The Chacon Zone: Using the Splits Leaderboard to Identify Closers-in-Waiting
Projecting future closers is always difficult. We can use a number of different frameworks that factor in environment, talent, pitch quality, and arsenals, and still scratch our heads marveling at how relievers are used. It’s a tricky proposition given the number of variables involved. Add to that the changing nature of bullpen roles, it’s not inconceivable, as we saw with Andrew Miller’s usage, that a progressive manager might not use his best reliever in a way that’s conducive to racking up saves.
In fantasy, saves are expensive and the inherent volatility of bullpens can make chasing them on draft day a dubious endeavor. The Chacon Zone’s goal is to identify non-closing relief aces. Those pitchers whose contributions in strikeouts, ERA, and WHIP, despite low innings totals, are significant enough to offset the lack of saves that you’d receive by rostering a closer in his place. By banking on talent, rather than simply opportunity, we can identify cheap relievers not only possessing high floors but also high ceilings should they be thrust into a ninth inning role. Think Edwin Diaz from last year. Luckily for us, the new Splits Leaderboard, provides yet another tool by which we can (attempt to) identify these pitchers. By isolating performance in high leverage situations, we can not only identify talented relievers but those whose managers entrust them in the most pivotal moments.
Below you’ll find a leaderboard of the top 34 non-closing relievers ranked according to performance in high leverage situations. In order to qualify for the list, a pitcher had to be a reliever last year and tally fewer than 8 saves, the average among qualified bullpen pitchers in 2016. By default, the list is sorted according to a composite Z-Score aggregated from HR/9, K-BB%, xFIP, GB%, Hard%. You can also sort the leaderboard by any of those individual measures as well.
| Name | Team | HR/9 | K-BB% | x FIP | GB% | Hard% | Z-Score |
|---|---|---|---|---|---|---|---|
| Brad Hand | SDP | 0.00 | 30.43% | 2.39 | 46.00% | 24.00% | 8.30 |
| Blake Treinen | WSN | 0.00 | 8.70% | 3.31 | 82.14% | 7.14% | 6.42 |
| Joseph Biagini | TOR | 0.00 | 15.15% | 3.85 | 61.90% | 13.64% | 4.55 |
| Taylor Rogers | MIN | 0.00 | 30.00% | 2.11 | 68.42% | 21.05% | 4.07 |
| Bryan Shaw | CLE | 1.04 | 17.19% | 3.43 | 54.05% | 12.82% | 3.34 |
| Hector Neris | PHI | 0.46 | 17.95% | 3.37 | 53.66% | 26.83% | 2.95 |
| David Phelps | MIA | 0.00 | 14.86% | 3.59 | 46.34% | 21.43% | 2.83 |
| Mychal Givens | BAL | 0.00 | 18.87% | 3.49 | 32.14% | 13.79% | 2.80 |
| Justin Wilson | DET | 1.26 | 19.40% | 2.82 | 64.10% | 25.58% | 2.59 |
| Juan Nicasio | PIT | 1.08 | 37.04% | 2.07 | 53.33% | 26.67% | 2.58 |
| Scott Feldman | 2 Tms | 0.00 | 25.00% | 2.06 | 55.56% | 26.32% | 2.55 |
| Dan Otero | CLE | 0.00 | 20.83% | 3.62 | 46.67% | 6.67% | 2.45 |
| Jake Diekman | TEX | 0.68 | 20.83% | 3.10 | 57.69% | 30.77% | 2.09 |
| Christopher Devenski | HOU | 0.00 | 39.13% | 0.73 | 33.33% | 33.33% | 2.07 |
| Ian Krol | ATL | 0.00 | 11.11% | 3.26 | 63.16% | 15.79% | 1.75 |
| Tony Barnette | TEX | 0.00 | 10.87% | 4.17 | 51.72% | 20.00% | 1.71 |
| Justin Grimm | CHC | 0.00 | 22.22% | 3.45 | 38.46% | 15.38% | 1.54 |
| Nate Jones | CHW | 0.77 | 24.73% | 3.06 | 39.22% | 32.69% | 1.52 |
| Raisel Iglesias* | CIN | 0.00 | 13.64% | 4.69 | 46.34% | 20.93% | 1.42 |
| Kyle Barraclough | MIA | 0.49 | 14.29% | 3.73 | 52.63% | 29.27% | 1.32 |
| Hunter Strickland | SFG | 0.52 | 21.21% | 3.63 | 42.50% | 31.71% | 0.91 |
| Nick Vincent | SEA | 1.32 | 28.00% | 3.32 | 32.26% | 21.88% | 0.88 |
| Mark Lowe | DET | 2.70 | 26.92% | 2.30 | 57.14% | 21.43% | 0.80 |
| Vance Worley | BAL | 0.00 | 13.33% | 3.15 | 70.00% | 30.00% | 0.72 |
| Kyle Ryan | DET | 0.00 | 5.88% | 3.95 | 55.56% | 10.00% | 0.71 |
| Chien-Ming Wang | KCR | 0.00 | 10.53% | 3.55 | 64.29% | 33.33% | 0.68 |
| Jhan Marinez | MIL | 0.00 | 23.08% | 3.51 | 42.86% | 28.57% | 0.56 |
| Drew Storen | 2 Tms | 0.00 | 22.50% | 4.18 | 40.00% | 34.62% | 0.42 |
| Neftali Feliz | PIT | 0.56 | 15.25% | 3.44 | 43.24% | 31.58% | 0.38 |
| J.P. Howell | LAD | 0.00 | 15.79% | 4.08 | 50.00% | 35.71% | 0.24 |
| Ryan Dull | OAK | 0.87 | 17.50% | 4.02 | 37.04% | 22.22% | 0.11 |
| Josh Smith | CIN | 0.00 | 8.33% | 4.85 | 44.44% | 22.22% | 0.09 |
| Antonio Bastardo | 2 Tms | 1.29 | 23.08% | 3.43 | 33.33% | 25.00% | 0.04 |
| Matt Bush | TEX | 0.68 | 9.80% | 4.35 | 51.35% | 25.64% | 0.02 |
Two more housekeeping items before we move onto some of the names. In case you’re wondering why I used a cutoff of the top 34, only 34 of the 99 qualified pitchers boasted positive Z-Scores. Secondly, I included innings pitched as a multiplier in the calculations so relievers who pitched well in more high leverage situations were rewarded with higher Z-Scores. Those who bombed in those same situations were commensurately penalized by their poor performance.
Right off the bat, we find relievers from two teams wading through murky closer situations. We like to see that. Brandon Maurer finished the year as the Padres’ closer and while he might be the favorite to start the season where he left off, he’s far from a lock. He certainly pitched better towards the end of the year but he also blew 6 of his 19 opportunities on his way to a 4.52 ERA. Meanwhile, Brad Hand struck out over 30% of the batters he faced while posting a ground ball rate approaching 50%.
Hand’s biggest impediment to his promotion might be his southern paw. Still, his numbers against righties are at worst comparable to Maurer’s in terms of both ERA indicators and batted ball data. Ryan Buchter, also a lefty, appeared to be first in line to take over closing duties last season, however he too was leapfrogged for the role. And don’t forget Chacon Zone darling, Carter Capps, who figures to be ready by Opening Day. There’s certainly stiff competition adding uncomfortable levels of ambiguity to this situation but I’m not certain that the status quo is the best option to close. Brad Hand certainly deserves a closer look by the Padres and fantasy owners alike.
With Mark Melancon leaving Washington to sign a monster deal with the Giants, the Nationals find themselves in need of a closer. Prior to acquiring Melancon mid-season, the Nats transitioned from eternal man-child, Jonathan Papelbon, to Shawn Kelley, neither of whom pitched nearly well enough to hold down the gig. In fact, you’ll notice by his absence from the list above that Kelley failed to provide positive value in high leverage situations and, among closers, Papelbon was arguably worse, posting a 5.14 xFIP in those same situations.
Blake Treinen on the other hand, was masterful at managing contact. His 82% ground ball rate, led all relievers but one, who happens to be pretty good himself. He also led all relievers in soft%. So, by limiting fly balls and hard contact, Treinen may have pitched far better at critical moments than even his 3.31 xFIP would suggest.
Overall Treinen struck out 24% of the batters he faced, the same as Melancon, but that’s not reflected in the table above. Given how frequently he pitched in high leverage moments, often with runners on base, it’s likely Treinen leaned on his soft contact and ground balls skills to induce inning-ending double plays, rather than going for the strikeout.
Still, Treinen struggles with the free pass particularly in low leverage situations. While slightly better than average in attacking the zone, his first-pitch strike percentage is too low. Given his strong swinging strike rates and low zone-swing percentage, Treinen could afford to be more aggressive with his first pitch. If he does, he could make for a sneaky source of saves in 2017 though we’ll have to wait out the rest of the off-season to see what the Nationals do to address their pen.
I really like Taylor Rogers for his ground ball rate and control but he profiles more as a LOOGY and doesn’t miss nearly enough bats. At least right now, Brandon Kintzler is clearly the play in Minnesota.
Hector Neris just keeps popping up on Chacon Zone lists. I first fell for his whiff and K-BB% rates back in April, awaiting the imminent demise of Jeanmar Gomez. He rewarded me in early May with a save but it took until the last game of the season to tally a second. Neris could go into the year as Philly’s closer but even if he doesn’t, snag him anyway. It’s just a matter of time as he was utterly dominant in 2016, particularly when it mattered most.
In one of my Ottoneu leagues, I’m starting the year with a pretty strong starting staff but I’m dangerously thin at relief. So unlike most fantasy players, I’m holding out hope that Don Mattingly leaves David Phelps in the pen. While great as a starter, Phelps was downright filthy in relief striking out 32% of the batters he faced. Now, he also benefited from an alarmingly high strand rate but then again so did a number of other elite relievers who maintain said rates through elevated strikeout totals. The Marlins look to have a deep bullpen headlined by A.J. Ramos, though he limped to the finish line last season, and everyone’s favorite sleeper relief ace, Kyle Barraclough. But if used in relief, David Phelps’ performance in high leverage situations could bode well for future save opportunities.
Other names I like on this list include Nate Jones, Ian Krol, the aforementioned Kyle Barraclough, and assuming he doesn’t make the Astros’ rotation, Chris Devenski. Regardless of save opportunities, all of these relievers are roster-worthy in deep leagues and even in some standard ones where over-zealous hoarders price you out of the closer market.
Rylan writes for Fangraphs and The Hardball Times. Look for his weekly Deep League Waiver Wire and The Chacon Zone columns this season.


Hi Rylan, Very interesting article. But as someone reasonably new to FanGraphs, could point me to where I can learn about Z-Scores? Been through the site and the glossary, but if it is there, I’ve missed it. Many thanks, Dave, sopacdave@aol.com
It’s just a sum of the standard deviations above or below the the mean. For example Brad Hand probably scored at least a 3 z-score in the HR/9 column because that’s how many standard deviations 0 HR/9 is from the mean. Add all columns to get a final z-score. This is how most fantasy values are calculated once projections are determined.
Thanks, David. Basically a Z-score is just the # of standard deviations in a specific stat better or worse than the mean. Because they use common units (standard deviations), you can add them across multiple stat categories to end up at an aggregate value. It gets a little tricky with rate stats when you have to account for innings or plate appearances but Zach Sanders covered it well back in a 3-pt series in 2011.
http://www.fangraphs.com/fantasy/value-above-replacement-part-one/
How are you defining High Leverage? Blake Treinen gave up a walk off bomb to Javier Baez in the bottom of the 13th inning on May 8th. Would have not expected a 0% HR/9 rate. He also still gets roughed up by lefties too much. Thanks for posting.
Any Leverage Index above 2.0 is considered high leverage. You made an awesome observation. Treinen’s LI that game…1.99.
http://www.fangraphs.com/wins.aspx?date=2016-05-08&team=Nationals&dh=0
If we lowered the threshold by .01, that’d absolutely impact his rating.
Also, here’s how it’s calculated. http://www.hardballtimes.com/crucial-situations/
I appreciate the effort to identify valuable relief pitchers in an environment in which projecting saves is problematic, but I think your methodology may need changes.
Take Devensky as an example. Last year, he pitched 108.1 innings total; of that, only 6.2 innings were high leverage.That raises two concerns. One, the statistical assertions you make are based on a sample size far too small to make any findings of significance about performance. As small as that number is, the real high leverage inning as a reliever is even smaller: Devensky pitched 24.2 innings as a starter and one inning total was in high leverage situations, so his total number of high leverage innings as a reliever was actually only 5.2 innings.
Fangraphs actually warns against extrapolating conclusions and making projections using leverage data: (http://www.fangraphs.com/library/misc/li/)
“Performance in high or low leverage moments isn’t a repeatable skill”
“it is important to remember that performance based on leverage is not a particularly repeatable or predictive skill. A .400 wOBA in high leverage situations one year tells you very little about the player that you couldn’t glean from their overall stat line.”
The only significant conclusion you should make (but don’t) about the tiny amount of high leverage innings that Devenski pitched is that the Astros have MANY relievers entrusted ahead of Devensky to pitch in high leverage at-bats, so his chances of picking up saves is remote. Among Houston relievers, Giles, Gregerson, Harris and Feliz all pitched more high leverage innings. In other words, you have ignored the one critical and helpful piece of information about understanding the pecking order in the Astros bullpen.
The better methodology would be to target high-performance arms based upon overall stats, regardless of leverage, then use the number of high leverage innings to establish the pecking order in the bullpen, which may be helpful in projection who might close for teams whose managers use their best relief pitcher in save situations. It appears you are so excited about using a new tool (the splits leader board) that you have attempted to use it for a task for which it is especially ill-suited.
This was kind of harsh but… pretty much spot on. It is best to target the best arms overall because they are most likely to be entrusted with the most “high Leverage” inning eventually. This is especially true in my dynasty points league because K’s are worth a lot and we award 8 points for Holds as well (10 for Saves).
You’re welcome for the article! I’m not sure where I ever said that success in high leverage situations is a repeatable skill and as such, small sample sizes aren’t really the point. The leaderboard is meant as a starting point by which we can identify players whose managers entrusted them in high leverage situations and who, by pitching well in those situations, did nothing to lose that confidence. I didn’t mean it to be a standalone predictive tool as to who’d be named closer. Otherwise, I wouldn’t have ignored Bryan Shaw (which I did due to lack of opportunity) or Scott Feldman (which I did because talent). Clearly, overall performance is much more valuable because of the body of work and you’ll notice that I only called out players whose overall body of work justified it.
As for Devenski specifically, you bring up a fair point about sample size implying pecking order but he’s also a unique example given his multi-inning usage. He was actually the most valuable reliever the Astros had last year by WAR and most effective by FIP. While he may not have pitched the same number of high leverage innings, that the Astros pen used three closers last season suggest it’s a murky situation and, as I’ve written in the past (http://www.fangraphs.com/fantasy/the-chacon-zone-and-a-closer-look-at-relief-pitchers-in-2016/), I really like Devenski’s talent as a reliever.
Thanks for your well-thought out comment though. I just think you may have grasped onto something that I didn’t intend to imply.
Thank you for your thoughtful response. I do appreciate you didn’t explicitly state that performance in high leverage situations is predictive (which it is not) but you do seem to suggest that managers view that performance as predictive in a way that is independent from how they pitch overall:
“By banking on talent, rather than simply opportunity, we can identify cheap relievers not only possessing high floors but also high ceilings should they be thrust into a ninth inning role. Think Edwin Diaz from last year. Luckily for us, the new Splits Leaderboard, provides yet another tool by which we can (attempt to) identify these pitchers. By isolating performance in high leverage situations, we can not only identify talented relievers but those whose managers entrust them in the most pivotal moments.”
It is possible that specific managers are swayed by small sample high leverage situations, but if we are to assume that most managers have a better grasp of statistics and the fact that high leverage performance is not a repeatable skill, than those managers will be evaluating overall performance and its more robust sample sizes, something that is readily available without the splits leaders board.
To the extent my assumption about managers is correct, it is the overall performance of relievers that is the best predictor of talent level and the truest indicator of which relievers have a better chance of closing given equivalent opportunities. Overall performance is also clearly more useful in selecting non-closing relievers whose stats will enhance your fantasy team’s performance.
None of this means Devenski is not a talented relief pitcher, but rather, that Hinch will select bullpen roles based upon overall performance, not high leverage performance.
And none of this means there may be some managers swayed by high-leverage performance despite the lack of statistical significance; if a manager has a history of such choices, than knowing high-leverage performance on those teams may be of some use.
Beyond overall performance and talent level among relief pitchers, I would think the next most important factor on some teams is contract status. Some teams will show a bias in using as a closer a relief pitcher who is already locked into a contract and whose cost won’t be elevated by virtue of getting saves. So we sometime see more talented relief pitchers in their arbitration years be placed in the back seat behind somewhat less talented and older relief pitchers whose contractual terms won’t change by accumulating saves.
Thanks–very interesting! How high would you go on phelps in ottoneu pts?
So, I’m brand new to Ottoneu points leagues so you might be better off directing your question to Justin, Joe, Trey or some of the other Ottoneu guys. But assuming he pitches the whole season in the pen, I like along the same lines as Barraclough, Hunter Strickland, and Nate Jones, putting him in the $12-15 range.
Great article, should be very helpful.
However, I think something very important is missing here: arbitration status.
Neris didn’t get more saves not just because Jeanmar Gomez stayed decent enough, but because Neris is in their long term plans and they don’t want him to become expensive. That logic will apply to quite a few players on this list (Devenski, Givens etc). They are unlikely to get saves because of how much it will cost their team vs. other, less skilled veterans who are past their arb days.
This won’t always be true for a team making a big push, but it will be a factor for all the others. Note that Philly did this with Neris and they are a large market team.