David Robertson and the Dangers of Reliever Volatility

Right-handed reliever David Robertson is earning $12 million per year on a rebuilding Chicago White Sox team that has little need for a high-priced closer. The Washington Nationals, meanwhile, might need a closer if they aren’t comfortable with internal options who, whatever their qualifications, lack proven closer experience. As a result, it isn’t surprising to find that the two teams have been discussing a trade. Robertson is owed $25 million over the next two years, a relatively reasonable fee given the cost of closers on the free-agent market. If the White Sox are looking to dump salary, Robertson might make sense for multiple teams, but if the Sox want prospects back, both Chicago and Robertson’s suitors might be better off waiting until July, even if the price for relievers is higher at that time.

From 2011 to -15, Robertson was one of the very best relievers in baseball. During that time, he averaged nearly two wins above replacement per season. The only relievers with a higher total WAR during that time frame were Aroldis Chapman, Greg Holland, Kenley Jansen, and Craig Kimbrel. That 2015 campaign, Robertson’s first with the White Sox, was also arguably the best of his career. He struck out 34% of batters while walking just 5%. A very low 66% left-on-base percentage gave him just a 3.41 ERA (compared to his 2.52 FIP), but the results were fine nonetheless. Entering the 2016 season, Robertson was again set to be one of the very best relievers in the game, earning a 1.9-WAR projection on our Depth Charts projection. The season didn’t go as well as expected.

Robertson put together a solid season, recording a good 3.58 FIP (82 FIP-) and a similar 3.47 ERA (82 ERA-). The result: a 1.0-WAR season, making him one of just a dozen full-time closers to hit the one-win mark last year. The results were good, but they represented a decline from his elite numbers the five years prior to 2016. His strikeout rate dropped from 34% to 28%; his walk rate more than doubled, up to 12%, after having remained below 9% since the 2011 season. Last season might be an outlier. It’s possible that Robertson return to form this year. It could be a new normal for Robertson going forward, though — or, worse, it could represent a decline that could continue into this season. The problem is that nobody really knows.

The projections peg Robertson at right around last season’s level. Steamer gives him a 3.49 FIP. Both ZiPS and Steamer forecast him to produce about a win’s worth of value. One win would represent a solid season. For a team in need of a relief pitcher and expecting to contend, the value of that single win is likely quite high, probably pretty close to Robertson’s salary. While projections are a very good way to estimate a player’s performance in the coming year, I would be very wary of relying on them when considering how a reliever might perform up until the trade deadline. The volatility of reliever performance year to year is significant.

Part of the problem with projecting relievers involves small samples: a few home runs here or there, or a short poor stretch, can greatly influence end-of-season numbers. Another problem is the frequency with which pitchers get hurt, especially pitchers who throw at higher velocities, which is often true for relievers. In combination, we end up with a fairly high bust rate for that population.

Last season, there were 65 relievers who’d earned a projection of at least 0.6 WAR by the very end of March, when we did our Positional Power Rankings for relievers. These pitchers account for roughly the top half of relievers, so we can classify these as the “good.” On average, these relievers were projected for 1.0 WAR. On the whole, the average performance in 2016 was considerably less: just 0.7 WAR. The problem doesn’t necessarily lie with the projections, which assume some playing time, but with the fact that so many relievers end up providing little to no value. Of those 65 relievers, 28 (43%) ended up with 0.3 WAR or less.

This problem is quite relevant to David Robertson’s projection for 2017. Elite relievers last season tended to reach their projections; merely “good” relievers had more trouble. The chart below separates the relievers last season by those who were projected for at least 1.4 WAR (13 pitchers) and those who were projected for a WAR between 0.7 and 1.3 (42 pitchers).

Reliever Bust Rate in 2016
Projected WAR AVG Proj WAR AVG 2016 WAR Bust Rate
0.7 to 1.3 0.9 0.5 50%
1.4 or greater 1.7 1.8 15%
Bust is classified as 0.3 WAR or less

The pitchers who were supposed to perform at a very high level last season did so. Only two pitchers projected for at least 1.4 WAR performed poorly: Jake McGee and Trevor Rosenthal had disappointing seasons. Among the other 11 pitchers, though, Robertson’s 1.0 WAR actually represented the lowest mark.

Meanwhile, among the 42 pitchers projected for merely “good” seasons, 21 were nearly worthless. Now, let’s compare that to the group of pitchers who weren’t projected to be all that good in 2016. We are obviously dealing with a larger sample of pitchers, but quite a few in that group turned in very good seasons.

Where Did Good Relievers Come From in 2016
Projected WAR 1.0+ WAR season 0.6+ WAR seasons
less than 0.6 24 43
0.6 or greater 21 32

There are a many more relievers projected to record less than 0.6 WAR than to record more than that figure. But the sheer number of good reliever seasons from unexpected places helps illustrate the uncertainty regarding reliever performance. Given that uncertainty, it might be best to see what a team has before attempting to acquire a merely good reliever at a high cost. The trades for Aroldis Chapman and Andrew Miller last season speak to the high cost of a relief ace at the deadline, but the cost was still high for pitchers a rung below — pitchers like Mark Melancon and Will Smith.

That high cost at the trading deadline might make it seem like trading now for a relief pitcher is a good idea. However, the high rate of busts from relievers makes any sort of short-term move for a reliever very risky. The price might be higher at the deadline, but a team also has the advantage of not investing in a player who hasn’t shown a considerable capacity for good performance in the season at hand. Reliever performance is always volatile, but of the top-50 relievers in the first half last season, more than half produced at least 0.4 WAR in the second half; more than 75% had an above-average FIP and an even higher percentage sported above-average ERAs. It’s probably better to pay a higher price at the deadline when you have a greater certainty regarding both the player for whom you’re trading and a better understanding of your team’s needs and position in the playoff race.





Craig Edwards can be found on twitter @craigjedwards.

6 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Jorge Fabregasmember
7 years ago

I found it interesting that Robertson had the same WARP, basically, in 2016 as 2015. He went from a good framer (Flowers) to a dumpster fire. Could that really have made that much of a difference in the rate stats? I wonder if BP’s fancy pitching metrics smooth out pitcher volatility at all.

trenkes
7 years ago
Reply to  Jorge Fabregas

Agreed – it would be interesting to see what portion of Robertson’s volatility was actually catcher volatility. My sense is that he lost the bottom half of the strike zone due to to navarro.