Early Season Pitcher Workloads
Traditional pitching metrics, such as innings pitched, and pitch counts, have often missed the mark when it comes to preventing pitcher injuries. As a result, I developed the Fatigue Units metric – which shows promise in illustrating how extreme workloads can influence pitchers in the subsequent seasons.
As a quick refresher – Fatigue Units are calculated by looking at an interaction between the number of pitches thrown, the velocity they are thrown at, the time taken between pitches, and the number of days between appearances. In the 2015 and 2016 season – these were your FU leaders.
| Rank | Name | 2015 | 2016 | Total |
|---|---|---|---|---|
| 1 | Travis Wood | 24.48 | 20.13 | 44.61 |
| 2 | Dellin Betances | 24.13 | 20.15 | 44.28 |
| 3 | Chris Sale | 21.92 | 21.51 | 43.43 |
| 4 | Max Scherzer | 20.38 | 20.16 | 40.54 |
| 5 | Chris Archer | 21.18 | 18.93 | 40.11 |
| 6 | Johnny Cueto | 21.85 | 17.92 | 39.77 |
| 7 | Jeurys Familia | 21.04 | 17.97 | 39.02 |
| 8 | Yordano Ventura | 19.49 | 19.24 | 38.73 |
| 9 | Jake Arrieta | 21.70 | 16.55 | 38.25 |
| 10 | Randall Delgado | 19.26 | 18.71 | 37.98 |
| 11 | Roberto Osuna | 18.00 | 19.82 | 37.82 |
| 12 | Cole Hamels | 19.93 | 17.57 | 37.50 |
| 13 | Brad Brach | 18.14 | 19.15 | 37.29 |
| 14 | Zach Duke | 17.12 | 19.84 | 36.97 |
| 15 | Addison Reed | 15.54 | 21.17 | 36.72 |
| 16 | David Price | 19.45 | 17.22 | 36.67 |
| 17 | Erasmo Ramirez | 17.74 | 18.83 | 36.57 |
| 18 | Hector Santiago | 19.95 | 16.60 | 36.55 |
| 19 | Kyle Barraclough | 15.99 | 20.50 | 36.48 |
| 20 | Madison Bumgarner | 18.35 | 18.03 | 36.38 |
Looking at the 2017 data for highest workloads, some pitchers have already starting racking up the fatigue units. Let’s dive in to see who has the highest workloads, and what teams appear to be mitigating those effects.
| Rank | Name | Fatigue Units | Days Between | Back to Backs | IP | Total Pitches | Game Apps | Pace (s) | Start IP | Relief IP |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Felipe Rivero | 6.20 | 2.00 | 8 | 19 | 297 | 19 | 21.6 | 19 | |
| 2 | Enny Romero | 5.71 | 2.62 | 5 | 15.1 | 276 | 15 | 24.5 | 15.1 | |
| 3 | Brad Brach | 5.57 | 2.06 | 7 | 18.2 | 297 | 18 | 23.8 | 18.2 | |
| 4 | Fernando Salas | 5.31 | 2.06 | 7 | 15.1 | 287 | 18 | 25.9 | 15.1 | |
| 5 | Danny Farquhar | 5.18 | 2.18 | 6 | 15 | 288 | 18 | 23.1 | 15 | |
| 6 | Bud Norris | 5.06 | 2.25 | 5 | 18 | 306 | 17 | 27.6 | 18 | |
| 7 | Anthony Swarzak | 4.96 | 2.09 | 5 | 14.2 | 195 | 12 | 25.3 | 14.2 | |
| 8 | Jhan Marinez | 4.89 | 2.77 | 5 | 14 | 274 | 14 | 26.7 | 14 | |
| 9 | Joe Kelly | 4.87 | 2.62 | 2 | 17 | 290 | 14 | 27.8 | 17 | |
| 10 | Yusmeiro Petit | 4.82 | 2.91 | 3 | 19 | 296 | 12 | 24.5 | 19 | |
| 11 | Chris Sale | 4.81 | 5.33 | 0 | 51.2 | 757 | 7 | 20.4 | 51.2 | |
| 12 | Tyler Clippard | 4.78 | 2.27 | 6 | 14.1 | 227 | 16 | 27 | 14.1 | |
| 13 | Josh Smoker | 4.71 | 2.29 | 2 | 16 | 311 | 15 | 22.4 | 16 | |
| 14 | Seung Hwan Oh | 4.66 | 2.64 | 6 | 16.2 | 285 | 15 | 22.9 | 16.2 | |
| 15 | Brian Duensing | 4.64 | 2.27 | 4 | 14 | 210 | 12 | 22.9 | 14 | |
| 16 | Miguel Diaz | 4.61 | 2.57 | 3 | 15.1 | 260 | 15 | 24.8 | 15.1 | |
| 17 | Daniel Hudson | 4.59 | 2.25 | 5 | 14.1 | 275 | 17 | 27.6 | 14.1 | |
| 18 | Joely Rodriguez | 4.58 | 2.43 | 4 | 17.2 | 287 | 16 | 23.3 | 17.2 | |
| 19 | Chris Devenski | 4.56 | 3.40 | 2 | 21 | 328 | 11 | 21.2 | 21 | |
| 20 | Hansel Robles | 4.53 | 2.25 | 5 | 18.1 | 302 | 17 | 24.4 | 18.1 |
Felipe Rivero leads the MLB in workload so far, driven primarily by an astonishing 8 back to back appearances in 19 games. That is not a lot of time for recovery! He averages an outing once every 2 days – also the shortest in the MLB. It’s easy to see why though – he has an ERA under 1, on a team that has a historic number of blown leads so far. There’s a reason why certain pitchers can get to high workloads – they’re good.
Brad Brach is also sky rocketing up the workload charts with the injury to Zach Britton – and has appeared in 7 back to back games. These are very high workloads early in the season, but should tend to decrease as their respective teams either start winning games big, or losing games big.
| Rank | Team | Fatigue Units | Days Between | Back to Backs | IP | Total Pitches | Game Apps | Pace (s) | Start IP | Relief IP |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Mets | 48.97 | 3.54 | 41 | 293.6 | 5052 | 159 | 22.93 | 173 | 119.9 |
| 2 | Pirates | 44.04 | 3.89 | 21 | 291.1 | 4776 | 124 | 23.20 | 182.2 | 108.9 |
| 3 | Cubs | 43.57 | 3.70 | 18 | 305.2 | 5268 | 147 | 22.52 | 176.4 | 128.8 |
| 4 | Angels | 43.36 | 3.93 | 18 | 310.1 | 5128 | 142 | 24.69 | 191.8 | 118.3 |
| 5 | Padres | 42.49 | 4.01 | 16 | 291.9 | 4788 | 134 | 22.56 | 184 | 106.5 |
| 6 | Diamondbacks | 42.40 | 3.66 | 19 | 296.5 | 5057 | 133 | 23.72 | 194.4 | 102.1 |
| 7 | Marlins | 41.73 | 3.79 | 21 | 284.2 | 4864 | 140 | 23.61 | 160.6 | 123.6 |
| 8 | Rockies | 41.53 | 3.79 | 14 | 304.2 | 4935 | 143 | 23.44 | 191.4 | 112.8 |
| 9 | Cardinals | 41.00 | 3.84 | 19 | 288.5 | 4798 | 137 | 22.96 | 188.5 | 100 |
| 10 | Orioles | 40.73 | 4.23 | 23 | 286.2 | 5027 | 129 | 24.13 | 173.7 | 112.5 |
| 11 | Rays | 40.23 | 3.90 | 23 | 309.7 | 5163 | 142 | 26.26 | 197.3 | 112.4 |
| 12 | Blue Jays | 40.18 | 3.85 | 17 | 296.4 | 4931 | 146 | 24.46 | 178.6 | 117.8 |
| 13 | Brewers | 39.98 | 3.84 | 26 | 279.7 | 4862 | 139 | 24.81 | 169.9 | 109.8 |
| 14 | Dodgers | 39.96 | 4.03 | 19 | 292.2 | 4755 | 138 | 25.92 | 183.5 | 108.7 |
| 15 | Nationals | 39.85 | 3.80 | 16 | 296 | 4985 | 132 | 24.09 | 201.1 | 94.9 |
| 16 | Giants | 39.78 | 3.84 | 18 | 297.6 | 4893 | 139 | 24.34 | 200.5 | 97.1 |
| 17 | Mariners | 39.76 | 3.74 | 15 | 291.2 | 4745 | 142 | 23.69 | 175.9 | 114.6 |
| 18 | Athletics | 39.16 | 4.20 | 15 | 287.7 | 4825 | 131 | 25.02 | 175.8 | 111.9 |
| 19 | Reds | 39.04 | 4.19 | 8 | 292.5 | 4906 | 131 | 21.89 | 162.4 | 130.1 |
| 20 | Rangers | 38.51 | 4.17 | 15 | 298.3 | 4972 | 132 | 24.49 | 198.9 | 99.4 |
| 21 | Yankees | 38.11 | 4.21 | 16 | 280.5 | 4525 | 123 | 23.54 | 183.6 | 96.9 |
| 22 | Astros | 37.98 | 4.06 | 14 | 297.6 | 4804 | 127 | 24.72 | 194.8 | 102.8 |
| 23 | Indians | 37.38 | 4.13 | 17 | 277.5 | 4610 | 121 | 23.27 | 186.7 | 90.8 |
| 24 | Royals | 37.20 | 4.05 | 16 | 286.5 | 4774 | 132 | 22.57 | 180 | 105.8 |
| 25 | Red Sox | 37.07 | 3.91 | 13 | 282.1 | 4787 | 128 | 25.30 | 186.4 | 95.7 |
| 26 | Phillies | 35.07 | 4.10 | 16 | 274.5 | 4591 | 127 | 23.74 | 169.6 | 104.9 |
| 27 | Twins | 34.92 | 4.15 | 14 | 263.5 | 4417 | 132 | 23.80 | 158.6 | 104.9 |
| 28 | White Sox | 34.78 | 4.03 | 15 | 268 | 4458 | 115 | 25.17 | 176.7 | 91.3 |
| 29 | Tigers | 33.70 | 3.66 | 15 | 269.7 | 4823 | 125 | 25.74 | 178.5 | 91.2 |
| 30 | Braves | 33.59 | 3.51 | 18 | 266.4 | 4293 | 127 | 22.78 | 169.8 | 96.6 |
When you look at the teams with the highest workloads, something very dramatic jumps out at you – the Mets already have 41 back to back appearances out of their bullpen. The next closest? The Brewers, with 26 back to back appearances. This is a landslide! Quite honestly, it brings to question the management of the bullpen – and as far as risk factors go, they have the second shortest time between pitching appearances – 3.54 days between appearance.
Of course – this is speculation when it comes to evaluating bullpens – particularly in the early going. Bad performances by the bullpen can really tax the pitchers who are performing in the early onset of the season – take for example, Felipe Rivero this year, and Roberto Osuna/ Joe Biagini last year. Skippers will turn to the guys who can get them the outs they require to finish off games. Over a long season – this tends to balance out. If not? Expect massive workloads on these high performing arms – but don’t be surprised if they end up on the DL in the coming seasons.
Ergonomist (CCPE) and Injury Prevention researcher. I like science and baseball - the order depends on the day. Twitter: @DrMikeSonne
This is great stuff, thanks.
Thanks for reading!
Your first sentence implies that some future metric will prevent pitcher injuries. I disagree wholeheartedly. That list of players with high FU (love the name) appears to be healthier than a random sample of 20 arms that actually accrue innings. I appreciate your work, and understand what you are going for, but the idea that a formula will prevent injuries is just wrong. Some players are just waiting to break and some players are not going to break. Sure, there is some useful data in between but these are humans that you are attempting to measure. I think it is extremely important to keep that in mind – more important than any exciting correlations that you may find. Regarding your last sentence, if you expect everyone to hit the DL in coming seasons, you will be correct more often than not!
If someone ever tells you they have created a single metric, or identified a single risk factor that predicts injury, you can tell them they are a f’ing idiot. That kind of reductionist thinking is why people are making dumb decisions based off of innings limits or 100 pitch counts. It doesn’t work.
The purpose of creating FU’s was to identify a method of tracking workload that is more effective than the traditional metrics of pitches, or innings pitched. My findings were, in extreme workloads, pitchers were more likely to have Tommy John Surgery. If you look at the original piece – these incidence rates are still fairly low. The point is, relief pitchers are at just as high of a risk of having injuries, if not higher, than starting pitchers. This workload metric does a better job of capturing the hazards they face, and it is based off of peer reviewed research.
Jeurys Familia is another hit.
I think single metrics are fine, like 100-pitch counts. They are flawed, just like more complex analysis has flaws. The problem is that people take “complex” analysis more seriously than they should. It is OK to dismiss simple data… As long as guys are throwing baseballs, they will be at increased risk of TJ. It probably makes sense to evaluate RP differently than SP.
I was just thinking about how you could best measure this type of wear and tear yesterday as Lance Lynn threw consecutive 30 pitch innings in the first and second inning of his start.
You could have “bad” innings and “good” innings. Good innings would have no base-runners. Medium innings would have base runners. Really bad innings would exceed 30 pitches… etc. That is overly crude, but you could easily define these buckets. You could do some math, create some ratio and multiply that by a starter’s inning count to create a more meaningful IP stat. This only works for starters because they all work off of the same rest. For RP, honestly, they should be treated as disposable more or less. The only RP that last are the ones that can succeed and work at less than 100% (Kenley) or the few freaks (Aroldis).
Everyone here should have a quick Google of Rick Peterson (Mets pitching coach) and prehab. He sold data and technology as a means of preventing pitcher injury. He revolutionized the game – or at least that is what he sold. That was well over a decade a ago and we have clearly made zero real progress, but that didn’t stop the excitement! Half the guys Peterson worked with were on the brink of being out of baseball. I get that guys throw harder today and the game is not exactly the same, but the idea that data solved any arm-health issues is pretty questionable. I am just pointing out that this is not a new idea and it hasn’t proved particularly valuable in the past. It is an attractive idea and I can see why people want to buy it. Some people have made a living from selling it – not accusing you of that at all. I think you are just doing research, which I think is great if you enjoy it.
I 100% am doing research… Sorry if it came off as anything else. That’s kinda what I do, haha. I’m not sure if you saw the list of all time (since 2008) workloads, but there’s more than a few hits on there as well…
http://www.fangraphs.com/fantasy/an-introduction-to-fatigue-units-a-new-method-for-evaluating-workloads/
Workload monitoring is probably more about your acute to chronic ratio than it is about 1 single value. If I went out and threw 60 pitches at max effort, there’s a really good chance I hurt myself. If I worked up to throwing 60 pitches at max effort over a month – that likelihood goes down. And about the long innings – I completely agree. I had written about that before too – http://www.mikesonne.ca/baseball/fatigue-inferences-on-a-100-pitch-limit/.
Curious to see what the list looks like of good/bad innings.
Thanks for the reply! Please don’t ever apologize to me – you certainly don’t owe me anything. I do appreciate your work. I stir the pot, as one of the few people around that have a legitimate interest in both the art and the science of baseball. I have a life of baseball experience and I have always been fascinated by the data as well. I literally quit a baseball coaching career to make a living in computer science and mathematics. Unfortunately, too many people place their faith in one or the other, but I can tell you that they are not mutually exclusive. I can tell that you get that, but some people don’t.
I assure you that I am not going to do any research, but I am full of ideas! I think binning is the best way to go. That is how defense is evaluated, no? 95%, 50% chance etc. Its not perfect, but any shades of gray, as opposed to black and white, are a win.
Not surprising the METS lead this list. I think Terry has misused his bullpen plus the METS starters have imploded more than most teams starters have, (I have no stats to back that statement up).
This reminds me, what happened to the MASH and/or injury reports that we previously had? I remember BJ Maack had taken over them from Jeff, but looks like he hasn’t posted since September, nor has their been anything on that front.
While the 2015/2016 FU leaders were split about 50:50 between starters and relievers, 2017 only has one starter in the top 20. Is this something that should even out as the season goes on, or is there something new going on here?
Most definitely, this will start to even out.
I guess I should be looking to move Chris Sale before he breaks, then?
Yes (and if you’re in my fantasy league, you should trade him to me without much in return).
Justin Verlander had insane workloads – but stayed relatively healthy. It’s just worth noting that for the number of innings Sale pitches, they’re relatively high workload innings. That comes with the territory of striking out a ton of batters.
Interesting that he is the only starter on the 2017-only list.
If you check back in a couple of months, that will change for sure.
I loved the original article and appreciate the original research but I suggest you reconsider your conclusion: “Don’t be surprised if (pitchers with massive workloads) end up on the DL in the coming seasons.”
Your initial study, which you linked, concluded the following: “Approximately 6.5% of the pitcher seasons that produced a 90th %ile workload (using fatigue units) resulted in Tommy John Surgery in one of the next two seasons.” That compared to 5.1% using innings pitched as a proxy for workload.
But even if they results from two years of data turn out to be repeatable, they do not support that notion that we should expect injuries for pitchers with the highest workload as measured by fatigue units. According to your own data, 93.5% of pitchers in the 90th %ile workload will NOT require Tommy John Surgery in the subsequent two seasons. Using your sample size of 20 pitchers, that means we should expect 1.3 of those 20 to require TJ surgery (as opposed to 1.02 if we use innings pitched).
Perhaps you have done subsequent research using fatigue units that try to link that metric to appearances on the DL, but if you have, I have missed it.
But absent additional research, I don’t think your conclusion is supportable. If even two pitchers in your list of 20 require TJ surgery the next two years, that would be more than your model predicted.
I should point out – I’ve isolated the top 20 in both of those situations. This was run on far more – it was calculated for everyone that threw a pitch in any of those seasons tested.
I haven’t looked at anything other than TJS so far – but that’s to come. If I do go down that road, I’m going to take it a bit further and go towards a journal submission.
I wouldn’t say this is a predictive model either – in the current testing, I’m really only looking at counts and means. As I mentioned above, I believe this workload metric is a better way of looking at workload in all pitchers – not just starters, or relievers.
I look forward to your further research and hope it gets published.
I just also would like to point out -it isn’t 6.5 percent of pitchers would suffer a UCL tear – it’s pitcher-seasons. Each season was treated as a separate entity – definitely a limitation of the analysis.
Whoa the Cubs have thrown 20%/1000 more pitches than the Braves. That’s a pretty wide variance
18 inning games can’t help that situation!
No doubt Collins mixes and matches on the excessive side, but if you further analyze the above chart you will see that the Met starters have only pitched 59% of the teams total innings. That is the 27th lowest in all of baseball.Add that to the fact that Salas, Robles, and Montero have been a horror show there is no question why Terry Collins keeps reaching for the same guys.