I can no longer sit idly by and say nothing – Derek Jeter has accepted a new role with the Miami Marlins. According to Twitter today, he is in great danger. What kind of danger, you say? Why, the danger of musculoskeletal disorders. THATS WHAT.
Let’s do an ergonomic assessment of Jeet’s office, and make some recommendations to help him out.
A new era begins. pic.twitter.com/WkqL2oyn2O
— Miami Marlins (@Marlins) October 3, 2017
This is the picture that lead me to believe that Derek is in grave danger. Take a look at that office set up – I’ve seen safer wood chippers. Derek – are you aware that your franchise is worth $1.17 billion, Giancarlo Stanton gets paid $25 million a year, and that chair is clearly from Staples. At most, they bought you a chair that cost $400.00. The biggest predictor of injury, is previous. With 8 trips (at least), to the DL in his career – he’s a huge risk for injury.
To take a look at how much of a risk Jeter’s office is, I’ve used the Rapid Office Strain Assessment (ROSA) – an ergonomic tool that examines how your office is used, and assigns a 1-10 scale representing an increasing risk of musculoskeletal discomfort. The research I did for my masters showed, if you have a ROSA score of greater than or equal to 5, you are at a greater risk of discomfort. Let’s dive in.
The Office Chair
I’ve expressed my displeasure in the chair itself, but how Derek is sitting is just as big of a problem. Not using the back rest? That’s going to cause his lower back muscles to just be on, firing at a low rate all day. “OH MAH BACK, HANNAH GET THE A535 – I GOTTA SLEEP IN THE RECLINER TONIGHT”.
At 6’3, we can make some assumptions about his chair. The first, is, it’s likely too small for him. It’s also not likely going to be adjustable for his gangly frame.
Most of these cheapo chairs have an adjustable seat height, and his knees look close to 90 degrees. We’re safe here.
Like I said – the chair depth is likely too short with him being so tall. Cheap chairs usually aren’t adjustable in the seat pan depth either.
The armrests – they’re not being used. He has his arms on the desk surface, and it looks like that surface is a bit high. There’s no way that cheap chair has adjustable arm rests, and it look a lot like a pressure point when his forearms are resting on the edge of that desk surface.
His chair back is likely not as much of a problem as is him leaning forward and not using the back rest. Also baseball executives?? Those are some loooong hours. Mark it up as a risk factor.
What in the honest hell am I seeing here. Seriously. How is this even functional?? This is as useless as a jump throw from short when some injured jabroni is crawling to first base. Two iPads – to the side, with his documents on the other side of the desk. MARK IT ALL UP.
I’m assuming he just uses his iPhone and talks with it in his hand when it’s on speakerphone and is a total annoying jerk about it. I don’t want to hear your conversations man, just put in your headphones.
Mouse and Keyboard
I have no idea what’s going on here. But, it looks like the bathroom of a Buffalo Wild Wings after all you can eat curry wing night. Does he have to use the touch screens all day? His arm is going to fall off. Does he type on those things? His wrists are going to get carpal tunnel, then that carpal tunnel will develop a small ecosystem inside of it, and everything in there will get it’s own separate disorders. Like Inception, but for wrist pain.
The final verdict
Hi Derek, come into my office. I’m Dr. Mike. I have your test results here, and, I’m sorry… but your ROSA score is above 5. You’re at a high risk of discomfort. I’ll leave you alone for a bit to process this, and then we can go through some of your options.
Mostly, before we can get a good “YEAH JEETS” for this office set up, he’s going to need a more ergonomic chair. ErgoCentric makes chairs that will be great for taller guys like DJ.
The biggest thing is going to be, plugging those iPads into real accessories, and elevating them so they’re at eye level. A bluetooth keyboard and mouse – then putting the screen right in front of him – that’s going to make his head and Neck in a way better position. He’s also going to take a lot of pressure off of his arms and wrists while typing or mousing, or going on YouTube and watching his own highlights while the team around him implodes and trades off yet another crop of stars.
Peer review is the necessary rite of passage that all academics go through before they achieve all of their riches and glory. Actually, just the glory. Well, really, just their name on a paper. But I guess that can be glorious at times.
When you submit one of your ideas to a journal for peer review, and hopefully, publication, what actually happens? In 2016, I submitted a paper to the Journal of Sports Sciences, where I aimed to use a fatigue model I had developed to investigate if Pitch Clocks would increase fatigue levels in Major League Baseball pitchers. This paper went through 3 intense rounds of revisions, faced the brink of rejection, before ultimately being accepted for publication. I wanted to write a little bit about the peer review process for those of you who aren’t familiar with it, just to show you how important the process is to understanding the way the human body works.
When I prepared the manuscript for submission, the journal asked that I remove all Author names and affiliations, to make it a blinded review and keep the authors from biasing the reviewers. This is a good first step – you shouldn’t get a paper accepted just because of the name that wrote it. That being said, I was using a fatigue model that hadn’t been used in practice before, and the only reference to it had my name all over it. I’m pretty sure that the reviewers could have put 2 and 2 together to figure out who the first author was – but at the same time, my name doesn’t really carry much clout. I’m not going to bias anyone.
The next step is to find the reviewers. The journal asks that you recommend three “arm’s-length reviewers” to closely critique your paper. These are researchers that you haven’t been on grants with, published with, or worked with. You very well may know these reviewers in a professional sense, but they should be able to review your paper without having any professional gain or detraction from their career by either accepting, or rejecting your work.
This is where the biggest downside of the peer review process takes place. I submit my first iteration of my manuscript on April 19, 2015. I received a request for revisions and feedback from the reviewers on July 8th, 2015. After resubmitting in July, I then waited until October 26th to receive the next round of revisions. By the time the paper was accepted, and the paperwork was signed to get the manuscript out to the general public – it was mid-February. Nearly a full year had passed from the time I submit the research, to when the public could consume it. This was a bit discouraging, as I had wanted to get the paper out before the Winter meetings so that GMs and rule makers in the MLB could have access to it – in the end, it all worked out, but it was a frustrating time while it happened.
So, what do these reviews look like?
There were 3 reviewers that looked through this manuscript. Each of them provided very detailed comments regarding the manuscript. Reviewer 1 had 10 comments, Reviewer 2 had 1 comment, and Reviewer 3 had 8 comments. Here is a sampling of some of the concerns they had:
The author(s) explore an interesting concept of potential pitcher fatigue within innings/games. With the new pace of play rules adopted by Major League Baseball, this is a very worthwhile endeavour. However, the results and discussion are based upon methods that are unclear. More details about the model used, and how exactly it was applied, are necessary to properly evaluate the contribution of this work. A discussion of the model’s sensitivity to assumptions made would also be beneficial. Finally, an analysis of velocity changes within a given inning or throughout a game may provide a more direct measure of fatigue.
The author(s) predicate their work based on the assumption that joint stability in the elbow is analogous to joint stability in the spine. Majority of references provided are for the spine, not the elbow. The argument that muscular fatigue alters joint position sense in the upper extremity, potentially leading to greater loading on passive tissues, is a much stronger justification for the present work; and hence, should be expanded upon in the introduction and discussion.
I applaud you on your topic selection as I think it is relevant and interesting. Unfortunately, I think you had to make too many assumptions in calculating your results. Also, as injury is ultimately what we are interested in, I found it difficult to relate your findings to increased risk of injury and truly quantify what that relationship is.
Although I agree with the arguments put forth by the authors in the introduction, as well as the intended purpose. I have difficulty seeing the contribution that this study provides to the current literature in its current form.
First, there is a complete lack of information regarding the model used to predict fatigue. From my understanding, the model was built from isometric hand grip, or isometric thumb contractions. This data is not appropriate to model the fatigue from pitching which is a complex movement between flexion/extension and axial rotations of the wrist, and elbow. Some of the muscles they include in their study have complex interactions at both joints (wrist and elbow). If the authors already had some data indicating that it is an appropriate model for flexion and extension of the wrist/elbow joint muscles, it would have been valuable to share. I think for this paper to be accepted this major limitation must be either validated before the fatigue related conclusions can be accepted.
I think the authors have some interesting conclusions to draw from their efforts, and so I am suggesting major revision. However, without this substantial amount of additional work I do not see the manuscript as publishable in its current form. The conclusions they draw are of substantial interest to those in the sports science and medicine occupation, as well as those working in Major League Baseball. Therefore this work must be completed to a high degree of scientific merit, which I hope the authors are willing to provide.
At the time I got these reviews back, I remember thinking – this is just about as harsh of a critique as I’ve received from any manuscript I’ve submitted. At the same time, the comments given by these reviewers were exceptional – every single point would go along way to strengthening the paper. The biggest concern, was that this was a completely theoretical approach to proving pitch clocks could cause an increase in muscle fatigue. This concern was later alleviated when another research group evaluated the physiological properties of fatigue in groups that used pitch clocks.
We took these comments to heart, and made significant changes to our manuscript – the findings remained the same, but we incorporated some of the literature that the reviewers recommended. We also found a few more papers that looked into muscle activation patterns of the forearm during pitching, and included those in to the fatigue model. And then? We waited…
Another large list of comments from the reviewers – once again, these comments were very constructive and bettered the paper. The biggest difference this time around, is that we didn’t necessarily make changes to the paper – but had to explain more in depth some of the assumptions, and limitations our model and study presented. Here is another sampling of a few of the comments the reviewers provided. Reviewer 1 had three major comments, and Reviewer 2 had 32 COMMENTS. Reviewer 3 was fine with the changes we made. I’m sure if you’ve cruised any of academic Twitter, you’ve heard the jokes about reviewer 2. Well… there’s a reason for that! (Reviewer 2 provided excellent comments, that’s for sure).
74% of academics hate reviewer #2. 36% of academics actually are reviewer #2.
— Fake Academic Stats (@FakeAcademStats) May 19, 2017
I appreciate the additions the authors have made to the paper. However, the authors seem to be missing an opportunity to make relationships to injury or biomechanics literature that is relevant to pitching. There is a lot to be said about how reductions in muscle force capabilities can affect the injury mechanism (valgus-varus joint stability). A 5 % force decrease might be substantial or insignificant, depending on the force-length, force-velocity. There is also a comment to be made about the muscles that stem from the common flexor tendon. Epidemiological, and cadaveric models indicate that these dynamic stabilizers are most important in the direction where joint stability is compromised during pitching. All the literature that you suggest stem from joint flexion/extension. This is not the suspected injury direction. In regards to this, I think the discussion of this paper will benefit from more comments on these muscles rather than the brachioradialis. How my comments fit into the comments from other reviewers will decide how much to add and take away.
The work of Holmes and Keir only looked at the flexion extension stiffness of the elbow. This is an inappropriate citation to apply to the valgus-varus stability of the elbow. There exist many other authors who have looked at muscle contributions to valgus-varus joint stability. This is a comment on all mentions of stability in this paper. The authors seem to be including all elbow stabilizers, independent of movement. I would argue that the brachioradialis is not a stabilizer of the valgus-varus movements at the elbow, in fact in the citations the authors reference, it increases UCL strain. The brachioradialis is however, a stabilizer of flexion extension stability. This is stability has never been identified as a reason for the most commonly seen baseball injuries. I urge the authors to be more conservative in their mention to stability in the elbow as a general term. Stability is dependent on movement direction.
-Lin et al. (2007). Muscle contribution to elbow joint valgus stability. J Shoulder Elbow Surg.
-Udall et al. (2009). Effects of flexor-pronator muscle loading on valgus stability of the elbow with an intact, stretched, and resected medial ulnar collateral ligament. J Shoulder Elbow Surg
-Park and Ahmad (2004) Dynamic contributions of the flexor-pronator mass to elbow valgus stability. J Bone Joint Surg Am.
The use of EMG as in an input does not validate a model to all movement types. The normalizing of muscle demands completely ignores the requirements of muscle energetics. With isometric muscle tasks you are limited by blood perfusion issues which are affecting oxygen uptake and metabolite flushing, this is not the same limitation in dynamic muscle contractions. Furthermore, you have built a model on local muscular work. The activity you are applying it to is a whole body. Whole body exercise provides other demands on the body’s energy sources apart from the demands of that local contraction. Simply having a normalized input, does not provide validity to a method. The reader must be made aware of this severe limitation, or citations must show that
1) EMG is appropriate for isometric, eccentric and concentric muscle contraction alike. Which includes being able to measure MVC and current motor unit usage during each movement.
2) That isometric fatigue models can, and have been applied to concentric or eccentric models.
If there exist no citations, then more care must be taken to show the reader the expected limitations. You are either over or under predicting fatigue with your model. I agree with the direction of the results you have found, but the number you land on must be put in perspective.
With more thought into the experimental protocol and analysis…I am unsure if null hypothesis significance testing is appropriate here. You have controlled inputs to your model. Your results are completely dependent on the inputs that you have put in. The only variability you have is the inputs of pace, but those are estimates at best. Since you have each pitcher using the same EMG, these results could essentially be arrived upon by inputting theoretical pitchers (Monte Carlo simulation). I feel confident that statistical analysis is unneeded and do not provide any additional understanding. The effect size is an appropriate tool for identifying important changes in muscle fatigue, but as mentioned elsewhere, a mechanical argument of joint valgus-varus stability is a better.
Once again, these were great insights into the paper. We removed the significance testing from the results section, and included the references on valgus/varus stability.
Our paper was accepted for publication in the Journal of Sports Sciences.
I wanted to write this post for a variety of reasons, but the main one was to discuss just how important this process is to better understanding how the human body works. Without the input from these reviewers, this paper may have been accepted into a lesser journal, but it would not have come close to having the reach and impact that it did in its final state.
Keep this in mind – the reviewers of these manuscripts made no dollars to do these extremely thorough reviews. Not an extra cent over their normal salary. Like I mentioned, reviewer 2 had 30 comments in the second round of revisions for this manuscript – that isn’t something that takes an hour to do. These reviewers were anonymous – I have my thoughts on who they likely are, but they will get no formal credit for the large amount of work they put in to this paper. This was an exceptional case for me, but all of the manuscripts I have submitted for publication have received great attention from the reviewers.
Keep this all in mind when you read about the evils of peer review. The blogging world has been a great experience for me – it allows me to express my ideas and findings to a larger community. However, you can’t replace the process of peer review when it comes to bringing your big ideas into the world. Yes, it takes a long time. Yes, there can be cronyism when it comes to presentations and conferences – but in the world of sports sciences, this process ensures that your ideas are thoroughly tested, and that your results carry the biggest impact that they can.
Whenever I write something that trumpets the importance of velocity for pitchers, someone jumps in and says “velocity isn’t everything. Gregg Maddux exists”. You’re damn right he exists. You know who else exists? Vesna Vulović. She’s the Serbian flight attendant that fell 33,000 feet from an exploding airplane and lived. Yes, she lived, but I don’t see many people lining up to see if they can replicate her landing strategy.
The moral of the story is, if you throw 89 mph – the chances are – you are a bigger freak than the pitcher that throws 97 mph. Throwing with pin point control is a much more difficult task than throwing hard – for example, the AL Cy Young winner in 2015 missed his spot by 12″ on average according to CommandFX data (and Astros MiLB pitching coach, Drew French)
@drivelinebases The AL cy young winner in 2015 missed his intended location on average by 12 inches
— Drew French (@Drew_French) May 28, 2016
How do you teach command? Look at the moments immediate before ball release in this paper by Matso et al., (2017). I’m am highly interested in the training techniques that can be developed to increase precision in this type of movement – but we do know, if you make your pitchers strong, there’s a relationship with them throwing harder.
So, let’s look at the relationships between velocity, command, and control, with outcome metrics, for the final time.
I have been mildly obsessed with the command and control metrics produced by baseball prospectus, because they tell a far more representative story about what a pitcher is trying to do than walk rates. You can have exceptional command and control, but if a hitter lays off that full count slider that you located perfectly outside the zone, that’s still a walk. The called strikes above average metric gives us an idea of a pitcher’s command – when they throw a pitch on the black, how do they compare to the rest of the MLB in getting it called a strike. The strike probability statistic gives us an idea of the likelihood of a pitch being in the zone. This is indicative of the control a pitcher has – can he throw his fastball for a strike – regardless of how pin point accurate that pitch is – when he needs to. I’m going to use these as the metrics for command and control in this article.
So, on an island, how does pitch velocity, command, and control relate to a pitcher’s performance? I’ve gone with a gigantic sample for this study – anyone who has thrown 50 innings since the 2008 season. Each pitcher-season represents one entry – so there could theoretically be multiple entries for one pitcher. This resulted in a sample of 2782 data points. Here are the correlation coefficients for all of those metrics. A negative value means that with an increase in one metric, there is a decrease in the other (for example, a -0.36 between velocity and xFIP means that as velocity increases, xFIP decreases). The r2 value represents how much variability in one metric can be accounted for with the variability of another metric.
|Correlation Coefficient (r)|
So, our take home point here, is that while no individual metric accounts for a huge amount of variance in output metrics, fastball velocity is by far the best predictor of success when compared to command or control.
But, there are pitchers who survive without having great fastball velocity. Remember? Greg Maddux? HOW CAN YOU FORGET.
So, let’s compare the top 25%ile velocity pitchers against the bottom 25%ile pitchers. Clearly, those who operate in the lowest quartile of fastball velocity (less than 90 mph) are just as successful as those who throw extremely hard (over 93.4 mph). Just ask Twitter.
|High Velocity (n = 722)||94.75 ± 1.14||3.61 ± 0.69||3.61 ± 1.06||3.43 ± 1.1||3.29 ± 1.04|
|Low Velocity (n = 702)||88.38 ± 1.47||4.24 ± 0.6||4.71 ± 1.22||4.17 ± 1.13||2.93 ± 0.95|
|K/9||GB%||Swinging Strike %||Command||Control|
|High Velocity (n = 722)||8.96 ± 2.09||0.46 ± 0.08||0.11 ± 0.02||-0.25 ± 0.7||0.47 ± 0.03|
|Low Velocity (n = 702)||6.51 ± 1.55||0.44 ± 0.09||0.08 ± 0.02||0.51 ± 0.86||0.46 ± 0.03|
The metrics that jump out to the first, are the fact that high velocity pitchers have more than a half run of ERA and xFIP advantage over the low velocity pitchers, and OVER a 1 RUN DIFFERENCE IN DRA. More than 1 full run! There is an advantage for the low velocity pitchers in walking fewer batters – by 0.3 walks per game. Compare this to the strike out rates, where high velocity pitchers strike out over 2 more batters per 9 innings than the low velocity pitchers. This means that the low velocity pitchers gain their effectiveness through bad contact, right? That is also not true – there was no statistical difference between the high velocity and low velocity groups for ground ball rates. Also, the low velocity group had better command – they got more called strikes on the edge of the strike zone compared to the high velocity group, but they didn’t have a better chance of throwing a ball in the strike zone.
Finally, I just looked at the proportion of pitchers in the high and low velocity group that had “elite” ERAs (in the top quartile) during a season. This value represented an ERA of less than 3.00 in a season. There were 38% of pitchers in the high velocity group (271 of 722 pitchers) that had elite ERAs. The low velocity group? Only 15% of pitchers (104 of 702 pitchers).
Even further – look at the proportion of pitchers who have elite ERAs in each percentile of pitchers, ranging from 87 mph to 95+ mph fastball velocities.
It’s tough to argue that velocity doesn’t rule.
I know this blog has become a bit redundant with this topic recently, but I want to highlight why I feel this is an important topic. There is extensive debate about how to develop effective pitchers. Broadly, this comes down to two camps in pitching twitter – those who try to say that velocity isn’t everything, and you should teach kids to throw with better command, and one that says throw as hard as you possibly can. This debate is over. Velocity wins. There is countless evidence to suggest that giving a batter less reaction time, limits their ability to effectively execute a movement pattern and hit the ball.
The challenge is no longer whether or not it is important to develop pitchers who throw hard. Those who make it to the major league level without a plus fastball are the exception to the rule, and not the alternative to throwing hard. If you want to make a difference – figure out how to help pitchers get stronger and keep their strength throughout a season (Toby et al., 2015), reduce their fatigue, and keep them healthy.
Matsuo, T., Jinji, T., Hirayama, D., Nasu, D., Ozaki, H., & Kumagawa, D. (2017). Middle finger and ball movements around ball release during baseball fastball pitching. Sports Biomechanics, 1-12.
Toby, B., Glidewell, E., Morris, B., Key, V. H., Nelson, J. D., Schroeppel, J. P., … & McIff, T. (2015). Strength of Dynamic Stabilizers of the Elbow in Professional Baseball Pitchers Decreases during Baseball Season. Orthopaedic Journal of Sports Medicine, 3(2 suppl), 2325967115S00163.
The hits just keep on coming from the Motus Sleeve and Driveline! Last week, the brains at driveline posted their data from 70 pitchers throwing 5 pitch, fastball bullpen sessions, and the Motus and velocity data to go along with them.
For those of you who don’t know what the Motus Baseball Sleeve is, it is a device that fits neatly into a compression sleeve, which can be worn by baseball pitchers to get the predicted ulnar collateral ligament (UCL) stress that occurs during throwing. It also provides pitchers and coaches with data on the arm slot (with respect to horizontal), the arm speed (in RPM), and the maximum predicted external rotation of the shoulder.
I have explored the relationship between pitch velocity and elbow stress on this site before, and the driveline team has done the same thing for both fastballs, and breaking pitches. Currently, pitch velocity appears to explain about 17% of the variance in elbow stress. That means – before you look at anything else – pitcher height, mechanics, fastball type – you have 17% of your elbow stress already accounted for. While this doesn’t seem like much, you can see there is a positive, significant relationship between velocity and elbow stress.
So, if 17 % of the elbow stress is accounted for by the pitch velocity, where are the other 83%? And, if two people throw at the same speed, but one has higher elbow stress than the other – what’s going on there?
I broke the driveline data down by velocity percentiles – just to help normalize things. I then broke this down into 5 % ile windows. At each 5%ile window for velocity, I looked at what the average predicted UCL stress was. I then determined if the pitch had greater than, or less than average stress at a given velocity range. Finally, I examined the average arm slot, arm speed, and shoulder rotation for those who had less than average, or more than average elbow stress, at a given velocity.
When examining the data by pitch, there were 420 samples (nice). Comparing the higher than average stress, and the lower than average stress groups at standardized pitch velocities, there were some interesting findings. To determine statistical significance between the low and high stress groups, a t-test was used with significance set to p < 0.05. Cohen’s d was calculated for significant effects.
Higher stress pitches at a given velocity had significantly lower arm speeds (figure 3A), and significantly higher arm slots (p < 0.05, d = 0.26, and 0.21, respectively). There was no statistically significant difference between maximum shoulder rotation angles in either group (p < 0.05). This is particularly interesting, as Whiteside and colleagues (2016), found there was a significantly higher risk of UCL reconstruction surgery in pitchers with less pronounced horizontal release points. These data seem to support their findings.
The next step is to include anthropometric measures of height and weight into these calculations, and see how those improve the prediction of elbow stress alongside the key motus metrics.
O’Connell, M., and Boddy, K. (2017). Bullpens, Tracking Elbow Torque, and mStress. Retrieved from https://www.drivelinebaseball.com/2017/03/bullpens-tracking-elbow-torque-and-mstress/, March 4, 2017.
Whiteside, D., Martini, D. N., Lepley, A. S., Zernicke, R. F., & Goulet, G. C. (2016). Predictors of ulnar collateral ligament reconstruction in Major League Baseball pitchers. The American journal of sports medicine, 44(9), 2202-2209.
If you follow me on Twitter, you’ve likely noted that I get quite upset about the idea of pitch clocks. Why would I hate the supposed silver bullet that will end all pace of play concerns, bring new fans to the game, cause all fans to lose 25 lbs instantly, and make McDonald’s a health alternative to vegetables? Despite what you’re hearing about the benefits of pitch clocks, there are some concerns related to the health of pitchers. Let me explain why I have taken such a strong stance against them.
Muscle fatigue is a very specific process that decreases the amount of force that a muscle can generate. It is NOT just an expression of how tired a person feels. If a muscle is active, small amounts of muscle force generating capacity are lost at the smallest function level of the muscle – the motor unit. Overall muscle force at a joint level can be maintained by switching between which muscles are active, or recruiting additional motor units, but you cannot escape muscle fatigue. So let’s be clear on that – muscle fatigue is a decrease in muscle force at the muscle fiber level. The other type of fatigue that we study as researchers, is central fatigue. This is related to decreased output from the system prior to the muscle – so that means, a reduction of the available oxygen in the bloodstream, an inhibition in neural drive from the brain (such as pain, or decreased motivation), or inability to remove metabolites from the blood stream. These factors can occur, or not be occurring at the same time as muscle fatigue.
Dr. Jeremy Bruce, and Dr. James Andrews wrote a review article on Ulnar Collateral Ligament injuries (Figure 1), and one of their primary findings centered around the role of muscle fatigue on UCL ruptures. The findings boil down to this: when you throw a baseball as hard as you can, the stress on the elbow is theoretically more than enough to tear your UCL. What protects the UCL, is the activation of the muscles of the flexor pronator mass. If those muscles become fatigued, they will not be able to exert as much force, and as a result, the UCL becomes at risk of rupture.
So, how do you get away from this evil fatigue monster? Once again, its complicated (it depends on the types of efforts you’ve performed, how long you’ve been working for, etc), but in its simplest form, the prescription is: Stop! Recovery is ultimately dictated by the time you can rest. The time between pitches, or the time between innings, is when pitchers have time to recover from muscle fatigue. This is where my disdain for pitch clocks comes from.
Last year, I published a paper in the Journal of Sports Sciences 2 on the implications of pitch clocks on the accumulation of muscle fatigue. Compared to the self selected paces pitchers had (from FanGraphs), forcing pitchers to throw every 20 seconds increased fatigue levels for starting pitchers (Figure 2).
Technically, the rules (Rule 8.04) state that the pitcher has 12 seconds to throw a pitch once receiving the ball from the catcher – this is obviously not enforced. If this was actually enforced, there would be even more fatigue in pitchers.
Shortening the amount of rest between pitches will increase the amount of muscle fatigue in pitchers. If you had 24 seconds of recovery, and now you only have 20 seconds of recovery – you will be more fatigued. How would this look in a typical major league baseball game? Not too long ago, I described a method to make some interesting graphs showing predicted fatigue levels in baseball games, using my fatigue model, and MLB’s GameDay data. As a Blue Jays fan, one of the most exciting moments I can remember as an adult, was when my favourite pitcher – David Price – was traded to my favourite team. In one of his first games as a Blue Jay, Price went 8 innings, and struck out 11 Minnesota Twins. Price has routinely been regarded as one of the slowest pitchers in the MLB – his pace often ranks in the top 3 of all starting pitchers. I replaced his long stretches during his games with the 20 second pitch clock limit, and this is what we’re left with:
In this situation, (and without assuming a significant influence in the other team’s pace), we see a reduction in time of 8.38% – going from 167 minutes, to 153 minutes – keep in mind, this doesn’t factor in the bullpen visits, or the time spent celebrating the victory afterwards. In this case, this was the time of the first 8 innings for the home team – so there would be an other inning and a half to come. At the same time, you can see in red (the condition where there is a pitch clock), that there is a lot more fatigue with the pitch clock – particularly in the 4th, and the 8th innings. Peak fatigue levels are more than 18% higher when the pitch clock is implemented.
As I have argued in the past – it’s not just injury to worry about. It is assumed that pitchers will perform at the same level when the pitch clocks are in place, when in reality – pitchers who are fatigued are shown to have less command, and less velocity. This would in turn result in more base runners, further extending the game. I hypothesize pitch clocks would have a minimal effect in reducing the time of games, all things considered.
Will the MLB eventually go with Pitch Clocks? I wouldn’t be surprised – it seems like Rob Manfred is hell bent on his crusade to speed up baseball games, turning them into NFL Blitz/ NBA Jam versions of themselves. Are these changes needed? Time will tell – but right now, it seems like those who like baseball, are fine with it in the current state.
Bruce, J. R., & Andrews, J. R. (2014). Ulnar collateral ligament injuries in the throwing athlete. Journal of the American Academy of Orthopaedic Surgeons, 22(5), 315-325.
Sonne, M. W., & Keir, P. J. (2016). Major League Baseball pace-of-play rules and their influence on predicted muscle fatigue during simulated baseball games. Journal of sports sciences, 34(21), 2054-2062.
When you watch football, it’s very clear by body type, what position a player likely plays. Are you 6’5 and 320lbs? That’s more than likely a lineman. 5’9 and 210 lbs? Chances are, you’re a running back. Baseball is a lot different – Marcus Stroman (5’8, 180lbs) and Chris Young (6’10, 255lbs) play the exact same position. Chris Sale (6’6, 180lbs) and Bartolo Colon (5’11, 285 lbs (sure… I believe you)) also play the same position. There aren’t too many times on the gridiron where a 100lb weight difference will line up against each other!
The point I’m trying to make, is with such huge variances in body shapes and sizes, there are many different ways to skin a cat. Marco Estrada (of 89 mph fastball fame) was massively more successful than Joe Kelly (punching a fastball in the high 90s, and over 100 mph). Pitchers of all shapes and sizes (of body and fastball) find ways to succeed. The question is – when you don’t have the clearly obvious advantage of that big fastball, how do you get major league hitters out?
Last week was pitching week over at Baseball Prospectus, and they revealed two new awesome statistics. The first one was Called Strikes Above Average: an attempt to quantify command (Long et al., 2017a). It reflects the ability for a pitcher to hit his spots and get strike calls. The second one was tunneling – a concept I tried to include in the Stuff metric by calculating the resultant distances in pitch breaks, but done in a way that is much more reflective of the demands on a hitter’s executive functioning (Long et al., 2017b). If you have studied Kinesiology, their description of tunneling, and the break:tunnel ratio is a great example of why hitting is so damned hard from a motor control perspective.
It’s pretty easy for someone to look at a pitcher, and say that they have great Stuff. Originally, I created the Stuff metric to try and win bar arguments about pitchers – it’s not necessarily shocking when you see the list of people with great Stuff. Of course Clayton Kershaw has better Stuff than Kyle Hendricks. I felt that combining the Stuff metric with these tunneling and command metrics would help better understand how pitchers of different Stuff levels ended up succeeding or failing at pitching.
The first step was to correlate Stuff, CSAA, and Break:Tunnel with one another. I wanted to see what each metric said about the others – what did the relationship look like?
From this correlation table, you can see that typically those who had better stuff had less command, but a bigger break:tunnel ratio. There weren’t really any significant relationships between command and tunnels, which is great – they’re telling us two different things.
To look at the relationship between outcome measures and Stuff, Command, and Tunnelling, I used the same process as in the last stuff article – just as simple correlation between the three pitching variables, and the outcome metrics of Deserved Run Average, Strikeouts per 9 (K/9), Earned Run Average, xFIP, Ground Ball Rate, and Swinging Strike Rate. To keep it simple at first, I just looked at qualified starting pitchers.
Interestingly enough, Stuff was more highly correlated than command or tunnelling, for all outcome measures. Better Stuff produced a lower DRA and ERA, and a higher K/9, and swinging strike rate. There were negligible relationships with GB% and Stuff. For command and tunnels, there were very few significant relationships with the outcome measures – with there being a slightly elevated relationship between Break:tunnel ratio and K/9. So, from the simple conclusions chapter, I have the following advice for pitchers. Throw hard, and be nasty. There’s no greater individual predictor for success than blasting a ball at 97 mph past a hitter. Go and watch the movie Fastball, and look at the science behind hitting a 100 mph pitch. But then again, we all knew that, right?
So, how do the guys who don’t throw absolute gas survive in the MLB? I think these three metrics provide some really interesting insight into those survival skills.
To put things on the same scale, I divided each metric’s range into thirds. The top third represents the most desired values (higher K/9 and Stuff, lower DRA and ERA). I then calculated the average CSAA and break:tunnel ratio by each third, which resulted in a 3×3 square. I performed this analysis with Stuff along the horizontal axis, and DRA, ERA, and K/9 along the vertical axis. I repeated this for CSAA and Break:tunnel, resulting in 6 squares. To expand this analysis to more pitchers, I set a threshold of 38.1 innings pitched in a season (and included pitchers from 2008 to 2016). The 38.1 innings limit represents the 75th %ile value for innings pitched in a season – 75% of all pitchers pitched at least this many innings in the season.
Table 3. Average Break:Tunnel Ratio, and Command metrics for Stuff, DRA, K/9, and ERA.
I’ll start with the tunneling story. This one doesn’t necessarily cast a whole lot of light on the situation. The best Stuff, and the best tunneling resulted in the lowest DRAs, and the highest k/9’s. The ERA section was even muddled, with the best stuff, and best tunneling resulting in the highest ERAs. That being said, the difference between the lowest tunneling value and the highest average break:tunnel ratio was only 22%.
The Stuff relationship with command is a lot more interesting. For those pitchers who had the least Stuff, but the best ERA, their average CSAA was 160% better than those who had great stuff, and a high ERA. On the flip side, check out what happens when pitchers have great Stuff. In general, those with great stuff don’t have great command – and those who are successful just keep that filth in the zone and let the hitter worry about making contact. This works out the same for K/9, and DRA – those who had the least amount of Stuff and the best results all had much better command. Once again, I don’t think this necessarily is ground breaking information (particularly for me, a known admirer of Marco Estrada, and all of his glory). However, I think this is a great validation of Baseball Prospectus’ command metric – and it has some great insight into pitchers who may age better than others. As players get older, their stuff drops off linearly. At the exact same time, the command values increase. Theoretically, a pitcher with great command would be able to overcome their lack of Stuff as they age.
Figure 2. Average ERA and DRA as Pitchers Age.
A great example of this is Felix Hernandez. Felix started his career throwing as hard as anyone in baseball, but has had his stuff drop off over the past years. Aside from one season where things looked a bit different, the trends for Stuff and Command look very similar to the population graph – a sign that Felix has adapted what he does as a pitcher to continue to be successful.
So, now that we have established a loose relationship between all of these metrics, let’s look at a few lists to help better understand the arms behind the numbers.
These are the pitchers who have it all together. Zack Britton had elite Stuff, elite command, and parlayed this into one of the league’s best ERAs of all time. Kenta Maeda had a breakout rookie season, and Justin Verlander reinforced the theory that he has discovered the fountain of youth.
These guys had great Stuff and great ERA, but they didn’t have elite command metrics. In a sense, these guys offered up a “try and hit this” mentality, and usually ended up blowing hitters away. I’m looking at you, Noah Syndergaard and Aaron Sanchez (oh, what a rotation that would have been in Toronto).
|Robbie Ross Jr.||0.92||-0.50||0.22||3.79|
These are the pitchers that didn’t have great Stuff, but had elite command and used that to have great seasons. It isn’t surprising to see Kyle Hendricks on this list, but to see Masahiro Tanaka here is a bit shocking. His low Stuff in the 2016 season is a far departure to what it was when he broke into the league, but is a testament to his ability to adjust and find new ways to be successful.
In this final list, you see pitchers that put everything together: Stuff, and Command, but didn’t parlay it into great results on the field. For Tigers fans, seeing Jordan Zimmerman on this list is encouraging – perhaps that wasn’t an albatros contract, and he will be back to his previous form in 2017. Sonny Gray battled injury, but if his health returns, he could be back to his old dominant self. Michael “420 Blazek” – shout out to the guys at Brew Crew Baseball who were the first non-Jays fans to ever read something that I wrote – had great stuff and command, but didn’t get great results. Next season, he could be a key piece in the Brewers bullpen.
So, in summary – combining the Stuff Metric with BP’s Command and Tunnels provides some insight into how pitchers get hitters out. These new metrics can be used to help look at how pitchers age and change their strategies to forge long careers.
Long, J., Judge, J., & Pavlidis, H., (2017a). Prospectus Feature: Introducing Pitch Tunnels. Retrieved from: http://www.baseballprospectus.com/article.php?articleid=31030, published January 24, 2017.
Long, J., Judge, J., & Pavlidis, H., (2017b). Prospectus Feature: Introducing Pitch Tunnels. Retrieved from: http://www.baseballprospectus.com/article.php?articleid=31022, published January 23, 2017.
I wanted to start this (the final post of 2016) off on a sappy note. This has been the first year that I’ve been involved in writing about baseball from a scientific perspective. When I started my PhD at McMaster, there was a department seminar meeting, which focused on science and knowledge translation. The speaker, who was a world leader in her field, told us that on average – research takes 14 years to make it from the lab to the field. I made it my focus to try and reduce those numbers. While my PhD had an ergonomics application, I have really enjoyed trying to tie the worlds of baseball and ergonomics together. Fatigue Units, and the fancy fatigue graphs, and the paper arguing against pitch clocks, all use the model I developed for my PhD thesis. If you have read anything I have written this year, I am eternally thankful! Interacting with people through Twitter, email, and this blog, has been one of the most fun and rewarding experiences I’ve had as a scientist. I really want to thank Dr. Stephen Osterer, Dr. Michael Chivers, Tavis Bruce, Kyle Boddy, Michael O’Connell, Trip Somers, Kevin Kennedy, and Eno Sarris for helping me get my ideas out there. You guys know so much more than me about baseball, but thanks for listening to me anyway!
Alright, enough of that – time for nerd stuff.
I’ve had a chance to look at Fleisig’s most recent paper 1 on arm forces during weighted ball training. This paper is a very important step towards determining the safety of a very effective method for pitcher development. If pitchers are getting better, but increasing their risk of injury – what’s the point in throwing harder? At the same time, maybe throwing these weighted implements serves as a unique stimulus for muscle growth and arm health. A lot of questions are still out there waiting to be answered, but this is an important first step.
So, let’s look at things as simply as we can. One of the first equations you learn in physics, or in the application of physics to the human body – biomechanics – is that force is equal to mass times acceleration.
F = M*A
One of the main findings of this study, was overweight baseball throwing produced lower forces on the arm, primarily, lower varus torques when compared to normal weight baseballs. With respect to injury prevention, and specifically, Ulnar Collateral Ligament tears, this is a good thing! Lower varus torques mean less strain on the ligament, and a reduced risk of tearing when compared to the normal weight baseballs.
How does throwing something heavier result in less force in a joint? It all comes back to F = M*A. If we keep acceleration constant, and want to decrease force, we’d need to decrease the mass of the segment (and implement) being rotated. Similarly, if we wanted force to be lower, but increased the mass, we’d need to significantly reduce the acceleration of the segment. From table 1 in the Fleisig paper, we can see that the accelerations of the shoulder, elbow, and pelvis are all lower in the heavier ball conditions when compared to the underweight, and overweight baseballs.
However, the overall varus torque on the elbow, and the directly transmitted torque to the ligament may not be the same thing.
Werner et al., (1993) 2, illustrated that torques incurred at the elbow during the pitching motion are greater than the known torsional torques required to rupture the UCL. Muscle activation is used to transmit these forces off of the ligaments, and prevent damage to the UCL. This is why it is dangerous for the flexor-pronator muscles to become fatigued, and lose their ability to produce muscle force during pitching (Bruce & Andrews, 2014) 3.
Joint rotational stiffness refers to how rigid a joint is while it is moved 4. In the human body, stiffness is controlled by increasing the co-contraction of muscles spanning a joint. For example, activating the flexor muscles of the elbow (the biceps), as the elbow is rapidly extending (using the triceps). A stiffer joint (with more muscle activation) theoretically protects the ligaments by transmitting rotational torques from “passive” tissues (those tissues which are not contracting), to the “active” tissues (the muscles which are contracting). However, by increasing stiffness of a joint, the body loses the ability to produce high rotational accelerations and velocities.
Think about it as a whip – a thicker piece of rope can not rotate as quickly, and snap as violently as a thinner piece of rope. At the same time, the thinner rope is at a higher risk of fraying and breaking than a thicker rope. In the body, a joint with less stiffness can move faster, and more smoothly than a joint with more stiffness.
When throwing a heavier implement as hard as you can, there has to be more joint rotational stiffness in place to protect the ligaments. However, as the motion becomes more natural, and more familiar to the thrower, stiffness will decrease. This will result in a ball velocity increase, and possibly, a re-mapping of muscle activation patterns for throwing the ball when returning to a normal weight ball. If you want to get deeper into this, and the concept of stability, stiffness, and joint rotational impedance, check out the work from some really smart guys – Mike Holmes, and Joshua Cashaback.
Does throwing a weighted ball decrease the total varus force at the elbow, but increase the amount of force that is actually transmitted to the ligament? For the sake of arguments, let’s say this is true. How do pitchers manage to stay healthy, while throwing harder? One of the possible ways to protect the UCL is to make sure that muscles are able to contract, and reduce the amount of force transmitted to the ligament. Putting in work in the weight room, and specifically, training the forearm muscles, could serve to cause hypertrophy, increasing the physiological cross sectional area of the muscle and allowing the muscle to pull with the same amount of absolute force, but at a lower effort level (associated with less muscle activation).
Research has shown there is very little effectiveness in restricting innings in pitchers with respect to preventing injury (Karakolis et al., 2016 5 ). Other research has shown that pitching velocity (and “Stuff” – sorry, shameless self-promotion) has increased linearly, year after year.
At the same time, the number of Tommy John Surgeries has also increased every year (particularly in younger pitchers) (Keri, 2015 6). Some people out there, who get their jollies on hoping people get hurt, so they can be proved right, even call this an epidemic.
(Figure from Keri, 2015).
So, let’s bring my crazy, mad scientist, aluminum foil hat-wearing hypothesis full circle. Back to this theory on stiffness – pitchers are finding ways to shut off their muscles through different training regimes – yes, even maximizing your throwing intent to throw as hard as humanly possible, could play a role in reducing co-contraction and joint stiffness. Over the long term, could this be causing a re-mapping of the muscle firing patterns, reducing stiffness in game-thrown pitches, exposing the UCL to greater stress? Furthermore, do strict innings limits, like the ones that Matt Harvey and Stephen Strasburg faced, lead to a de-training effect? Pitching is a unique motion – one that is difficult to replicate in a gym. As a result, could there be atrophy occurring in the forearm musculature due to reductions in throwing? Fleisig’s paper hypothesized that the “holds” drill, and throwing of weighted baseballs, could be an effective resistance training strategy. Now imagine, hypertrophy occurs, allowing for protection of the UCL during high velocity throws. This happens in the offseason, when pitchers are preparing for their next year of competitive baseball. They throw frequently, with the intent of getting better. Now, they enter the MLB season, and arbitrary limits are placed on their pitching. To stay in the major leagues, they have to keep their velocity high – but now, they have lost some of that muscle mass they grew during their training. When they return to throwing – the velocity remains the same, but the protective mechanisms are now reduced.
This isn’t meant to be a stance for or against weighted balls. I don’t coach baseball – I’m a fan of the game, and truly enjoy applying what I know about how people move to the game I passionately watch. Keep in mind, this post is a perspective – it is not tested, or even extensively researched. Selfishly, I like writing these things, because there are people out there who have the access to pitchers, equipment, and can answer these types of questions. Consider this – a challenge point to unlock more information about how the body works. If you are using this as some form of gospel to take a stand against weighted balls – don’t do that. Don’t be a dick.
So, that’s my theory on how using weighted baseballs for training, and pairing it with workload restrictions in season, could lead to an increase in UCL injuries.
It’s either that, or it’s aliens.
I’ve done a bunch of work modelling fatigue in pitchers, including the study used to suggest Pitch Clocks would lead to increased fatigue 1 , how pitches per inning influences fatigue levels 2 , and a new workload metric based on fatigue 3 . For a lot of these simulations, I used average pitch counts and paces from FanGraphs. This provides an overall picture of the estimated fatigue in a pitcher, but doesn’t really account for the individual variability associated with a game. I was originally trying to get a bigger picture idea of how fatigue accumulated during pitching – so this was appropriate. However, I missed out on some pretty cool looking graphs – which depending on who you talk to, is the entire point of science.
Last week, I tweeted out a few graphs that looked at predicted forearm fatigue in pitchers from games during the 2014, and 2015 season. Here was the first one:
Playing around with data visualization – Marco Estrada’s predicted forearm muscle fatigue during his game 5 gem from the 2015 ALCS. pic.twitter.com/fWqH0JaqWm
— Mike Sonne (@DrMikeSonne) December 2, 2016
In game 5 of the ALCS in 2015, Marco Estrada went 7 and 2/3 innings, staving off elimination and sending the series back to Kansas City. Estrada started out fast, and never gave up a run until the 7th inning. It was a sign of things to come – absolute clutch pitching performances that buoyed the Blue Jays pitching staff in the 2016 season. Here’s how I was able to get these simulations.
The first step in this method was to get the pitch types, sequence, and time of each pitch during an outing. From the MLB’s gameday data, I was able to pull this sequence out by querying the database by pitcher and game. Here’s the MySQL Query I used for that purpose:
select pitches.sv_id, atbats.inning, atbats.outs, pitches.pitch_type, atbats.des, atbats.pitcher from atbats, pitches, games where pitches.ab_id = atbats.ab_id and atbats.game_id = games.game_id and games.game_id = ‘23910’ and atbats.pitcher = 462136 order by atbats.inning, atbats.outs ASC
This produced a result that looked like this:
Figure 1. Sample output from the MySQL query posted in the snippet above.
Every pitch, organized by inning and sequence, for an individual pitcher in an individual game. That value on the left – the sv_id from the gameday atbats table, contained the date (YYMMDD_HHMMSS). Everything I needed to calculate the amount of time between each pitch.
From the paper on pitch clocks and fatigue, I had created a time history of muscle demands for a series of forearm muscles. These were defined as either fastball type pitches, or breaking type pitches. In the graphs, figure A are the demands for fastballs, and figure B are the demands of curve balls / breaking pitches. These time histories are from a series of studies, summarized in the pitch clock papers.
So, back to the Gameday database. I exported the demands from the query into Matlab. I coded each pitch as being either a fastball or a breaking ball, then attached the appropriate amount of rest after each pitch. Here’s an example from Marco Estrada’s ALCS game:
Figure 3. Pitch selection and rest time from Estrada’s game 5 of the 2015 ALCS. All pitches coded as a 1 were fastballs. All pitches coded as 2 were breaking pitches. These were used to simulate the time histories of demands for Estrada, and then determine the predicted fatigue level.
Figure 4. Muscle demands by pitch during a 12 pitch first inning. The first pitch of the game lead to a ground ball out. The next 7 pitches resulted in a strike out, followed by a 4 pitch strike out. You can see the larger gaps between pitches as being reflective the time between batters.
The largest numbers in this time history represented the time between innings. What I have learned from this is, is just how fatiguing it can be for a pitcher to have a long inning, followed by a short half inning where his team goes three up three down.
At this point, I had everything I needed to predict fatigue in a pitcher. Using the EMG demands from figure 4, the three compartment fatigue model was able to generate an example of how much muscle force would be lost at a given time based on the previous history of muscle demands. For Estrada, his peak fatigue came during the 5th and 7th innings of the game. During these innings, he threw 14 and 17 pitches – the highest per inning in the game. In the 5th inning, he threw more fastballs than breaking pitches, which resulted in higher fatigue levels.
Interestingly. innings where pitchers throw more fastballs have higher fatigue levels. Greater homogeneity in pitch selection has been identified as a risk factor for UCL reconstruction by Whiteside and colleagues (2016) 4 have higher fatigue levels occur when pitchers throw more fastballs.
This methodology could now allow for examination of fatigue during starts where a pitcher was hurt, or from a series of starts where a pitcher became hurt. If at the start of the season, a pitcher had his EMG activation levels recorded during a bullpen session, these accurate demands could be substituted into the model for a more accurate fatigue prediction. For now – this is a fun new way to look at a pitching performance.
If you have any requests for simulations, let me know!
A lot was made of the heavy workloads that elite relievers performed in the 2016 post season. Aroldis Chapman, Andrew Miller, and Roberto Osuna had extended appearances of 2+ innings, when they had only thrown an inning at a time in the regular season. With everything on the line, teams were willing to push these elite arms to their limits. For the most part (save for one scary moment in the AL Wild Card game with Osuna), they all emerged unscathed, and probably went a long way to raising reliever salaries for future seasons.
By now, you have probably learned that my Fatigue Unit metric heavily favours relief pitchers. A study by Whiteside and colleagues (2016) 1, and my analysis of Driveline’s publically available data 2,3, have indicated that throwing on consecutive days, and throwing harder, are risk factors for Ulnar Collateral Ligament reconstruction. Relief pitchers are throwing more and more innings, and their rate of injury appears to be much greater than those of starting pitchers 4. These are the hard facts of relief pitching – see examples from Aaron Sanchez (94-95 mph as a starter, and 97-100 as a reliever).
Using Fatigue Units5, I wanted to look at the top workloads in the past post season, and furthermore, explore how those workloads compared to regular season workloads.
I was able to analyze the 2016 MLB regular season thanks to Michael Copeland (@jelloslinger). He pulled in pitch counts, by inning and game, for every game since the start of the 2015 season. For simplicity’s sake, I looked at the peak velocity and the average pace for all pitchers, downloaded from the Fangraphs website. Those were the data required to calculate FU’s for each pitcher. I then broke these data down by regular season games and post-season games.
Once cumulative fatigue units were calculated for each pitcher, I also calculated the average workload per game. The rest of the analysis was performed on pitchers who only pitched in both the regular season and the post season. I compared the average workload between regular season and post season performances. For an example, I also looked at how workloads changed throughout the season, and I’m going to present the data on how workloads changed throughout course of the season for a few select pitchers.
Results and Conclusions
Not surprisingly, the highest workload of the 2016 playoff season belonged to Andrew Miller and Corey Kluber. For Andrew Miller, the biggest change in his workload between the regular season and the post season, was his multi-inning appearances. In the playoffs, he averaged 0.8 FU’s per game, and in the regular season, he average 0.4 FU’s per game. His workload per game was doubled in the playoffs when compared to his regular season numbers. Andrew Miller’s playoff workload represented 25.2% of his entire regular season workload – during 10 games (compared to 70 games in the regular season).
As for Corey Kluber, his high workload was driven by short rest. Kluber pitched game 1, 4 and 7 of the World Series in 2016. In the regular season, Kluber pitched 100% of his starts on regular rest (a 5 day separation, minimum, between starts). During the playoffs, 50% of his appearances came on less than 5 days of rest. Kluber averaged 1.23 FU’s per appearance in the 2016 playoffs, compared to 0.74 FU’s per appearance in the regular season.
Going to game 7 of the World Series, the Cubs had the greatest sum of fatigue units of all 2016 playoff teams. However, when looking at sum of fatigue units for every playoff teams, the Cubs accumulated 147% more fatigue units than the second highest team (the Toronto Blue Jays). When looking further into these findings – what drove these really high FU accumulations in the Cubs and Blue Jays, were a very high number of back to back relief appearances for both teams. The Cubs had 35% of all of their pitching appearances in the post season occur on back to back nights, a total of 26 pitcher appearances in the 2016 post season. Comparatively, Cleveland only had 11% of all of their pitching appearances occur on back to back nights.
I’m not one for hot takes, but I feel that I have to say something here. If workload has any sort of relationship to injury, Terry Francona’s management of his pitching staff, from a workload management perspective, is one of the most impressive baseball strategy events to occur in my recent memory. While the Cubs emerged the eventual World Series Champions, Joe Maddon had to repeatedly rely on the same pitchers (primarily, Aroldis Chapman) in high leverage situations, which lead to high usage during back to back games. If extreme workloads to lead to injury, the work done by Francona in the playoffs does appear to be a great template for managers in the future. If we’re being realistic though – flags fly forever, and when they haven’t flown in 108 years, it’s hard to say that there were any errors in workload management.
I still have to look further into the utility of FU’s as an injury prediction model. Currently, I believe it is a metric that more accurately represents physiological demands on the body, and more importantly, the musculature responsible for predicting the UCL.