Wednesday night in the 11th hour, MLB owners and players agreed to a new collective bargaining agreement that will cover five seasons through 2021.  While many of the items eventually agreed upon were tweaks and not major overhauls, one of the items that was of interest to me was the reduction of the disabled list from 15 days to 10 days.  On the surface, this could look like a win-win for both the players and the owners.  After all, players get to come off the DL and back on to the playing field five days sooner than they would have in past seasons, and owners can save coaches and fans from having to watch replacement-level players play while a most likely better player is on the shelf.

Using DL data compiled by baseballheatmaps.com, I took a look at length of stay on the DL by all players who landed on the list from 2010-2016.  Since 2010, 319 players have spent exactly 15 days  on the DL.  In total, this is 4785 days spent on the DL in seven seasons.  Now, for fun, let’s assume those same 319 players were ready to go after the new minimum of 10 days on the DL.  Simple math here will tell you those players spent 3190 days on the DL.  So in theory, over the course of seven seasons, reducing the DL to 10 days could save players 1595 days on the DL and owners the same number of days using most likely replacement-level players.  On a per-team average basis, reducing the DL by five days could actually save a team 7.6 days of DL time.

Seems like a win-win, right?  Again, players come back sooner, GMs don’t have to call up as many players from the minors and burn options, and owners save money by not having as many extra players come up from the minors accumulating MLB service time.  Not so fast.  In the same seven-season stretch, 3324 players spent 15 days or more on the DL and only 319 came off after 15 days.  So only 9% of all players on the DL spent the minimum amount of time out of action.  Why would this be?  Well, the obvious answer is if a player is hurt, they are hurt.  No one knows a player’s body better than the players themselves and they will return to action when they feel they are ready.

But the other answer is it pays to be on the DL in the majors.  There is protection.  Players still earn their salary and collect service time, so why rush back from an injury?  In the minors it is a different story. If you get hurt it becomes the next man up for a promotion to the big leagues.  There’s a reason there is a saying in the minors: “you can’t make a club in the tub.”  Now, just because there is protection doesn’t mean players want to spend time on the DL.  If they could, they would spend no time on the DL, as time away from the playing field can hurt future earning potential. Injuries are an inevitable part of the game but most seem to prevent players from feeling they are healthy enough to come back sooner than 15 days to compete at their best.  By reducing the DL to 10 days, I can see increased pressure from fans and media to come back quicker.  What we have to remember is this is the new minimum.  Players will return when they and the medical staff feel they’re ready.  I wouldn’t give your hopes up to see players return from the DL sooner than they have in the past.

“Make baseball fun again” is Bryce Harper‘s outcry against baseball fundamentalists who continue to police emotions and enforce baseball’s expressionless professionalism.  “Shut up and play the game right” might be something you’d hear uttered from the fundamentalist’s side — ideally through tobacco-glazed teeth — and maybe by Brian McCann.  The discourse is of course more involved than that, covering everything from retaliatory plunk balls to bat flips, and anytime something marginally inflammatory happens, it’s beaten so hard that we’re reminded how boring our lives are that we have to discuss the same things over and over and over.  I know you can picture the media package that accompanies the discourse: a young, brash, exquisitely coiffed, generational talent, who was hit in the ribs in his first ever plate appearance (then proceeded to steal home), is unabashedly passionate about a “fun” revolution in baseball.  His eye black is adorned like war paint; he has emojis on the bottom of his bats; his helmet never stays on his head when he runs the bases; and yes, he “pimps” his home runs.  Cut to Joey Bats‘ ALDS bat flip and the ensuing brawl and then connect it with Rougned Odor’s haymaker; cut to Brian McCann standing at home plate waiting for Jose Fernandez after his first career home run; then enter the commentator: “Is this wrong?”

While baseball’s moral code on gaining an edge is unpredictable, there’s always been the idea that individuals conform to the game, not the other way around.  Harper’s sermon won’t shatter the code of conduct, but it might move the needle, if it hasn’t already.  For example, I can’t think of a standout incident this season because of a bat flip.  That’s good! Because bat flips are really fun!  There’s really no need to overthink it.  There were plenty of memorable bat flips this year, and in an effort to make some fun out of baseball when there is no baseball being played, I’m breaking out my bat flip tracking equipment (a ruler, a stop watch, and a parabolic trajectory calculator) that I introduced last year, and booting up BatCast for a look back at the year’s most memorable wood-chucking moments.

A brief recap: arriving at these numbers is a sloppy and wildly imprecise affair.  I pull videos, gifs, and stills of a bat flip and start by measuring the height of the player as he appears on my screen.  I convert that measurement into the player’s real-life size and reference this ratio, as well as measurements on the baseball field, and rough estimates, to arrive at some of the data I present to you in meters and feet: initial height, apex, and distance.  Using a stopwatch or the time stamp on YouTube, I can declare a fairly accurate hang time of the bat.  Angles are roughly noted using the batter and the ground to form a 90-degree angle and are adjusted in the parabolic trajectory calculator.

Let’s kick this off:

Exhibit A – The one that’s probably at the forefront of your mind:

Le Flip

How about in slow motion?

Ejaculatory!

How many of his teammates do you think saw that flip?  They may have seen the tail end of it, but I’m willing to bet zero saw the flip in its entirety because everyone in the dugout was gazing at the ball in flight.  But this was a no-doubter.  Edubray Ramos resigned to the outcome likely before the ball had reached its apex.  The Phillies weren’t playing for anything at this point, but the Mets?  Before this pitch, the Mets were tied with the Giants and Cardinals for the top wild-card spot.  Before this pitch, in the 9th inning, Jose Reyes erased a two-run deficit with a home run of his own, only to see that lead given up again when Jeurys Familia and Jim Henderson allowed two runs to score in the top of the 11th.  After this pitch, this game ended and they had a half-game lead on any team in the National League for the first wild-card spot.  That bat flip is a team effort.  There’s some “I did it” in there, but the way he looks towards the dugout and offers his bat up towards his teammates makes this feel like “We did it!”

The numbers:

Cabrera is listed as 6′ tall.  On the freeze frame I measured, he’s 1.9″ tall.  So our key tells us that 1″ on the screen is 37.9″ in real life.  When he releases the bat, he does so from about shoulder height and we’ll call 5′ (1.52 m) in real life.  The acme is, it appears, not a great deal lower than the top of Asdrubal’s head, so we’ll tally that down at 5′-7″ (1.71 m).  To me, the launch angle looks to be right around 30 degrees, and we’ll refine this number once we get them in the parabolic trajectory calculator.  The duration of flight I’m using is the average number I’ve come up with through timing the video 10 times — 0.79 seconds.

Parabolic Trajectory Calculator:

Exhibit B – A Man Possessed:

If this picture was part of an emotional intelligence quiz, I’m sure the answers given as to what facial expression is being displayed would vary greatly.  To accurately assess the information in this picture it may behoove one to understand that, in baseball, home teams wear white and that the man in the background is most likely a fan of the home team and that his hands are held high in jubilation.  If you’re only looking at the horrifying ogre in the foreground who appears to be screaming at 67 Hz+, the pitch only a dog can hear, you’d be hard-pressed to say that is a happy man.  In fact, he may not be happy yet — he’s likely evoking a form of relief, having just exorcised the demons one faces when up to bat in the 16th inning of a tie baseball game; he looks like pure adrenaline.  Most of us don’t get to experience a moment like this in our lifetime so we don’t have a really strong reference point for what he’s feeling, but luckily you know what this article is about and there’s a gif:

PUMP! PUMP! PUMP IT UP!

That’s all lizard brain right there.  It’s a little undignified, but that’s the beauty of it.  Matt Adams is a dense, hulking man, and that makes it a little scarier and a little sillier.  Look:

Sassy.

The numbers:

This one is especially hard to measure because of Adams’ primitive (yet graceful) movements.  I extracted these numbers using the still image and the video:

Exhibit C – Into the Batosphere

Yoenis Cespedes made it into my BatCast segment last year with his nifty flip in the NLDS.  This flip follows a similar trajectory but he varies his look this time with a cross-bodied toss.  It’s rude:

“Hold my drink, bitch.”

While the lesson here is obvious, the mistake is not as easily avoided: get the fastball ball UP and in on Cespedes.

Because of the evidence we have, the numbers for this bat flip will be even more rough than the others — by the way, I hope you’re not a mathematician, and I apologize if you are.  The data we can gather is the launch angle and at what time stamp the bat reaches it’s highest point.  Here’s a better view of the angle:

Can we agree on shoulder height for the initial launch height to make things easier?   Let’s call it 5′ since Cespedes is 5′-10″.  We’ll say the bat was launched at a 70-degree angle and in the gif the bat appears to reach it’s apex at just before 0.4 seconds.

Exhibit D – The “I probably didn’t even need this bat to hit this home run” flip

After my long-winded intro it’s fitting to get to feature Bryce Harper in this piece.  He probably didn’t have as much fun this year as he did in 2015, but he appears to have gotten some enjoyment out of this shot.

Correct me if I’m wrong, but I believe that is what the kids call “Swagadoscious.”  I’ll just get right to the point this time.

Those are the ones that stuck out to me as the best flips of the year and I hope you were able to move past the rough estimates and get some enjoyment out of that as well.  I should note that Joc Pederson’s bat flip in the NLDS is omitted because I cannot find substantial evidence of an acme or distance.  And while a lefty going across his body like he did is pretty exotic, the uncertainty he exudes, combined with his panicked sashay, makes this effort pretty uncool.

(Scherzer looks super imposed here)

So what can we pretend to glean from this?  Based on WPA, it’s probably not surprising that Harper had the most disproportionate bat flip.  Looking at the Statcast data, Harper’s home run was also the “weakest” out of the group.  So I guess even if Bryce Harper says what he says just so he can get away with being a little douchey, he’s holding up his part of the deal.  Of course, bat flips aren’t what make baseball fun.  Baseball is fun because we can see so much of our own lives in the game — it’s the humanity.  It provokes endless curiosity and it will reward you if you know where to look.  It’s the only game that can end, not because of time, but with one swing, and flip, of the bat.

Don’t be afraid to clue me in to bat flips in the future — my Twitter handle is in my bio (below).

We all love baseball. And, since we’re on FanGraphs, odds are we also love baseball stats. A stat is always intended to measure one thing — how many home runs did Miguel Cabrera hit, how many bases did Ricky Henderson steal, how often does Joey Votto get on base, and so on.

The savvy fan knows that no one stat tells the whole story. Even WAR, which is our best estimator for how much value a player brings, requires us to dig further into how the player got there. Was it defense? Offense? If it was offense, how’d he get there — lots of walks, lots of home runs, or did he have a high BABIP? And which of these are most likely to repeat?

That’s where correlation comes into play. If there’s a high correlation season to season, then odds are what we’re measuring* is repeatable, and we can expect more of the same going forward. Otherwise, we should expect a regression (positive or negative) toward the league mean the next season.

Or maybe a player has a high line-drive rate. How’d they get there? Do they swing a lot, do they tend to be power hitters, etc. There are a lot of relationships within a season as well.

*Even at a seasonal level, we’re not necessarily measuring true talent so much as performance. Estimating true talent involves a lot of regression, and that’s a fun and important study, but not what I chose to focus on for this tool.

A few years back, Steve Staude published a hitting stat correlation tool that let anybody explore these correlations at their leisure. It was a fun way to explore the data, not to mention a neat piece of Excel engineering. I wanted to bring it up to date a bit, and in the process switch from Excel to Tableau. I can’t embed the view in this post, but you can view it by clicking through to the view on Tableau Public.

I decided to include every season in the FanGraphs database (I’m sure I owe a database admin somewhere an apology). By default, it filters on 300 PA for both metrics (either intra-season or next season), but you may drop the floor to 1 PA. It’s a terrible idea and you really shouldn’t do it, but I’m not here to tell you how to live your life.

I also added a yearly trend of the correlation. For most stats, this doesn’t add a lot, but there are some interesting stories. For instance, the yearly correlation between BABIP and AVG for players with 300 PA has been slowly dropping in the last 20 years. Reflective of more emphasis on walks, or perhaps defensive positioning?

The player with the highest swing rate and lowest strikeout rate? Randall Simon in 2002, with a 63.6% swing rate and a 5.9% K rate, which seems like some sort of joke. All that swinging meant a low 2.9% walk rate, so it’s not like he got away with anything.

The usual caveats about correlation not equaling causation apply. Just because you get a high r-squared doesn’t mean there’s a causal relationship; one always has to apply a common-sense analysis as well. That said, dive in and have some fun.

Statcast has revolutionized the way we look at batted-ball data. We have been spoiled with exit velocity, launch angle and so much more. After looking into this treasure trove of data, I began to wonder, how closely is a hitter’s overall production tied to their exit velocity? More specifically, I wanted to uncover whether production was tied to differences between Air EV and Ground EV. First, I calculated the difference between Air EV and Ground EV from Baseball Savant. Next, I filtered the list to only include those with at least 100 batted-ball events to not skew the sample. I also calculated AIR% by adding together LD% and FB% to see who is maximizing their contact and see who may need a change in approach.

This first chart illustrates which players have the largest difference between Air and Ground EV:

Byung-ho Park leads the way by nearly 4 MPH, with a difference of 17.3 MPH. With the exception of Eibner, Park and Arcia, this is a list of hitters whose primary BIP type is FBs. Each of these hitters has an AIR% over 60%, with Ryan Schimpf pacing the group at an incredible 80%. With such a stark difference in EV, each these players should focus on hitting the ball in the air to maximize their overall production. For Park, Eibner and Arcia, putting the ball on the ground severely limits how often they can make harder contact. All things equal, hard contact is better than soft contact and these players should adjust their approach accordingly to maximize hard contact, which could help their overall production.

As we move on, the next chart displays players with the smallest differences in Air and Ground EV:

The speedy Billy Burns tops this list, complemented by a group of players no one will mistake for sluggers. This group comes with considerably less ceiling and overall production. Of this group, only three guys managed to post a league-average or better wRC+ (Prado, Cabrera and Peraza). Lorenzo Cain has been better in the past but was hampered by injuries this past year. Cain and the three previously mentioned provide the blueprint for how this profile can work. By spraying the ball and making enough contact, these guys maximize their limited power but have a razor-thin line between their bats being productive and unplayable.

As an aside, there was only one player who had zero difference in his EVs. The culprit? Nick Markakis, which for some reason makes perfect sense. Anyways!

So now that I have shown the extremes we can begin to answer the original question: does EV difference even matter for overall production? To find out, I ran a couple different tests. First, I took the data and divided them evenly into quarters. The results look like this:

The top 50% of hitters with large differences in EVs hit average or slightly better. Meanwhile, hitters in the bottom 50% produced slightly below average. To give each group a face, I took the best hitter by wRC+ and here we have four elite hitters. So far we have a very minor indication that says players with larger EV differences hit better than those with smaller differences. What we do not have is a concrete reason to disqualify a hitter from being elite based on their EV differences.

Next, I took the data and plotted players’ EV Differences and wRC+ to see if there was any correlation.  The graph is about as random as it gets with an R squared value of .022. This shows that there is a relationship between EV differences and overall offensive production but nothing significant.

All things equal, you probably take the guy with the larger differences but that does not guarantee any kind of success. We now know that their differences of how hard they hit balls in the air or on the ground do not preclude them from being elite. Hitting is both art and science and what we have learned today only reinforces that hitters can have very different profiles and still have excellent results.

Jeff recently ran two articles about the season’s worst and best home runs, as measured by exit velocity.  As a small addendum to that, I’d like to include both exit velocity and launch angle to try to determine the season’s least likely home run.  So how do we do such a thing?  Warning!  I’m going to spend a bunch of time talking about R code and machine learning.  If you want to skip all that, feel free to scroll down a bit.  If, on the other hand, you’d like a more in-depth look at running machine learning on Statcast data, hit me up in the comments and I’ll write some more fleshed-out pieces.

As usual, we’re going to rely heavily on Baseball Savant.  Thanks to their Statcast tool, we can download enough information to blindly feed into a machine-learning model to see how exit velocity and launch angle affect the probability of getting a home run.  For instance, if we wanted to make a simple decision tree, we could do something like this.

# Read the data
my_csv <- 'hr_data.csv'
data_raw <- read.csv(my_csv)
# Make training and test sets
library(caret)
inTrain <- createDataPartition(data_raw$HR,p=0.7,list=FALSE)
training <- data_raw[inTrain,]
testing <- data_raw[-inTrain,]
# rpart == decision tree
method <- 'rpart'
# train the model
modelFit <- train(HR ~ ., method=method, data=training)
# Show the decision tree
library(rattle)
fancyRpartPlot(modelFit$finalModel)

That looks like what we would expect.  To hit a home run, you want to hit the ball really hard (over 100 MPH) and at the right angle (between 20 and 40 degrees).  So far so good.

Now, decision trees are pretty and easy to interpret but they’re no good for what we want to do because (a) they’re not as accurate as other, more sophisticated methods and (b) they don’t give meaningful probability values.  Let’s instead use boosting and see how well we did on our test set.

method <- 'gbm' # boosting
modelFit <- train(HR ~ ., method=method, data=training)
# How did this work on the test set?
predicted <- predict(modelFit,newdata=testing)
# Accuracy, precision, recall, F1 score
accuracy <- sum(predicted == testing$HR)/length(predicted)
precision <- posPredValue(predicted,testing$HR)
recall <- sensitivity(predicted,testing$HR)
F1 <- (2 * precision * recall)/(precision + recall)

print(accuracy) # 0.973
print(precision) # 0.792
print(recall) # 0.657
print(F1) # 0.718

The accuracy number looks nice, but the precison and recall show that this is far from an amazingly predictive algorithm.  Still, it’s decent, and all we really want is a starting point for the conversation I started in the title, so let’s apply this prediction to all home runs hit in 2016.

Once you throw out some fairly clear blips in the Statcast data, the “winner”, with a 0.3% chance of turning into a home run, is this beauty from Darwin Barney.*  This baby had an exit velocity of 91 MPH and launch angle of 40.7 degrees.  For fun, let’s look at where similarly-struck balls in the Rogers Centre ended up this year.

* I’m no bat-flip expert, but I believe you can see more of a flip of “I’m disgusted” than “yay” in that clip.

Congrats Darwin Barney!  There are no-doubters, then there are maybes, and then there are wall-scrapers.  They all look the same in the box score, but you can’t fool Statcast.

In “Hardball Retrospective: Evaluating Scouting and Development Outcomes for the Modern-Era Franchises”, I placed every ballplayer in the modern era (from 1901-present) on their original team. I calculated revised standings for every season based entirely on the performance of each team’s “original” players. I discuss every team’s “original” players and seasons at length along with organizational performance with respect to the Amateur Draft (or First-Year Player Draft), amateur free agent signings and other methods of player acquisition.  Season standings, WAR and Win Shares totals for the “original” teams are compared against the “actual” team results to assess each franchise’s scouting, development and general management skills.

Expanding on my research for the book, the following series of articles will reveal the teams with the biggest single-season difference in the WAR and Win Shares for the “Original” vs. “Actual” rosters for every Major League organization. “Hardball Retrospective” is available in digital format on Amazon, Barnes and Noble, GooglePlay, iTunes and KoboBooks. The paperback edition is available on Amazon, Barnes and Noble and CreateSpace. Supplemental Statistics, Charts and Graphs along with a discussion forum are offered at TuataraSoftware.com.

Don Daglow (Intellivision World Series Major League Baseball, Earl Weaver Baseball, Tony LaRussa Baseball) contributed the foreword for Hardball Retrospective. The foreword and preview of my book are accessible here.

Terminology

OWAR – Wins Above Replacement for players on “original” teams

OWS – Win Shares for players on “original” teams

OPW% – Pythagorean Won-Loss record for the “original” teams

AWAR – Wins Above Replacement for players on “actual” teams

AWS – Win Shares for players on “actual” teams

APW% – Pythagorean Won-Loss record for the “actual” teams

Assessment

The 1979 New York Mets 

OWAR: 50.7     OWS: 262     OPW%: .479     (78-84)

AWAR: 24.8      AWS: 188     APW%: .389     (63-99)

WARdiff: 25.9                        WSdiff: 74  

The “Original” 1979 Mets ended the season in the cellar, yet the club outpaced the “Actuals” by fifteen victories! Ken Singleton earned runner-up status in the MVP balloting on the strength of a .295 BA with 35 circuit clouts and 111 ribbies. Lee “Maz” Mazzilli (.303/15/79) nabbed 34 bags and merited his lone All-Star appearance. Tim Foli set personal-bests in batting average (.288), base hits, runs and RBI. John “The Hammer” Milner contributed a .276 BA with 16 jacks while splitting time between left field and first base. “Actuals” right fielder Joel Youngblood posted a .275 BA and raked 37 doubles. Richie “The Gravedigger” Hebner added 25 two-base knocks and drove in 79 baserunners.

Tom Seaver and Nolan Ryan rated sixth and twenty-fourth, respectively, among pitchers in the “The New Bill James Historical Baseball Abstract” top 100 player rankings. “Original” Mets teammates registered in the “NBJHBA” top 100 ratings include Ken Singleton (18th-RF) Paul Blair (66th-CF) and Bud Harrelson (88th-SS). “Actuals” third baseman Richie Hebner ranked fifty-sixth while center fielder Jose Cardenal placed seventh-sixth.

  Original 1979 Mets                                  Actual 1979 Mets

Jerry Koosman reached the 20-win plateau for the second time in his career. Tom “The Franchise” Seaver (16-6, 3.14) led the National League with 5 shutouts and finished fourth in the Cy Young Award balloting. Nino Espinosa delivered 14 victories with a 3.65 ERA. Nolan Ryan aka the “Ryan Express” tallied 16 victories and struck out 223 batsmen. Craig Swan augmented the “Originals” and “Actuals” rotation with 14 wins and a 3.29 ERA after securing the National League ERA title during the previous campaign.

  Original 1979 Mets                                  Actual 1979 Mets 

 Notable Transactions

Ken Singleton 

April 5, 1972: Traded by the New York Mets with Tim Foli and Mike Jorgensen to the Montreal Expos for Rusty Staub.

December 4, 1974: Traded by the Montreal Expos with Mike Torrez to the Baltimore Orioles for Bill Kirkpatrick (minors), Rich Coggins and Dave McNally. 

Jerry Koosman 

December 8, 1978: Traded by the New York Mets to the Minnesota Twins for a player to be named later and Greg Field (minors). The Minnesota Twins sent Jesse Orosco (February 7, 1979) to the New York Mets to complete the trade. 

Tom Seaver

June 15, 1977: Traded by the New York Mets to the Cincinnati Reds for Doug Flynn, Steve Henderson, Dan Norman and Pat Zachry.

Nino Espinosa

March 27, 1979: Traded by the New York Mets to the Philadelphia Phillies for Richie Hebner and Jose Moreno.

Nolan Ryan

December 10, 1971: Traded by the New York Mets with Frank Estrada, Don Rose and Leroy Stanton to the California Angels for Jim Fregosi.

Honorable Mention

The 2012 New York Mets 

OWAR: 27.7     OWS: 262     OPW%: .492     (80-82)

AWAR: 24.1       AWS: 221      APW%: .457    (74-88)

WARdiff: 3.6                        WSdiff: 41

The “Original” 2012 Mets placed third, fourteen games in arrears to the Nationals. David “Captain America” Wright (.306/21/93) raked 41 two-base hits and received his sixth All-Star invite. Angel “Crazy Horse” Pagan topped the circuit with 15 triples and set career-highs with 38 two-baggers and 95 runs scored. Jose B. Reyes swiped 40 bags and rapped 37 doubles while double-play partner Daniel Murphy contributed a .291 BA with 40 two-base knocks. Nelson R. Cruz nailed 45 doubles and jacked 24 round-trippers. First-sacker Ike B. Davis established personal-bests with 32 taters and 90 ribbies. A.J. Burnett paced the starting staff with 16 victories along with a 3.51 ERA and 180 strikeouts.

On Deck

What Might Have Been – The “Original” 2013 Marlins

References and Resources

Baseball America – Executive Database

Baseball-Reference

James, Bill. The New Bill James Historical Baseball Abstract. New York, NY.: The Free Press, 2001. Print.

James, Bill, with Jim Henzler. Win Shares. Morton Grove, Ill.: STATS, 2002. Print.

Retrosheet – Transactions Database

The information used here was obtained free of charge from and is copyrighted by Retrosheet. Interested parties may contact Retrosheet at “www.retrosheet.org”.

Seamheads – Baseball Gauge

Sean Lahman Baseball Archive

The best relief pitcher in baseball is not who you think he is. Most of you probably would not even include him in the top 10. If I were to take a poll on who is the best relief pitcher in baseball, the top voted would likely be Zach Britton, Dellin Betances, Aroldis Chapman, Kenley Jansen, and Andrew Miller. I will say that it is none of them. To illustrate my point, I will compare this mystery pitcher’s numbers to all of their numbers. Nothing too scary, just xFIP, K/9, and ERA. I also will not just tell you which pitcher produced which numbers. Where would be the fun in that? I will compare the numbers of all six pitchers and walk you, the reader, through determining which one is the best.

Pitcher A: 1.18 xFIP; 14.89 K/9; 1.45 ERA
Pitcher B: 1.92; 13.97; 1.55
Pitcher C: 1.17; 16.84; 1.16
Pitcher D: 1.75; 15.53; 3.08
Pitcher E: 2.41; 13.63; 1.83
Pitcher F: 2.09; 9.94; 0.54

At first glance, Pitcher F’s ERA of 0.54 is likely what stands out most. Alas, even calling him only by a letter cannot mask Britton. He has the lowest K/9 by far and the second-highest xFIP, so Britton is effectively taken out of consideration.

Pitcher D has an ERA over a run higher than any of the others. His K/9 and xFIP fit in the range but do not stand out. Thus, Dellin Betances is out as well.

Of the remaining four, Pitcher E rates the worst in each of the three categories. Goodbye, Kenley Jansen.

That leaves us with Pitcher A, Pitcher B, and Pitcher C. In this group, B is the worst across the board. Aroldis Chapman leaves the conversation.

Pitcher C is better than Pitcher A in all three statistics. Andrew Miller bows and exits.

Carter Capps stands victorious.

Yes, I know Capps did not pitch in 2016. I used his 2015 numbers. They stack up just as well against the elite relievers from that year as well. It is true that Capps pitched only 31 innings in 2015, but the stats I used are rates. Maybe a larger sample would have dragged him into mediocrity, but I doubt it. Capps was ahead of the field by such a large margin that even with regression in his 2017 return he would be #1.

I am crazy for saying Carter Capps is the best relief pitcher in baseball. Or am I, really? If Capps pitches as well in 2017 as he did in 2015, just over a larger sample, I believe many of you will agree with me. Some of you may even agree with me after reading this.

So, let me be the first to say it: Carter Capps is the real best relief pitcher in baseball.

This is a question that intuitively would seem to be answered by: Sure, why not?  The assumption was recently made in the comments section of this article by an FG writer:

Think about it — if you are Rougned Odor and you are on first base and, say, Joey Gallo is at the plate, there’s a good chance he’s going to cool down the stadium with some high-powered fanning.  He’s not exactly known as a high-contact guy.  There’s a roughly one-in-three chance that his at-bat is going to end in a backwards K sign being held up by someone in the stands.  So ‘Ned might decide this is a good time to steal because the ball isn’t likely to be put into play in the air, where, if caught, he would have to double back to tag.  Maybe he’s also thinking that, like Brad Johnson alluded, the break-even point for a steal (famously ~75% success rate as calculated by Bill James in Moneyball, ~66% in this more recent FG article) is lower if the guy at the plate is likely to cause an out, specifically a strikeout which normally doesn’t allow a runner to advance like a bunt, grounder or long fly might.

On the other hand, maybe Odor doesn’t have such a cynical view of Gallo, and doesn’t change his mindset on the basepaths.  Maybe he doesn’t try to assume what Gallo might do, so he doesn’t go for any more risky of a steal than he otherwise might.  So maybe he isn’t stealing at a higher rate than normal if the guy at the plate is a K machine.  Heck, maybe Joey Gallo is a specifically bad example here, because, though he does whiff a lot, he also hits a lot of home runs, which might cause a runner to take fewer risks when waiting on the outcome of his plate appearance.

So, let’s looks at what the numbers have to say.  I ran a simple correlation analysis between team stolen-base totals and team K%.  Here’s what I got:

So, no real correlation to be seen here.  But perhaps that shows that it could be a market inefficiency.  In 2016, the Brew Crew led the league in both K% and stolen bases.  Even without John Villar’s big SB season, they are a top-five SB team.  Below is a chart from last year — in yellow are the top five teams in both total SBs and K%.

Perhaps the Rays should have been trying to steal some more?  Though some of these anomalies could just simply be explained by personnel issues — maybe teams like the Orioles just have no one who can steal on the entire squad?

Here’s the same chart, for 2015, just for sugar and giggles:

For the Astros, this is starting to look like a trend — Orioles too.  I think my final answer to the question posited by this post is — Hmm, not sure exactly.  But maybe?

As Cubs fans and non-Cubs fans alike celebrate the end of the 108-year drought, we have overlooked the fact that in winning, the Cubs also bucked two trends in major league baseball:

1. 100+ win teams struggle in the postseason and rarely win the World Series, especially since the wild-card era began in 1995
2. Losers of the ALCS and NLCS (Cubs lost 2015 NLCS) historically decline the following season, both in win total and playoff appearance/outcome

Below is a table to quantify a team’s performance in the playoffs:

*The LCS was a best-of-five-game series from 1969 through 1984

It is important to acknowledge how close a team comes to winning a particular round. Based on a 0 to 4 scale, with 0 indicating the team missed the playoffs and 4 indicating the team won the World Series, the table credits fractions of a whole point for each playoff win. For example, in a best-of-seven-game series, each win (four wins needed to clinch) is worth 0.25. In a best-of-five-game series, each win (three wins needed to clinch) is worth 0.333 (1/3). Any mention of playoff result or average playoff result in this article is derived from this table.

THE STRUGGLE OF 100+ WIN TEAMS IN THE POST-SEASON

Playoff baseball, due to its small sample size and annual flair for the dramatic, historically has not treated exceptional regular season teams well. Jayson Stark recently wrote an article for ESPN titled, “Why superteams don’t win the World Series.” He noted that only twice in the first 21 seasons of the wild-card era had a team with the best record in baseball won the World Series (1998 and 2009 Yankees). Those two Yankee teams are also the only two 100-win ball clubs in the wild-card era to win the World Series. Research in this article will span the years 1969 to 2015, with 1969 being the first year of the league championship series (LCS).

Entering the 2016 season there had been 47 100+ win teams since the start of the 1969 season. Of those, 10 (21.3%) won the World Series. Other than those 10 World Series winners, how did 100+ win teams fare in the post-season?

Below are the average playoff results for 100+ win teams in each period of the major league baseball playoff structure from 1969 to 2015. The playoff structures were as follows:

1969-1984: LCS (best of 5 games) + World Series (best of 7 games)

1985-1993: LCS (best of 7 games) + World Series (best of 7 games)

1995-present: LDS (best of 5 games) + LCS (best of 7 games) + World Series (best of 7 games)

The wild-card game (2012-present) is omitted because a 100+ win team has yet to play in that game, although it certainly would be rare if we ever see a 100+ win team playing in the wild-card game.

As the data shows, 100+ win teams during the 1969-1984 period on average made a World Series appearance. This could be partly due to the fact there was only one round of playoffs (the LCS) ahead of the World Series, with the LCS being a best of five games. It was certainly a much easier path to the World Series once a team made the playoffs, yet on average 100+ win teams were finishing with a World Series sweep.

Changing the LCS from a best-of-five-game series to a best-of-seven-game series had a negative impact on team post-season performance, as 100+ win teams during the 1985-1993 span on average lost a deciding Game Seven in the LCS.

When the league added the wild card and LDS in 1995, it expanded the opportunity to make the playoffs but made the path to a World Series title more difficult, for a team now had to win 11 games to hoist the trophy. In the wild-card era, 100+ win teams are on average losing 4-1 in the LCS. This period also has the lowest percentage of 100+ win teams winning the World Series.

Using average playoff result standard deviation and a normal distribution, we can also see that the likelihood of a 100+ win team to win the World Series has had a significant decrease over the past several decades, left at under 7% during the wild-card era. The longevity of 100+ win teams in the playoffs has been trending downward over the past several decades. Despite being on the verge of a World Series defeat, the Cubs were able to successfully break through and buck a trend that had haunted outstanding regular-season teams for decades, especially since the wild-card era began in 1995.

THE CURSE OF THE LCS DEFEAT

The 2015 Cubs lost to the Mets in the NLCS yet bounced back in 2016 to have an even better regular season and win the World Series. This, however, was a rare feat. Teams that lose in the LCS historically win fewer regular-season games and perform worse on average in the post-season (if they make it) the following year. Below are two charts (1969-2015 and 1995-2015) that display average win differential, average playoff result, likelihood win differential is greater than +5 (2016 Cubs were +6), and the likelihood of winning the World Series.

Due to the 1981 and 1994 strikes, a few data points for win differential and playoff result are not included in the calculation. The data set includes 82 LCS losers for win differential and 88 LCS losers for average playoff result. The 1980-81, 1981-82, 1993-94, 1994-95, 1995-96 win differentials are not included for LCS losers in both leagues. The 1994 and 1995 playoff results are not included for LCS losers in both leagues because there was no post-season in 1994, hence no LCS loser. Regardless, there is a notable trend among LCS losers to perform worse the following season.

The 2016 Cubs not only won six more regular-season games than in 2015, but they became only the seventh team in history to lose the LCS one season and win the World Series the following season (1971 Pirates, 1972 Athletics, 1985 Royals, 1992 Blue Jays, 2004 Red Sox, 2006 Cardinals). Two of the previous six teams repeated as champions: 1973 Athletics and 1993 Blue Jays. Most recently, the 2005 Red Sox lost 3-0 in the ALDS and the 2007 Cardinals failed to make the playoffs.

LOOKING FORWARD

The Cubs have already been pegged favorites to win the 2017 World Series, which isn’t surprising given the fact nearly every key player is under team control. Is history on their side? Winning back-to-back titles is difficult in today’s competitive league, as new baseball thinking has somewhat evened the playing field and the small sample size of post-season baseball has the ability to lend unexpected results.

The 10 100+ win teams who have won the World Series since 1969 historically have not been successful in their attempts for back-to-back titles. Below are the average win differentials and average playoff result for these teams in the season following their championship:

Only three of these 10 teams (1975-76 Reds, 1977-78 Yankees, 1998-99 Yankees) have repeated as champions. Can the 2017 Cubs be the fourth? No matter the numbers, the 2017 Cubs still have to perform on the field. They were on the brink of losing the World Series in 2016, so we must not take anything for granted. But despite this, there’s no doubt the 2017 Cubs will be in a good position for a repeat. The Cubs are expected to be MLB’s best regular season team in 2017, according to FanGraphs and Jeff Sullivan’s analysis in his November 11, 2016 article. Only time will tell.

The case for never letting pitchers bat in the NL has just gotten a whole lot better. I now estimate that if a NL team were to always pinch-hit for their pitchers they would expect to pick up a whopping 7.2 wins per year. And that, my friends, is a game-changer.

In my initial post two weeks ago I laid out a strategy in which a National League manager pinch-hits for his pitchers every time their turns come up in the batting order. I called it the “Pitchers Never Bat” strategy. The manager would keep a pitching staff of 11 “relievers” and no “starters.” The major benefit of doing this, I estimated, would be an improvement in the team’s offense.

I addressed what I considered the two major “components” of the analysis and estimated that the impact of this strategy was worth an extra 3.6 wins per year if the team was the only team in the National League to implement it. I also identified four other components of the analysis that could possibly add to, or take away from, my initial estimate of 3.6 wins per year.

In this follow-up post I will do two things. First, I will make some improvements by estimating the impact of two of the four components that I previously left unaddressed. And second, I will address some concerns raised by some members of the FanGraphs community via their thoughtful comments on my initial post.

Here is where I left off at the end of my original post:

Estimated Change in Wins Per Year by Component –

Component #1:   +3.6

Change in Runs due to pinch hitters batting for all pitchers

Component #2:   +0.0

Change in Runs Allowed due to using pitching staff in a new way

Component #3: Not Evaluated

Change in Runs Allowed due to added flexibility in selecting pitchers based on how they are warming up prior to or during a game

Component #4: Not evaluated

Change in Runs Allowed due to opponents’ inability to “stack the lineup” to take advantage of the starting pitchers “handedness” (i.e., lefty or righty)

Component #5: Not evaluated

Change due to reducing size of pitching staff by 1-2 men

Component #6: Not evaluated

Change in Runs Allowed due to the “times through the order” effect

TOTAL:                +3.6 Wins per Year

IMPROVEMENTS

So now, let’s make some improvements to the prior analysis. Here, I’m going to add estimates for the impacts of Components #4 and #6:

Component #4 – Handedness

In my “Pitchers Never Bat” strategy, the starting pitcher leaves the game when his turn in the batting order comes up, as a pinch-hitter takes his place. In this approach the starting pitcher will typically throw 1-3 innings, averaging two innings per start. Compare this to the conventional starting pitcher who will throw six innings, on average. If the opposing manager were to “stack” (or “tilt”) his batting order to have more lefties (LHB) to face a righty starting pitcher (RHB), or more RHB to face a LHP, as they do now, the value of his tilt would only be in effect for two innings, not six. The manager of the team using the “Pitchers Never Bat” strategy would most likely bring in the next two relievers with the opposite hand of his starter. Example: A lefty starter goes two innings, and is replaced by two consecutive right-handed relievers who would pitch two innings each.

After reviewing league averages for wOBAs for each of the four “handedness combinations” (i.e., LHP/LHB, LHP/RHB, RHP/LHB, and PHP/RHB) as well as how much managers “tilt” their line-ups to take advantage of the starting pitcher’s hand, I estimate that the opponent would lose his current handedness advantage for, on average, four PAs per game, with each of these PAs reducing his batters’ expected wOBA by 18 points for these PAs. Over 162 games, that amounts to 648 PAs per year. Using the rule of thumb that a decrease of 20 wOBA points decreases team run production by 10 runs per every 600 PAs, I estimate that the opponents will lose 9.8 runs/year (that is 18/20 * 10 * 648/600). And since every 10 runs is worth a win, on average, that’s a positive impact to the team implementing the “Pitchers Never Bat” strategy of about 1.0 wins/year (= 9.8/10).

But, since opponents will quickly catch on to the new strategy that they are facing, they should immediately stop trying to “stack” or “tilt” their line-ups. If the opposing manager puts up a line-up that is set up with absolutely no regard to lefty or righty pitching, he can reduce the negative impact to his offense by about 25%, down to a loss of 7.3 runs/year, or a loss of 0.7 wins/year. Since I assume that the opponents will take this less damaging approach, I will use +0.7 wins/year as a conservative estimate for Component #4.

Component #6 – Times Through the Order

Times Through the Order (TTO) refers to differences in pitcher performance due to how many times pitchers have faced the opposing lineup. I recently read an excellent piece on this topic by Mitchel Lichtman, published on Baseball Prospectus on 11/5/13, entitled “Everything You Always Wanted to Know About the Times Through the Order Penalty.” I will draw on one of his many key findings to estimate the impact of TTO on the “Pitchers Never Bat” strategy.

Lichtman presents data (drawn from 2000-2012) which shows that starting pitchers are, on average, at their best the first time through the line-up, are worse the second time through, and even worse the third time through. Using “wOBA against” statistics (adjusted appropriately for batter quality), he shows that pitchers suffer a decay of about 10 points in wOBA against when going from the first TTO to the second TTO, and then decay another 10 points when going from the second TTO to the third TTO. He also estimated the wOBA against statistic for the second TTO is equal to the pitchers’ overall wOBA against. So, in other words, starting pitchers are about 10 points better than average for the first TTO, about average for the second TTO, and about 10 points worse than average for the third TTO.

In the “Pitchers Never Bat” strategy, starters will occasionally work into the beginning of the second TTO, so I’ll assume that 80% of the batters they face will be in the starter’s first TTO, and 20% will be in the second TTO. This means that their wOBA against should be about eight points better (=10 points * 80%) than they would see if they were used in the conventional six-plus inning approach. This advantage will be repeated again by the relievers who replace the starter and pitch through the sixth inning, or until the time that the starter would typically be pulled when using a conventional pitching staff. Think of it this way – instead of a starter throwing a wOBA against of .320 for the first six innings, you get a starter plus two relievers each throwing a wOBA against of .312 for the first six innings. And this benefit is strictly due to the TTO effect.

Improving your wOBA against statistic by eight points for the first six innings of every game means that these pitchers will face about 4,374 batters per year (= 27 PA per game X 162 games.)  Again, using the rule of thumb of 20 woBA points equates to 10 runs per 600 PA, I estimate the impact of this improvement to be a decrease in Runs Allowed of 29.2 runs per year (=8/20 * 10 * 4,374/600.) And using the rule of thumb that 10 runs per year equates to one additional win per year, I can finally estimate that the positive impact of the TTO effect to be 2.9 additional wins per year (=29.2/10).

Now, let’s revisit where we stand with our six components:

Estimated Change in Wins Per Year by Component –

Component #1:   +3.6

Change in Runs due to pinch hitters batting for all pitchers

Component #2:   +0.0

Change in Runs Allowed due to using pitching staff in a new way

Component #3: Not Evaluated

Change in Runs Allowed due to added flexibility in selecting pitchers based on how they are warming up prior to or during a game

Component #4:   +0.7

Change in Runs Allowed due to opponents’ inability to “stack the lineup” to take advantage of the starting pitchers “handedness” (i.e., lefty or righty)

Component #5: Not evaluated

Change due to reducing size of pitching staff by 1-2 men

Component #6:   +2.9

Change in Runs Allowed due to the “times through the order” effect

TOTAL:                 +7.2 Wins per Year

CONCERNS FROM COMMENTERS

Commenters to my original post raised no objections with my estimated value of +3.6 wins per year due to Component #1, which is the expected change in runs due to pinch-hitters batting for all pitchers. Their two primary concerns were regarding Component #2, which is the change in runs allowed due to using the pitching staff in a new way. Commenters were concerned that my proposed staff of 11 pitchers, averaging 130 innings pitched (IP) per year each, would not be able to handle that large a workload, and therefore the pitchers’ performances would be worse than they would be as part of a traditional pitching staff.

On the issue of workload I see it as follows: Say half of the new staff comes from current relievers who are used to throwing 50-80 IP per year. The new strategy would ask them to average 100-130 IP per year. And let’s say that the other half of the new staff comes from current starters who are used to throwing 160-200 IP per year. The new strategy would ask them to throw 130-160 IP year. So, yes, one would expect that the old relievers would probably pitch worse if they were asked to throw an extra 50 IP per year. But, by similar logic, the old starters would be expected to pitch better if they were asked to reduce their workload by 30 or 40 IP per year. Do these two effects offset each other? Does one dominate the other? I don’t know. Even if Component #2 resulted in a negative net effect, how big could it be? Could it be large enough to outweigh the +7.2 wins estimated from Components #1, #4, and #6? I don’t think so.

And what if, instead, the GM hired 11 guys for the staff that were all starters previously? Would that lead to a net gain to the staff’s performance due to reduced workloads per person? Potentially. Also, note that the impact we are talking about here is solely due to workload and has nothing to do with Handedness (Component #4) or Times Through the Order (Component #6).

For those still concerned that an average of 130 IP for each of 11 pitchers is still a big negative, here are three ways to reduce the average workload:

First, due to call-ups from the minors, visits to the DL, and expanded rosters in September, team workloads are actually shared by far more than the current 12-13 pitchers on the roster at any one point in time. In 2016 the median number of pitchers used by NL teams was 27. If you ranked each team’s staffs at year-end by IP, and then added up the IP thrown by their top 12, you’d find the top 12 typically account for about 80% of their team’s total IP. So you could safely reduce my 130 IP per person that I required for the “Pitchers Never Bat” strategy by 10% to adjust for that. That brings the average workload required down to 117 IP (= 130 * 90%).

Second, some commenters suggested I keep a 12-man staff, not 11 as I proposed. Doing this would decrease the average workload per pitcher by another 8%, or about 9 IP. That would bring down the average workload from 117 IP to 108 IP. (Of course, this would require that the number of position players be reduced by one, and there would be some negative impact because of that.)

Third, as I mentioned in my first post, a team could keep an ace starter that is allowed to bat for himself. He would be used exactly as an ace is used now, pitching 6+ innings every fifth day. In this variation, the “Pitchers Never Bat” strategy would only be used on the four days that the ace is resting. So, here the ace would pitch about 180 innings, reducing the workload for each of the other pitchers by another six innings per year, bringing their average workload down to about 102 IP. (By the way, I roughly estimate that the ace would need to have an expected WHIP of 1.05 or lower to justify allowing him to bat. At a WHIP of 1.05, the added benefit of letting him pitch 6+ innings would just offset the benefits from Components #1, 2, 4, and 6.)

So, to recap, with all three of these changes incorporated, the staff would consist of an ace throwing 180 IP, plus 11 others averaging about 102 IP, and another 15 or so pitchers that come and go throughout the season to support the 12 “primary” guys by sucking up the remaining 20% of the entire team’s IP. This should alleviate the concerns about pitcher workload.

I’m still not totally comfortable quantifying the impact of Component #2 yet, but I’m going to go out on a limb and say that if the staff was developed from 11 guys who were previously starters throwing 180 IP, the smaller workload should improve their average performance. My hunch is that the impact might be slightly positive, whereas the commenters thought it was negative. At this point I’m still going to leave the impact of Component #2 at 0.0, or no change, pending further evaluation.

CONCLUSION

By adding estimates for the impacts of “Handedness” (Component #4) and “Times Through the Order” (Component #6) my total estimated value of the “Pitchers Never Bat” strategy has jumped dramatically from +3.6 wins (in my initial post) to +7.2 wins per year. If this were to hold up, this would be an astounding gain to any NL team that implemented the strategy. At the going rate of $8 million per year that teams currently pay per win, this equates to about $58 million per year. I look forward to hearing your comments regarding this analysis.

Oh, and by the way, if any NL team would like to discuss additional analysis and/or implementation of this strategy please feel free to contact me at howardsrubin@gmail.com.