The Rays’ Unique Ability To Mitigate Risk
I’ll be the first to admit that I don’t understand the Tampa Bay Rays. I don’t understand how they win as many games as they do. They’re definitely good, but it never feels like they should be as good as they are, or recently have been. But at a certain point, if they are consistently better than expected, I’m the one in the wrong and it’s on me to try to understand.
The offense I get. The current American League leader in runs scored is a little over their skis, as they hold that lead despite ranking sixth in on-base percentage and fourth in slugging (they’re fourth in team wRC+ and fifth in OPS). That’s mostly due to the fact that as a team they have greatly improved results with runners on base compared to when they’re empty. That’s more likely luck-related than some kind of mysterious clutch skill they possess, but what the Rays do have is monstrous depth. Nobody in the lineup is going to garner MVP consideration, but their ability to almost never throw out a lineup with dead innings is unmatched in baseball thanks to a roster filled with average or better players. Just look at the Rays compared to the rest of their AL East competition:
| Team | Players |
|---|---|
| Tampa Bay Rays | 11 |
| Boston Red Sox | 8 |
| Toronto Blue Jays | 7 |
| New York Yankees | 6 |
| Baltimore Orioles | 6 |
More often than not, the Rays lineup is nine deep, while also being filled with guys who can leave the yard; the Jays are the only American League team with more home runs this season. Easy enough.
What confuses me is the pitching. The team ranks first in the league in FIP and second in ERA, but they’re basically doing it without a rotation at this point. Only four Rays pitchers have exceeded 100 innings on the year and that number is unlikely to grow by the time the regular season comes to a close. What’s more, of those four “bulk” guys, only one has really been good:
| Pitchers | Innings | ERA |
|---|---|---|
| Ryan Yarbrough | 136.0 | 4.90 |
| Shane McClanahan | 110.1 | 3.59 |
| Josh Fleming | 100.2 | 5.01 |
| Michael Wacha | 100.2 | 5.54 |
The success, as everyone knows, comes from a small town’s worth of bullpen arms who deserve some kind of group MVP award. “Eleven to 12 relievers on the IL plus another 10 in their current ‘pen kind of tells you what you need to know,” an American League staff member who has advanced the Rays in the past told me. “And then they are getting anywhere from useful to exceptional overall contribution from them.”
That brings me not to the games the Rays are playing this month, but next. Potential playoff teams spend the season planning their postseason rotation. The Dodgers will roll out Walker Buehler, Max Scherzer, Julio Urías and hopefully a healthy Clayton Kershaw. The Brewers will have a terrifying first three of Corbin Burnes, Brandon Woodruff and Freddy Peralta. White Sox opponents will have to face Lance Lynn, Carlos Rodón and Lucas Giolito. And the Rays? Well, opponents will have to prepare for about six or seven pitchers a night. That approach creates a risk for Tampa Bay that has always concerned me, but as we’ll discuss in a moment, that concern was greatly assuaged after taking a closer look at their roster.
Shane McClanahan is an obvious choice to start a playoff Game One, as he’s the Rays’ most likely starter to get deep into a game and avoid contact. From there, it’s going to a mish-mash of arms. The strategy of needing more than twice the number of pitchers to perform on a given night than a more traditional club would carries an inherent risk that has always made me uncomfortable. Pitchers — and more precisely for this conversation, ‘pen arms — can be quite volatile, with the words “reliever” and “consistency” often feeling like oxymorons. Nearly every reliever is prone to a bad outing here and there. Sometimes it’s bad luck, sometimes they just don’t have it. But it clearly happens. There’s always a chance that the next arm your team puts on the mound is just going to crap the bed. The theory then goes that the more pitchers one uses, the greater the chances that such a crap-the-bed inning will occur, which can be disastrous in the playoff game.
Let’s start with a team like the Yankees, who despite some recent struggles, have one of the better bullpens in baseball. Here are their five most-used relievers during the 2021 season, as well as a CTB% (Crap The Bed Percentage), with a CTB outing defined as one featuring more baserunners allowed than outs recorded and/or an in-game ERA of 18.00 or higher:
| Reliever | G | CTB | CTB% |
|---|---|---|---|
| Aroldis Chapman | 50 | 6 | 12.0% |
| Chad Green | 56 | 6 | 10.7% |
| Jonathan Loaisiga | 54 | 3 | 5.6% |
| Lucas Luetge | 51 | 6 | 11.8% |
| Wandy Peralta | 36 | 6 | 16.7% |
This metric is far from perfect; Aroldis Chapman, for example, had mid-year hiccups that are contributing to a larger than anticipated CTB%. But I still think it’s useful, so let’s assume for a moment that these percentages are mostly accurate. If the Yankees behaved like the Rays in terms of bullpen management and used all five of these relievers in a single game, the odds of them getting non-CTBs from the entire quintet would be roughly 54%, meaning it’s only a little more than a coin flip that none of the five arms, all of them quality, would screw up a close game. The math here shows the risk that any team takes when it uses more than handful of arms in a game by design.
But we’re not just talking about a good playoff-caliber bullpen, we’re talking about the Rays.
And they don’t merely have a deep ‘pen, but an excellent one, making the math change considerably. Here are their five most used relievers, based on those most likely to be available to them come the postseason:
| Reliever | G | CTB | CTB% |
|---|---|---|---|
| Pete Fairbanks | 38 | 3 | 7.9% |
| J.P. Feyereisen | 25 | 2 | 8.0% |
| Andrew Kittredge | 49 | 0 | 0.0% |
| Collin McHugh | 31 | 2 | 6.5% |
| Matt Wisler | 24 | 2 | 8.3% |
While it’s worth digging into the historical data (perhaps for a future piece), the Kittredge number feels remarkable, as he’s avoided disaster the entire season. Indeed, the Rays as a whole avoid disastrous reliever outings better than anyone in the league thanks to great scouting, impressive analytical work, talent-enhancing coaching and development, and a manager who knows exactly when to use each player based on matchups and game situations. These lower percentages lead to much smaller overall risk. While the Yankees would be nearly even odds to get a disastrous relief outing out of their five most-used ‘pen arms, the odds of this Rays group getting through a game en masse without a CTB outing are nearly three-fourths. By developing this pitching staff, along with strategies to maximize performance, the team has positioned themselves to survive postseason games with a ‘pen game strategy at an efficacy rate of nearly 50% greater than most clubs. And while much of that is because of pure-minded baseball thinking, there is also an underrated cultural aspect to it.
“The fact that they have either gotten buy-in or forced it on relievers who they need to be ready anytime and who don’t have more traditional roles is probably a long-run contributor to increased pitching success,” explained the American League staffer. “I think defined bullpen roles in general are a good thing and I’m not sure you could do what Tampa is doing universally across personnel, but maybe the lack of a traditional role can be a role in and of itself.”
There’s certainly still a risk to what the Rays do without traditional starting pitchers, but they’ve done everything in their power to mitigate that risk, and to great effectiveness. I went into this piece intending to show why the Rays can’t get through the playoffs doing what they do in terms of managing their pitching. After looking at the data, I’m not as convinced as I once was. That’s becoming the norm when I try to find something wrong with the Rays and their unconventional ways.
Kevin Goldstein is a National Writer at FanGraphs.
I will start by admitting ignorance of the intricacies of pitcher preparation or the significance of knowing one’s role in aiding success.
But it has always seemed to me-as an outsider-that the essential role of a pitcher is to avoid letting the other team score runs-to get batters to make out. And it also seems to me that the fundamental factors there are the batters faced (their tendencies and weaknesses), the situations (that is, men on base or not) and the skills of the pitcher. I can understand that getting the last outs might add some stress for some pitchers, but given the fact that so many pitchers have gotten final outs over the years, I think that factor has been exaggerated.
If all that is so-or to a large extent so-I don’t know why it matters if the pitcher comes in at 9:00 one day and 10:00 the next or in other words, in the 6th inning and then the 8th inning. His job should be to get ready to pitch when called upon, and then to do his job of getting outs. What is the 8th inning pitcher doing during the 6th and 7th innings that he could not be doing in the 4th and 5th innings? If a manager signals to the bullpen in the 5th that he wants pitcher A to start warming up, how is that different than if he signals it in the 7th inning?
I am sure there is a psychological factor, but am guessing that it has become a self-fulfilling prophesy. Having been told for decades now that pitchers do better with given roles rather than being able to get outs when called upon, they now think it is so. Apparently the Rays have told their pitchers to be ready when called upon and it is working. It is unlikely these are somehow unique people. More likely that the expectations are unique.
There may also be a financial reason in that the “Closer” will lose money in negotiations and arbitration if he gets fewer saves. But surely that can’t be what drives the manager’s thinking.
If I am right, then the best way to use relievers is by evaluating matchups. This batter can’t catch up to high fast balls and pitcher A has a plus plus high fast ball. And in the 5th inning today, with men in scoring position, i want a strikeout which is more likely with pitcher A, who is supposedly my closer, than any other pitcher on my staff. Of course, other factors come in such as how many innings can he go and who will hold our lead over the next 3 or 4 innings, but none of it depends on who is my 5th inning man.
By using a variety of relievers in high leverage spots across game situations and inning, the Rays are building a form of depth designed to mitigate risk.
Businesses (especially IT departments) do this same kind of thing, they cross-train individuals so that one product owner (for example) can step in and take over for a different product owner in the event that person leaves or is otherwise unavailable.
The Rays are essentially training all of their relievers to do each others job, so if one guy goes down, the rest can confidently pick up the slack without having to worry about being thrust into unfamiliar territory.
At its core, the idea of assigned ‘roles’ is less about the role / title and more about being comfortable in the situation you are entering into. Most people like feeling comfortable, it’s partly why we have routines, we know what to expect.
Assigned closers know exactly when they are entering a game and there’s a level of comfort in that. The Rays are simply trying to get as many pitchers ‘comfortable’ in as many situations as possible.
Bullpens with set roles also create more room for interpersonal issues. Roles are tied to status and ego can take precedence over what is best for the team – think the guy who “has to close” based on the pitcher he used to be not who he is now.
The ‘flat’ bullpen structure (one without defined roles) may not be everyone’s cup of tea but it helps to mitigate potential interpersonal issues tied to role while increasing the sense of responsibility for the bullpen as a whole versus feeling like the success of the group is tied to the 2 or 3 guys at the top of the hierarchy.
Excellent points.
I wonder to what degree teams select personality type for this. “Closer mentality” is a phrase that gets thrown around a lot as a determination point on whether a guy can pitch the 9th inning. But as often as not those guys are very prickly pears who if they don’t have a predefined, (big money) roll, make trouble. Thinking of the Jonathan Papelbon type. As good-great of a reliever he was, you do not want him on the 2021 Rays.
I realize there’s a human factor in baseball and that there’s more to the game than just numbers, but the ridiculousness of ultra-rigid bullpen roles never made any sense to me. The idea that your “8th inning” guy would be incapable of getting outs 20 minutes earlier in the 7th inning, or the deliberate holding back of your best reliever for a fairly low leverage clean 9th inning was absurd and I can’t believe it lasted as long as it did.
It’s mostly out of the game now of course, it I can’t believe it took so long for teams to realize the inning a pitcher is used to pitching isn’t relevant
They have the highest O-Swing%, second best BB%, 6th in Hard Hit %, and 5th best HR%. Basically they don’t shoot themselves in the foot.
Great article dude.
I love how the Rays operate and respect the innovation. They’re such a great team to read about but actually watching their games feels kind of soulless.
I don’t know why you’re getting downvoted a bit here. Rays games are indeed pretty soulless and, imo, boring as hell. Baseball games are and will always be crafted around the starting pitchers, and since Tampa removes them from the game, there’s a lot less to attach yourself to when watching.
I remember when I was a kid in the late 70s, and I saw Chuck Tanner move his right fielder into the infield for one batter. Never saw anything like it before, and I was convinced for a while that Tanner was an incredible manager. It never crossed my mind that the Pirates were soulless because they didn’t keep the outfielders out of the infield 100% of the time.
Have no idea where the idea comes from that baseball is built around the starting pitcher — that’s an opinion, not at all a fact.
Come on, man. Baseball games are based around the starting pitcher because he’s the one who holds the ball. He’s the one who stands alone in the middle of the diamond, the one who initiates play by pitching the ball. Lineups get introduced as a unit on a broadcast, the starting pitcher gets his own graphics and spotlight. And, for most of baseball history, a team’s fate in a game was decided by the game-controlling figure that is the starting pitcher.
The starting pitcher and the matchup between the two starting pitchers is easily the biggest narrative element of a baseball game. One of baseball’s biggest problems right now is that starting pitchers are so sterilized as dominant figures they have massive trouble becoming stars. Go look at any list of stars, any montage MLB makes, it’s almost exclusively position players. That’s a huge problem.
Again, you’re simply stating conventional wisdom. Which is, uh, mass opinion? Opinion nonetheless. There is a hitter in every at bat just as there is a pitcher.
Hitters rotate through, the pitcher is the one who remains to face them all. I understand the Fangraphs crowd likely doesn’t consider it, but the death of the starting pitcher is a main factor in baseball being a poor product in 2021.
It’s a huge problem to you, but I haven’t met many people who care how long the first pitcher of the game pitches
Counterpoint: nah
I would not give you a down vote, but I do disagree.
I do understand and respect the feeling many have that the game is being played in front offices rather than on the field, that the athletes are simply pawns being manipulated, moved about by computer analysts rather than heroic figures testing themselves against heroic competitors for preeminence.
Indeed, it is unlikely we will see another Bob Turley pitching a complete game shutout in World Series game 5, then coming in to get the last out two days later in game 6 and finally pitching 6.2 innings the very next day in game 7 for the victory. I doubt we will see another Koufax pitch a complete game shutout in game 5 of the World Series and then three days later do it again in game 7. Has any pitcher thrown a 10 inning World Series game since Morris did it?
Of course, nobody any more starts 39 games as Whitey Ford did in 1961. Although it was unusual in 1961 for a pitcher to manage just 11 complete games out of 39 starts. That year 40 year old Warren Spahn threw 21 out of 38 complete games and two years later lost a 1-0 game to Juan Marichal in which both pitchers threw 16 innings. (Well, Spahn threw 15.1 only but it was a complete game.) At the time, many pooh poohed Ford’s 25 victories because he relied so heavily on Arroyo to preserve his wins rather than do it himself.
We don’t see that kind of heroic action, but the game is no less exciting to watch unless you are stuck in the past and cannot appreciate the excitement not just of the physical play but of the intellectual game that is going on. I consider the Rays the most interesting and most exciting team in the majors. Some people have talked about how they have a brilliant bullpen, but not one of their relievers has a history of stardom. In fact, most are discards who, when acquired by the Rays, suddenly turned it around. I find it exciting to see journeymen used to such great effect and wonder at the Rays ability to extract what they can do well while mitigating their weaknesses.
And the game on the field never ceases to excite no matter who is out there. Players like Phillips and Choi are a joy to watch as is Arozarena and Franco and McClanahan, Lowe, Kiermaier and now Rasmussen. Watching them play and also recognizing the background that put them there, enhances my enjoyment of baseball, it does not dampen it.
Yes, starting pitching has always been the central story of games. To some extent, it still is with the Rays when Glasnow pitched or now when McClanahan or Patino do. But there is more to the game and if fans are going to keep up with changes, we have to alter our perspective. I know fans tend to attach themselves to individual players, and that is tough with the Rays, who churn their roster so regularly, but I don’t think it makes them soulless. I root for the team which I find exciting, creative, surprising and a trailblazer. And while they are hear, I revel in the success of players who do not win the big awards but do win a lot of games.
Not to downplay, but Madison Bumgarner did something similar to that quite recently, and I have no doubt that any modern team would lean on their clear best pitcher similarly in the final days and innings of the World Series.
Randy Johnson also in 2001. His fastball was topping out at ~90 and he looked like he wanted to die, but he gutted through 1.1 to get the win in Game 7 the day after throwing 104 pitches.
(Note: the DBacks were up 15-0 after four innings in the game Randy went 7. The had Schilling going the next day and were doubtless feeling confident, but the smarter move probably would have been to pull Johnson earlier than they did.)
As i wrote above, i don’t feel that way at all when the giants do this. You must not be paying much attention if this is how you feel.
They have so many pitchers that make cameos over the course of a single season that I can see why a non-Rays fan would think the bullpen games are kind of soulless. I like seeing the variety of pitching styles and keeping up with the roster churn, but that’s just me. And I get that the Trop aesthetic isn’t for everyone. But I think that the defense, hitting, and baserunning more than make up for the other stuff.
Off the top of my head, Kiermaier, Phillips, Margot, Walls, Franco, Wendle, Brandon Lowe, Zunino, and Choi all routinely make dazzling and flashy defensive plays. When they go to the Margot-Kiermaier-Phillips outfield, it’s by far the best defensive outfield I’ve ever seen.
They have their share of stoic players (Wendle, McHugh, Wisler, Wacha, Zunino, plenty more) but all of Franco, Phillips, Diaz, Choi, Kiermaier, Arozarena, Cruz, Patino, and Margot have a very charismatic, joyful style of play.
Wander Franco seems to have figured out Major League pitching and he has been extremely entertaining on his own. 152 wRC+ with a 6.3% (that’s not a typo) strikeout rate since August 1st (142 PA).
I really don’t have any idea what you’re talking about when you say “soulless.” If you’re not a Rays fan, that’s cool, but if I watched the Rangers or Cubs on any given day I could just as easily say that such a game did nothing for my soul — I just don’t CARE about the Rangers or Cubs.
I know the Rays very well, know who they have on the bench and in the bullpen, and for me it’s interesting to try to “manage” the game from my couch. The Rays have just as many “fun” players as any other team, and the roster turnover isn’t as extreme as some would have you believe. And when they do churn the roster, it’s to make room for a superb prospect you’ve waited two years to see or watch some Orioles or Marlins retread reliever turn into Bruce Sutter for a month or two.
You’re simply adding to a national narrative that there’s something “wrong” with the Rays’ Way. That’s just an unwillingness to adapt, like the Rays continue to do. And even when MLB changes all the rules in the new CBA, I bet Neander and company still take advantage of the new rules “firstest and bestest.”
I dislike the Rays brand of baseball because they’re a three true outcome filled lineup and, most importantly, they have eradicated the figure of the starting pitcher from their organization. Bullpen games are background noise for me. I have no interest in seeing an endless string of nameless, uninteresting, limited pitchers who’ll face 4-6 batters and be out of the game. I do enjoy their defense and baserunning, tho, but pitching matters way more to me.
If Honeywell and McKay hadn’t been injured in the past couple of years, the Rays would have a very nice, standard-looking rotation. The strategy this year has simply been to roll with what they’ve got — it was never the intent to abandon the starting pitcher as a concept. If Glasnow doesn’t get injured, the discussion here would be considerably different.
This. 2021 was always going to be a weird year for SP given the lack of IP in 2021. Teams are unwilling to push young SP to huge IP increases and injuries have been brutal. Patino might go 5-6 IP in the playoffs but his control is still a wild card. Archer is another possible option who has been injured most of the year. Wacha and Yarbrough have given all they could and will finish with 130-150 IP. I don’t know how much more one could expect. We could also very well see some tired arms from other teams come October. We’ll see.
Rays starters have averaged roughly 18 batters faced per start since 2017. They play a boring brand of baseball.
Oh I disagree. Everyone’s different, but to me they are far and away the most entertaining team in baseball because you never know what you’re going to see when you turn them on. They’re the only team besides my own that I’ll go out of my way to watch
The giants do pretty much the same thing and i can say that as someone who follows all their games, it doesn’t feel soulless at all, because it isn’t just a bunch of interchangeable names, you think things like ‘oh, littell has been doing well in the last week, let’s leave him in for another batter’ or ‘maybe we better wait to use jackson until the bottom of the order, the way he’s been going’. But because i don’t follow the rays, when i do catch one of their games it looks kind of soulless, as you put it. But that’s just my lack of familiarity.
Do you think the Rays are better at sniffing out when their guys “don’t have it” on a particular night and pulling them (or not putting them in)? Are they rhapsodo-ing in the bullpen to see if someone’s stuff is playing up? Or using real-time statcast in the dugout? Is that allowed?
Ooh that’s an incredibly interesting idea. I wonder if there are any patterns- if we could define a “role” for each guy (pitcher A pitches against the heart of the order, or Pitcher B pitches against sequences of batters that struggle with high fastballs) we could then look for situations when they didn’t pitch when their number should have been called.
Or just ask someone on the Rays
For years I’ve read stories about pitchers having nothing in the pen, then being lights out in the game. And vice versa. But I suspect these are exceptions. Generally if a pitcher is sharp in the pen, he’s probably sharp in the game also.
I’ve fascinated by this. I’ve always wondered if you can tell it a reliever “doesn’t have it” when he’s warming up in the bullpen, and one would think the technology exists now to where you can
This is a neat idea, but you would kill your bullpen if you repeatedly have relievers warmup just to see if they have their best stuff working on a particular night. Especially with how much the Rays rely on the bullpen, this strategy put even more stress on the bullpen arms.
I…think this is a fallacy:
“The theory then goes that the more pitchers one uses, the greater the chances that such a crap-the-bed inning will occur”
That’s not true in the sense that there’s some chance that any given reliever is likely to throw a bad inning. A pitcher is as likely to perform poorly as their talent dictates. If you have a deep well of good relievers (like the Rays do) there’s no reason why you would expect one of them to blow up just…because. Trying to get a 7th inning out of Ryan Yarbrough is probably more risky than bringing in J.P. Feyerstein.
What makes the theory true-ish, is that the more pitchers one uses, the greater the likelihood that one is forced to take a chance with a not very good one. Lots of teams just don’t have a very good 4th or 5th reliever, so when they’re forced to use lots of pitchers they are forced to take a chance on some mediocre arms.
This doesn’t explain how talented guys like Aroldis Chapman and Chad Green blow up more than Andrew Kittredge
Well Crapping The Bed is an arbitrary metric made up just now for this article. We don’t actually know how useful it is to describe a pitcher, or how much it’s just luck, or if it’s predictive. It’s ALSO possible that Andrew Kitteredge is a better pitcher than either Green or Chapman at this point in their respective careers.
Crapping the Bed is a very real problem that afflicts millions worldwide
It’s probably just that Chapman and Green get more opportunity to blow up. They are in traditional roles, Yankees are winning by 2 in the 9th, Chapman comes in and pitches the 9th. If he gives up 3 runs and blows the game so be it, they will let him do it because he is the “closer”. Same thing with Green, he is the “set-up guy” he comes in during the 7th or 8th and come hell or high water and either bridges the gap Chapman, or blows the lead. The Rays probably pull the plug much faster on their relievers since they haven’t annointed anyone as the closer or set-up guy. They are all fungible.
It’s not a fallacy at all. Pitchers in general, and relievers in particular (because, in essence, they’re very limited pitchers), are always prone to poor outings where they just don’t have it, and the more variables you introduce in a game, the more likely it is that at least one of them will have one of those outings. If your plan is to ride the starter for 130 pitches, then he either has it or not, and only he will determine the outcome. When you use 6-8 limited pitchers to get those same outs, you’re opening yourself up for failure more often.
“Just doesn’t have it” is not a variable independent from the pitcher’s general level of talent and ability. The reason using 6-8 limited pitchers is suboptimal isn’t that there’s something magically risky bout introducing a new pitcher, it’s that the deeper into the bullpen you get, the more limited your options become. If your bullpen was made up of 8 identical true talent 1.50 ER pitchers it wouldn’t make any difference at all which or how many guys you threw in any given game (subject to matchups and usage fatigue, of course. I’m talking about the abstract case)
Fair, but in reality, that’s not the case. Also, there’s the simple fact that relievers are very limited, and the more looks teams get at them in a series, the worse they’ll perform. I’m surprised no one has done a study on that (the reliever equivalent of “times through the order”), because it’s extremely apparent.
And this is the risk that the Rays seemingly have successfully mitigated by building an ultra deep bullpen. Unlike most teams, they have a plethora of good arms who can nearly all be called upon at a moment’s notice in a variety of situations. They are not a team who’s starting rotation has fallen apart and they’re forced to use relievers out of necessity. They are a team that has been game planning this the entire time.
This is correct, but I also don’t see how this refutes mariodegenzgz. You’re saying that a measure of dispersion or reliability matters because it is necessarily a deviation from a measure of central tendency. But it also seems that you’re arguing that a measure of dispersion or reliability is also inherently related to the measure of central tendency which is…probably not right.
Actually, what I’m doing is having trouble making a distinction between volatility and talent. The theory claims, in essence, that relievers are inherently more volatile than starters, and using more of them is thus more risky regardless of the level of underlying talent.
And I don’t think that’s true. I contend that a true talent 3.00 ERA reliever is no more volatile than a true talent 3.00 ERA starter, and all else being equal they are both just as likely to throw a good (or bad) next inning.
Ironically the best examples I have of this are starters… Freddy Peralta and Tyler Glasgow. Sometimes they come in and are completely unhittable, and other times they come in and they scan not hit the side of a barn. I think the reason why it is easier to see with starters is that they pitch more innings but there are some nights where they are utterly dominant and others where you’re lucky if they can complete an inning.
Not all relievers are like that but since relievers are often guys who cannot repeat their mechanics to pitch for longer periods of time it should apply here too. Let’s say you have a guy who has 50 appearances of one inning each giving up no runs and almost no hits in 48 of them, but gives up a grand slam in two of them. Then there is another guy with the same usage patterns but who gives up one or two singles in each appearance, giving up one run in 8 of them. Same runs scoring average. Different ways they affect the game. It’s not clear to me one is better than the other, at least in terms of results. It is clear that the first guy might might occasionally lose you the game but otherwise will lock it down, while the other guy might allow a run but will limit the damage.
The idea that you’re more likely to run into a pitcher who “just doesn’t have it” by changing pitchers, rather than by letting the current pitcher go another inning, is the questionable part. Pitchers are great for an inning or two or three and then blow up all the time. Examining that proposition using available data seems like it would be fairly easy.
This. Exactly this.
The mere act of changing the pitcher does not engender the risk. When you pull the starter and the reliever blows up, everyone moans about how they should have just left the starter in. But, of course, there’s no certainty that the starter wouldn’t have also blown up, or that a different reliever may have fared better. Every time a manager makes the decision about changing a pitcher and which pitcher to select, they’re doing the math in their head. Certainly they can be wrong about what the starter has left, or what the reliever brings to the table, or pick the wrong matchup, or end up stuck with a lesser pitcher at some point later in the game, or any of a thousand other bad outcomes. But that’s where the risk lies, not in some inherent instability introduced by putting a different arm on the mound.
I don’t think that’s a fallacy, but it is not particularly clearly worded and so it’s hard to evaluate. Building off of your second point (which is absolutely true), it’s also true that sometimes pitchers are just “off”. and that’s what the CTB% is attempting to capture. Let’s say Yarbrough goes 4 innings and then you have to bring in all of Fairbanks, Feyereisen, Kittredge, McHugh, and Wisler to finish off the game (one inning each). That means you used six pitchers that game, and any one of those six pitchers could melt down and ruin your game. If Yarbrough goes 6 innings, then you only need four pitchers in total.
The point that the starter is more risky than the reliever the third time through the order is well-taken, but I also don’t agree that looking at the innings as completely separate from each other is right either because the same pitcher can pitch multiple innings. If Yarbrough is having a “good night”, then you would hope he could go later in the game and reduce the risk of a “bad night” from someone else. In more technical terms, without taking this into account we have a correlated errors violation. We would expect any starter to have some nights where they can go longer than others.
The CTB% clearly isn’t perfect and who knows if the thresholds set are right but it does seem like something that managers should watch out for. That the Rays have largely avoided outings where a starter completely melts down is therefore very interesting, and may point to player acquisition as much as the strategy around pitching changes or coaching.
I think a big part of why Rays “starters” melt down less than normal is because they’re often not given a chance to do it. They’re not allowed to pitch through traffic in the 6th, 7th, or further, which is where a few of those blow ups happen.
As much as I like the idea of Crap The Bed Percentage, I think you might be over thinking this a bit. The Rays’ reliance on depth over star power applies as much to their pitching staff as it does to their hitters. The Rays have given 61.3 innings to pitchers with negative WAR. The Jays, in contrast, have given almost 3x more innings to negative WAR pitchers. The more bad pitchers you have, the more Crap the Bed innings you will have
I don’t know their CTB%, but the other day I was curious wrt how the Mets could possibly be this badly off in such a weak division. Through August 29th they’ve gotten ERAs over 5.00 from 16 different pitchers, for a total of 178 innings with a 7.12 ERA.
Between allowing that and dumping Matz, failing to recognize that he had turned it around during the winter and thanks to his work with Phil Regan, that’s surely more than the margin by which the Mets trail the Braves.
I’ve often thought the same thing. If you need 6 or 7 pitchers to get through a game, that’s 6 or 7 chances for someone to have a bad night and since you are dealing with small sample size players, you probably know less about those guys than you think you do.
One counterargument is that if the starter has a bad day, you are in the hole also. It’s an “all your eggs in one basket” situation. Also, since you have 9 innings ahead of you and probably trust starters more than relievers in general, you can be prone to let him dig that hole deeper before pulling him. But I think that if that happens, you have more time to climb out of that hole. It is sort of like the Monty Hall logic problem – if the starter is good, you win. If he is bad, you get a second chance. It doesn’t really work that way if you plan from the start to only get 15 or 18 outs from the starter.
The other problem is what this does for the post season. You go into October with a list of guys you can trust, but you know anybody can have a bad game so one bad outing probably doesn’t take that guy off the trusted list. But players can have bad runs, like Chapman did earlier this year. If someone like Kettridge goes through his first rough period in the ALCS, he could CTB 3 times and the series could be over before you fully lose trust in him and stop putting him in high leverage situations. You don’t have the luxury of letting him “figure things out” over a couple of weeks in July to get him back where you need him.
All of this is magnified by the three batter minimum rule.
Despite all of this, I don’t see any team going back to letting 4 starters throw 250 innings or routinely letting them go through the order a fourth time just because they have only allowed 3 runs. The lure of bringing in a parade of guys who throw 97 with a nasty slider who all look slightly different and no hitter will see twice is powerful.
“Despite all of this, I don’t see any team going back to letting 4 starters throw 250 innings or routinely letting them go through the order a fourth time just because they have only allowed 3 runs. The lure of bringing in a parade of guys who throw 97 with a nasty slider who all look slightly different and no hitter will see twice is powerful.”
Sadly true… unless MLB gets serious about the state of their product and severely limits the number of pitchers on a roster. That would be interesting to see, and I hope it happens in the near future.
I think this frames the subject incorrectly. After all, the Starter has to get through 5, 6, 7, 8, 9 innings no less than 5 , 6, 7, 8, 9 relievers have to get through a game. Once we break down the Starter’s workload into components the way we do unique relievers (one inning at a time, at least for the moment, as a useful generalization) why isn’t the Starter no less prone to CTB in a given inning than a reliever?
I think teams are shifting and may continue to shift back to using long relievers, except instead of the role being thrust upon low leverage relievers in low to (at best) low-mid leverage situations, they will use long relief as more of a tandem starter model, where they count on someone to get them through an entire turn of the batting order two to three times per week. Guys like Cristian Javier and Garrett Whitlock jump to mind. Loasiaga, Petit. It’s one way around the issue of counting on several pitchers per night. 12 outs – 8 outs – two to three setup guys basically gets you out of the CTB problem. Realistically, bullpenning almost every game just doesn’t work if you have only short relievers, plus one mop up guy who can’t be used in any undecided game.
Look for more guys who could passably start – but probably not for 30GS / 200 IP – to act as med lev long relievers who often double as 6th and 7th SP options.
Also, along similar train of thought…
I’d love for stats that better tracked reliever usage than, I guess, appearances and leverage. Average batters faced or outs recorded, as well as number of appearances of each batters faced or outs recorded, would add context to how relievers are being utilized. Some guys can regularly deliver two or more innings; some can’t, or at least aren’t being asked to.
I’ve never understood the statement that in a bullpen game you are in danger of ONE of your relievers having a blow up game. This is true, but when you realize that the average ERA for a 4th or 5th starter is in the neighborhood of 4.00 – 4.50, you should then also realize it would take 2 or maybe 3 relievers having a bad day to do worse than that over the course of a game. In other words, you would consider it a good day if a reliever blows up and gives up 2-3 runs in an inning as long as the rest of the guys put up zeroes.
I would also assume a reliever having a bad day would get pulled a lot quicker than a starter having a bad day in the 1st or 2nd innings and thus more likely to keep you in the game and another way to mitigate the risk. Or a team that relies on a few good relievers to carry the load is more likely to stick with a guy who clearly doesn’t have it that day hoping he can get through it.
Usage becomes a performance issue in that today’s usage affects the next two days of availability, and we can’t quantify that because it changes every day and it incorporates information, including subjective information, that we aren’t privy to. More usage today means more opportunities for pitching on zero days rest, three out of four, four out of five, etc., which on average across all relievers, will generally mean worse performance. While your performance argument makes sense, a reliever isn’t necessarily going to be more likely to be pulled if the plan going in was to use six or seven different pitchers after the starter. In general, I think there are limits to the kinds of usage Tampa utilizes that are avoided by frequently shuffling the last few arms in their pen. They recently optioned a reliever with a sub 3 ERA for the eleventh time this year. Most teams don’t have that luxury.
It is cherrypicking but this analysis makes me think of Nick Anderson. He appears to have had a 0% CTB in the regular season last year and was absolutely dominant and then proceeded to befoul the sack several times in the WS, including losing Gm 6.
I think the Rays (rightly) really care about the times-thru-the-order penalty, but teams seems unwilling to acknowledge the same can more or less apply to relievers in the playoffs. If batters are seeing the same relievers 4-5 times over a 7 game series there is likely a similar ‘familiarity’ penalty.
Are relievers actually more inconsistent or is it that a starter’s inning’s help them move to the mean over the course of a game? In other words, do starters actually have less CTB INNINGS than relievers that are being masked by other great (un-CTB?) innings elsewhere in the start?
Would it be possible to look at the standard deviations of relievers and starters’ FIPs per INNING to see how much more inconsistent relievers actually are?
Hey, if you guys are going to do this “comment awaiting moderation” thing with me, you may as well delete my account, because I won’t be offering any comments in the future. After 8 years here, I say Ashley’s “What’s your favorite baseball movie” list belonged on buzzfeed rather than the premier sabremetric site, and you guys moderate my comments now?
Please delete the flag on my account that requires “comment awaiting moderation” from my account, or delete my account.
Drew Rasmussen will be their game 2 starter. He’s quietly been getting stretched out and has made five consecutive starts. His average fastball velocity (97.5) would be third in MLB behind only Alcantara and Cole if he had enough innings to qualify.
It could be that defined roles actually make the pitchers worse since 9th inning means Rod Beck whether or not Rod Beck has a good matchup or Rod Beck feels 100% or Rod Beck has pitched 2 days in a row. If it’s not Rod Beck then it’s some guy uncomfortable in the situation because it’s made a bigger deal than it needs to be to pitch the 9th. Kittredege can pitch against whatever matchup benefits him when he’s 100%, and be on the IL when he isn’t. The old adage of 80% of Star X is better than 100% of Joe Nobody may not apply among all but the very best relievers.
Living in Tampa has given me a unique perspective on the Rays and pitching deployment-I could spend 50,000 words on how $tu has circumvented & manipulated the rules to BOTH win games by manipulating arbitration/arbitraging players salaries combined with having a bottom 5 payroll maximize making profits …yes he is accomplishing BOTH and according to a lawsuit from his minority partners vs him he is cheating them also…and this ruthless, profit & equity growth approach is the foundation for this “on the field product”.
The starting pitching ERA is a bit skewed. Based on previous previous games usage Kevin Cash allows the starter to go out there and pitch until he gives up a certain bumber of runs based on is on his current bullpen availability(often times the last pitcher in the bullpen is sent down to minors to get fresh arms next day).
For example if Wacha is in the 4th inning and he has given up 2 runs and doesn’t seem to have it that day then his is pulled for a advantageous reliever match-up and his line is 3.2 inn giving up 2 earned runs or ERA of 4.96 , but the key is run prevention and you pull the starter BEFORE he gives up too many runs that puts the team at a disadvantage to win. What’s different this year is that the offense is ranked number one, so he can have have a little longer leash than he did in the past 2 years since they are scoring more runs.
The starting pitching ERA is skewed because it looks like the starter is allowing in X amount of runs per 9 innings but in reality he is never allowed to pitch long enough to give up X .
When you have Glassnow or or any other starter that has a deep arsenal then you revert back to a more familiar pattern of usage.
The challenge the team is going to have during the postseason is they don’t have the ability move pitchers up-and-down from the roster vs having a set amount. The Rays average 12-13 pitchers during the regular season roster and in a 5 game series will have 12 or 13. Since they don’t have Charlie Morton(back to the bottom 5 in payroll vs really trying to win concept, $tu wouldn’t allow them the 15 M to sign him and he lives an hour away and really wanted to stay!) or any consistent traditional starter, can they minimize run prevention using the previously discussed approach without running out of available arms?
Looks like they will need 1 or 2 of their starters to go at least 5 innings during a 5 or 7 game series in order to keep their top 4 relievers to pitch enough innings to keep run prevention down … Or they do what has been occurring during the regular season which is score score more runs and this allows room foroom for error in the run pervention current formula.
You have to give credit to people who have developed this incredibly intelligent and ruthless design, but don’t forget is was born from $tu relentless greed.
Since he is not getting the money he is asking for which is pretty much 75% of the cost of building a new stadium from either Tampa or Saint Petersburg, he is now saying he has to have a stadium in both Montreal and Tampa and create a 2 city plan in order for him to Make more money than he already is now because let’s face let’s face it his current situation in Tampa is causing him a lot of work to be this greedy.
Is it feasible to go with a 3 man rotation and 10 relievers in a 7 game series? Or a 3+3 piggyback rotation? They also use a bunch of position players so they can’t go to 15 pitchers or anything like that. Another option would be to have a different set of relievers in the ALDS vs the ALCS. The offense is the strength of the team so it should probably get priority.
Disregarding the other items you wrote about, I think the two items of most impact/interest are as follows:
“(often times the last pitcher in the bullpen is sent down to minors to get fresh arms next day)”
“The challenge the team is going to have during the postseason is they don’t have the ability move pitchers up-and-down from the roster vs having a set amount”
—
I agree with you. I have noticed that the Rays really shuffle their bullpen around, bringing up guys from the minors and sending guys down very often. I would fully expect that they have some sort of plan to bring up “pitcher A” from the minors in two days, so he shouldn’t get in a game in Durham before then. Such an approach really extends their staff…in the regular season.
With the limited number of moves allowed in the playoffs, I do think that the pitchers on the Rays will be taxed more. If this means going deeper into the games (by the starters) or trying to get one more out (from a reliever), it will be a challenge. The longer the series goes, the less the Rays can use this advantage.
The Rays are the team everyone thought the A’s would be.
I would expect that they Rays will position themselves to (if they need to) use the postseason injury roster exceptions to their fullest advantage to shuffle out overtaxed relief arms.
The a’s are stuck in the past (and i write this as someone who basically likes the a’s).
From poking around on B-R team relief pitching stats: Tampa leads MLB in outs per reliever appearance and pitches per reliever appearance. They are third in the majors for number of appearances recording three or more outs and multiple innings. Notably, they have the second fewest relief appearances recording fewer than three outs, ahead of only Milwaukee. They are well below league average for number of appearances on zero days rest (the Dodgers and Giants are both tied for the MLB lead).
This is an interesting take, but what I’d like to know is how they’ve done this without completely burning out their relievers and/or finding an endless stream of reliable relievers to replace the ones that get hurt. I think the main problem with this strategy is attrition, but the Rays seem to keep pushing the envelope on this and it keeps working.
It is very much a regular season vs playoffs roster limitation issue here. How will they get the innings form their bulk pitchers to limit the appearances and length demanded of their relievers that can no longer be switched out – absent an injury – in a given playoff series. It did seem that this issue wore on them in the playoffs last year, culminating in the World Series.
That said, what they have achieved this year is simply remarkable. Just looking at the injury attrition their pen has taken with 10 relievers on the 60 IL it is amazing that they have maintained the performance level they have. They clearly have an advantage in their ability to evaluate what castoff guys from other orgs could do – look at recent addition Shawn Armstrong,, perhaps the poster boy for “castoff reliever”. He came form Baltimore with an 8.55 ERA and 1.90 WHIP to Tampa and has a 2>57 ERA and 0.71 WHIP in about 2/3’s the IP. He is a flyball pitcher – you might think Baltimore was punitive but his HR rate is still quite similar, but with less guys on they are more often of the solo variety. Kinda small sample size here too of course. But then you look at their other “castoffs” Feyereisen, Wisler, Chargois, and Conley and they seem to repeat it over and over
Or it is just remarkable luck – but that seems unlikely.
You can’t use last years post-season at all. It was both longer than the regular postseason format and less days off. You get a ton of off-days in a normal post-season compared to last year. The Rays will be fine. They will win while the Yankees will not.
The Rays do one thing better than everyone else. Identify what a pitcher does well (typically at least elite at one thing – Armstrong has ridiculous spin rates on his pitches and his 4S is elite. His slider sucks. Cutter is in the middle. Ergo, Rays have him throwing more 4S and cut the slider out completely) have them do more of that and less of the things they suck at and have a tremendous biomechanics/movement analytical staff. (How arm slot, tunneling, pitch selection is conveyed)
They’ve found an endless stream of reliable relievers.
I don’t know if it’s some sort of market inefficiency, or better scouting, better coaching, or better situational usage. I guess it’s all of the above to a degree. But it’s really quite something.
The giants also seem to have done this. Not sure how.
I keep predicting they can’t keep doing this forever, that their relievers will get hurt or tired and therefore ineffective, but it seems like their reliever usage has gone up every year and it still works. Next year I will probably say they can’t keep doing this forever, and I’ll think that it will be the year it fails. And if history serves I will be wrong again but I really think there is a finite limit to this strategy somewhere, even if it’s not where I thought it was. It’s one of the most fascinating storylines in baseball strategy. How far can they push it?
please add CTB% to fangraphs stat pages — I think we need something like this to accurately value volatility of RPs
Seconded – I like the information that it provides and the name is fun to boot.
There is the meltdown stat (losing .06 WPA during an appearance) they keep but is rarely mentioned. However, it’s not a rate stat and it’s name is far inferior to CTB.
It isn’t the same thing because CTBs in a bullpen game won’t normally produce enough of a WPA loss to count as a meltdown (unless the team is already ahead AND they’re really catastrophic, OR if they’re late in the game).
They sound a lot like the giants, who just won their second bullpen game out of their last 5 games. I don’t know about their ctb%s, but they seem to deal with the danger of ctb by pulling relievers fast when things start going in the wrong direction and being very flexible about how many batters each pitcher will face. Yesterday leone pitched 2 innings and alvarez pitched 1.2 because they seemed to be doing well.
This article is why I love this website. Fantastic observation and ability explaining it in simple terms!