decisions on the base paths. Defensive runs saved per 150 games (DRS/150) captures how well a team’s defence executes…well, defensively.


In my mind, xRD/G can provide useful context to a team’s performance. A team with a great record, but a negative xRD/G might be due for negative regression as the season wears on (like the Diamondbacks). A team with a less-than-great record, but a strong xRD/G could finish the season a lot better than it started (very much the Dodgers).


Before I dig further into the insights that xRD/G can provide us this season (and would have provided for us in recent seasons), let me lay out the case in favour of xRD/G’s usefulness as a predictive tool. Keep in mind that, as we are using Statcast data, I have to limit my focus to the 2015-17 seasons.


xRD/G predicts future performance better than a team’s record or run differential per game

A team’s record (Win%) in one season explains only 19% of that team’s record the next year, while a team’s run differential per game (RD/G) in one season explains 21% of its next-season record. On the other hand, a team’s xRD/G explains 37% of its record a year later.


xRD/G is even more predictive of a team’s future RD/G than RD/G itself. A team’s RD/G one season explains only 20% of its RD/G the next year. xRD/G explains 36% of a team’s RD/G a year later.


The same is true if we focus on half-to-half predictiveness within seasons. A team’s Win% at the all-star break explains only 27% of its post all-star break record, while a team’s first-half RD/G explains 34%. However, a team’s xRD/G in mid-July explains 39% of a team’s second-half record. First-half xRD/G is also a better predictor of a team’s second-half RD/G (explaining 33% of it) than is first-half RD/G (which explains only 24% of it).


xRD/G might be as good (or better) a predictor of future performance than FanGraphs’ team projections

In the age of fancy stats, comparing xRD/G’s predictiveness to basic stats like Win% or RD/G is setting the bar extremely low. Impressively, xRD/G appears to be a match even for an advanced tool like FanGraphs team projections. These projections are currently available for any date since the beginning of the 2016 season.


On the one hand, FanGraphs has the edge in terms of preseason projections. Its projections at the beginning of the 2016 seasons explained a whopping 54% of teams’ records that season. For 2017, the projections explained a more modest 35% of teams’ records. Overall, FanGraphs’ preseason projections over the last two seasons explained 43% of teams’ records.


In contrast, a team’s 2015 xRD/G explained 40% of its 2016 record, while a team’s 2016 xRD/G explained 34% of its 2017 record. Overall, a team’s previous-season xRD/G explained 37% of its record, falling a little short of the bar set by FanGraphs’ projections.


However, xRD/G seems to hold the edge in terms of mid-season prognostication, at least in our small sample of two seasons. At the 2016 all-star break, FanGraphs’ rest-of-season projections explained 34% of a team’s second-half record. On the other hand, a team’s first-half xRD/G explained a meatier 38% of its second-half record. Similarly, while FanGraphs’ rest-of-season projections at the 2017 all-star break explained 45% of a team’s second-half record, a team’s first-half xRD/G explained 52% of it.


The idea that FanGraphs’ projections are more useful than xRD/G before the season starts, but less useful at the all-star break makes a lot of sense. The pre-season FanGraphs projections account for all of the transactions that a team has made during the off-season, while a team’s previous-season xRD/G is (literally) stuck in the past, limiting how well it can predict the future. This isn’t an issue at the all-star break.


xRD/G is even a useful postseason predictor

Over the last three seasons, xRD/G has accurately predicted more series winners than both RD/G and Win%. That’s made extra impressive by the fact that the team with an edge in Win% almost always also has home-field advantage—the one exception being the Cubs in the 2016 World Series.


xRD/G has been even more accurate over the last two postseasons. Of particular note, xRD/G has matched the postseason accuracy of FanGraphs’ projections, which (like Win%) is able to account for home-field advantage.




xRD/G really liked the 2015 Blue Jays and hated the 2015-16 Rangers

Plenty of pixels have been (completely and totally justifiably) spent discussing how awesome the 2015 Blue Jays were. xRD/G concurs—no team has produced a better half-season xRD/G than the Blue Jays did in the second-half of 2015 (2.2), out of 180 half-seasons since 2015. For the season, the Jays produced a 1.0 xRD/G, the best mark in 2015 and the sixth-best mark overall. The 2016 Jays produced a still-respectable 0.3 xRD/G, tied with the Orioles for eighth-best in baseball that season.


On the other hand, xRD/G did not look kindly at the Rangers. The worst two teams to make a divisional playoff series were the 2015 (0.0) and 2016 (-0.1) Rangers. They’re the worst. The Blue Jays even had the edge in 2017, with their -0.1 xRD/G rating more highly than the Rangers’ -0.2 mark. I mean, really, they’re just the worst.


The 2015 NL Champion Mets were backed by xRD/G

While the 2015 Blue Jays produced the best xRD/G in the majors that season, the 2015 Mets were the NL leaders. That presaged their run to the World Series, something that was less evident when focusing only on the team’s record (fifth-best in the NL) and RD/G (also fifth-best).


xRD/G didn’t see the Royals getting past the LDS, let alone winning it all

The 2015 Royals ranked tenth in the majors in terms of xRD/G and fourth in the AL. In terms of xRD/G, they were underdogs in all three series. They did have the best record in the AL, which brought with it home-field advantage, a useful aid in the postseason (see: the kid with the beard).


When it comes to the 2015 Cubs, xRD/G owned RD/G and Win%

In 2015, the Cubs won the second wild card spot after finishing third in a stacked NL Central. In the Wild Card game against the Pirates, they were clear underdogs—playing away from home against a team with a superior record and RD/G. They did have a slight edge in terms of xRD/G. They beat the Pirates.


Then, they faced the Cardinals, their division’s champion, in the NLDS. They were even bigger ‘dogs in terms of record and RD/G, but again retained an edge in terms of xRD/G. They beat the Cards.


In the NLCS, they faced the Mets. While they lacked home-field advantage, they were now the ones with the edge in record and RD/G. However, they were modestly outmatched in terms of xRD/G. They lost.


On the other hand, xRD/G fancied the 2016 Cubs above all rivals

The Cubs went into each of their three 2016 series with an edge in terms of xRD/G (and RD/G and record). They ran the table and won the World Series. If I recall correctly, it had been a while.


The Dodgers have the best cumulative xRD/G in the Statcast era

The Dodgers are really due for a World Series. No team has a better 2015-17 xRD/G than they do, with their 1.1 mark leading the majors. Each of the last three seasons, they finished with a Top 3 xRD/G, rising from third in 2015 to second in 2016 to first in 2017. Each year, they’ve progressed further and further into the playoffs, getting dumped in the NLDS in 2015, the NLCS in 2016 and the World Series in 2017 (game seven, no less).


In 2015, only one NL team was superior to the Dodgers in terms of xRD/G. That would be the Mets, their NLDS opponents/eliminators. In 2016, only one team in the entire major leagues had a better xRD/G. Obviously, that would be the Cubs, their NLCS opponents/eliminators. Finally, in 2017, they were #1. That quality took them to within one win of winning it all, but it was not to be.


The Red Sox lead the way in 2018; the Blue Jays are better than their record

Two and a half months into 2018, we can see that the usual suspects are leading the xRD/G leaderboard. At the very top are the Red Sox, owners of the second-best Win% (after the Yankees) and second-best RD/G (after the Astros). Once again, the Dodgers lead the way in the NL, comfortably ahead of the Nationals.



xRD/G effectively captures the difference between the AL and NL this season. In the AL, there are four super-teams, way ahead of the next-best teams. In the NL, there is a great deal of parity. After the Dodgers and Nationals, there are six slightly above-average teams with only a tenth of a run separating top from bottom. The NL East looks impossibly competitive.


Finally, our Toronto Blue Jays. All season, xRD/G has been higher on the team than either Win% or RD/G. In particular, the team’s batting xwOBA is more positive than the team’s runs scored per game, while the bullpen’s xwOBA is better than its ERA. That’s not going to put them in the playoffs or anything, but it’s worth pointing out as evidence that the Jays may have performed more like the average team alluded to by their xRD/G, than the comfortably below-average team alluded to by their Win% and RD/G.


Most importantly, for the fourth straight season, they are comfortably ahead of the Rangers.


Image result for jose bat flip gif





*Featured Image Courtesy Of DaveMe Images. Prints Available For Purchase.






I’m an economics professor in the GTA whose lifelong love for the Jays was reignited by that magical August of 2015 and the amazing moments since.

Jeff Quattrociocchi

I'm an economics professor in the GTA whose lifelong love for the Jays was reignited by that magical August of 2015 and the amazing moments since.