Trusting the Numbers: Emilio Pagán vs. Griffin Jax

3 hours ago, jimbo92107 said:

Emilio Pagan is a prime example of what a manager and FO are NOT supposed to do: Give a guy a free baseball education at the cost of losing games. First we heard he had "electric stuff," but he needed to work on his command. Okay, send him down to AAA to...Oh wait, let's stick him in the major league bullpen and just see what happens. Disaster, of course. Low leverage situations? Hey, now he looks pretty good! Let's try to work him back into... Walk walk, BOOM BOOM, the Pagan Sacrifice comes back. 

Sometimes a guy just doesn't work out. DFA him, and if he clears waivers (any other stooopid FO's out there?), send him back to the Saints. If he mows down hitters for the rest of the season with almost no walks, then consider...trading him. 

Jax is completely different. Weakly hit balls will regress to the mean. However, he should polish up his change-up. A heater and a slider are not enough to fool hitters 1 through 9. That is, unless Jax can peg the four corners with his heater, which so far he does not do. 

Plus 100.

1 hour ago, Wizard11 said:

It is mind boggling that so many fans who claim to be watching this team can identify Emilio Pagan as the problem with the Twins season.  Pretty sure his 2 bad outings are not the reason for this team’s gross underperformance.

Are you sure it's only two?

I thought it would be more, but I found 4 games in his log where the win-probability-added was negative.  That's far from a perfect stat, but for the purposes of identifying bad games for a reliever it's a decent place to start. It's not always the earned runs charged either - an inherited runner scoring can change the tide in a close game.

Aside from those, WPA gave him a nearly neutral outcome when he allowed an inherited runner to score in that 16-3 laugher we won against the Cubs.  "Only" letting that one run in but garnering two outs moved us closer to that win, LOL.

Your general point that the majority of Pagan appearances haven't been implosions is right, though. The offense has been the culprit far more often than the bullpen, and Pagan isn't the only bullpen member who has produced clunkers. 

He's not "THE" problem with the Twins season.  But that's a kind of low bar to set, isn't it?

On 5/29/2023 at 12:19 AM, ashbury said:

Are you sure it's only two?

I thought it would be more, but I found 4 games in his log where the win-probability-added was negative.  That's far from a perfect stat, but for the purposes of identifying bad games for a reliever it's a decent place to start. It's not always the earned runs charged either - an inherited runner scoring can change the tide in a close game.

Aside from those, WPA gave him a nearly neutral outcome when he allowed an inherited runner to score in that 16-3 laugher we won against the Cubs.  "Only" letting that one run in but garnering two outs moved us closer to that win, LOL.

Your general point that the majority of Pagan appearances haven't been implosions is right, though. The offense has been the culprit far more often than the bullpen, and Pagan isn't the only bullpen member who has produced clunkers. 

He's not "THE" problem with the Twins season.  But that's a kind of low bar to set, isn't it?

I was looking at WPA the other day and realized that I probably need a primer on how it operates. I get the whole, “check the win probability before and after each at bat and add them up” thing.

But what do they mean? On the surface, I’ve been assuming that a positive WPA in a game means that the player gave his team a better chance to win and a negative means that he gave his team a lesser chance to win. Is that a correct interpretation? 

Secondly, the participants in a given game always total 0.00, does there tend to be a similar number of guys with positives and guys with negatives? Or does it often skew to have a bunch of guys with positive and a comparably few guys with big negative values? 

I’m asking because I look at Pagan and see what looks like a bad season, because he’s got a total of -0.173 (though I don’t know enough to know whether that’s bad or REALLY bad). 

But, as you note, if you look at him on a game-by-game basis, he’s only been negative in 5 of his 20 games. That includes the debacle in L.A. and the game in Boston when he was left out to dry. Take out those two, and his season total is easily positive. Another of the negatives was Sunday, when he hit a batter and was pulled because of injury. It seems hard to count that one, but technically, he did lessen the team’s chance to win with the one batter he faced.

They’ve generally been careful on how they’ve used him — I get that — but it seems like a “won-loss record” of 15-5 (i.e. positive vs. negative) is actually pretty good. (And I do recognize that a number of the positives were very small, but that seems to be greatly affected by the situations they’ve put him in.)

I guess my main point is to say that I agree with Wizard11. It sometimes (often?) feels like he could be pitching like the second coming of Mariano Rivera and folks on TD would still be calling for his DFA. 

———————

And to clarify, Ash, though I used quoting you as my starting point, I’m not pointing at you in particular with my last paragraph. I honestly don’t know what your comments about Pagan have been. That paragraph is about TD in general. And if you or others show me a better understanding of WPA that helps me understand that I’m using it wrong, I’m fine with that learning. 

22 hours ago, IndianaTwin said:

I was looking at WPA the other day and realized that I probably need a primer on how it operates. I get the whole, “check the win probability before and after each at bat and add them up” thing.

But what do they mean? On the surface, I’ve been assuming that a positive WPA in a game means that the player gave his team a better chance to win and a negative means that he gave his team a lesser chance to win. Is that a correct interpretation? 

Secondly, the participants in a given game always total 0.00, does there tend to be a similar number of guys with positives and guys with negatives? Or does it often skew to have a bunch of guys with positive and a comparably few guys with big negative values? 

I’m asking because I look at Pagan and see what looks like a bad season, because he’s got a total of -0.173 (though I don’t know enough to know whether that’s bad or REALLY bad). 

But, as you note, if you look at him on a game-by-game basis, he’s only been negative in 5 of his 20 games. That includes the debacle in L.A. and the game in Boston when he was left out to dry. Take out those two, and his season total is easily positive. Another of the negatives was Sunday, when he hit a batter and was pulled because of injury. It seems hard to count that one, but technically, he did lessen the team’s chance to win with the one batter he faced.

They’ve generally been careful on how they’ve used him — I get that — but it seems like a “won-loss record” of 15-5 (i.e. positive vs. negative) is actually pretty good. (And I do recognize that a number of the positives were very small, but that seems to be greatly affected by the situations they’ve put him in.)

I guess my main point is to say that I agree with Wizard11. It sometimes (often?) feels like he could be pitching like the second coming of Mariano Rivera and folks on TD would still be calling for his DFA. 

———————

And to clarify, Ash, though I used quoting you as my starting point, I’m not pointing at you in particular with my last paragraph. I honestly don’t know what your comments about Pagan have been. That paragraph is about TD in general. And if you or others show me a better understanding of WPA that helps me understand that I’m using it wrong, I’m fine with that learning. 

These are all fair questions and I'm not sure I'm up to the challenge of giving you a definitive answer.  I have opinions, and it's going to be hard to avoid rambling on a topic like this.  I don't know whether enough others are even reading this thread by this point, to make it worthwhile to open a dialog, but I do find it hard to resist potential rabbit holes. 😀

I could invest many words defining WAR and its relative Wins Above Average, contrasted to WPA, but trying to be brief I'll say that they approach the ultimate analytics question of "Where Do Wins Come From?" from different starting points and nearly completely different philosophies. One tries to compare a log of a player's outcomes to a mythical replacement player (say, Ryan Lamarre or Tyler White?), one compares to a generalized average player (Max Kepler?), and one starts with the assumption that a team is 50% to win any particular game and looks for the contributions that day that bring a given game to 100% or down to 0% by game's end.  Expressed as fractions, that is .500 to divvy up among the players' contributions on the winning side, and -.500 for the losing team.  Of course if you have a lot of bad players, you're not 50-50 to win, and their WPA for the season will eventually reflect that.

Let's take a step back and look at Cole Sands for an example that will be meaningful to Twins fans and probably no other team.  He pitched on Tuesday, and his season ERA is now an impressive 0.73, across 7 games and 12.1 IP.  Is he good?  I dunno.  Obviously it's a small sample.  His WAR on b-r.com stands at 0.4, which is pretty good - multiply everything by 10 for a moderately used pitcher across a full season and you've got a WAR of about 4.0 in 120+ innings which is verging on All-Star performance, so he's going at a good rate right now  This is reflected in the WAA on that same web page, currently 0.6, even higher than his WAR which is unusual.

Even though "Wins" is in the name of both those stats, has Cole actually contributed to any wins this season?  Is he actually above average?  Take a look at his game log. His first three appearances were 2 inning jobs to close out blowout wins (11-1, 11-2, 11-1).  Then he was brought in to soak up some innings in losses where the offense wasn't doing much (7-3, 4-1, 3-0, 5-1).  He's only given up a run once, but whether he pitched well or not, he wasn't likely to do so badly as to blow those three wins, nor will putting up zeroes for an inning or two usually bring a win when you are behind by multiple runs in mid to late innings.  (We all remember the dramatic comebacks or the choke jobs, but overall they are pretty seldom.)  It may not be Cole's fault that he's been used this way, but a rookie needs to establish some trust and his 2022 didn't do that, so in the early going he's been used gingerly.

And WPA (at least IMO) reflects this reality.  The in-game WPA for each of his 7 appearances has been positive, but close to zero.  0.001, etc don't add up to much.  Yesterday was his high-water mark so far, because he was brought in for the fifth inning of a loss instead of late, and even so it only improved his team's chances by 0.017 (again, the team that wins accrues 0.500 together while the losing team accrues -.500).  For the season those little numbers add up to just 0.043, which is about an order of magnitude lower than his WAR or WAA.

In terms of the Twins season record, I think the tiny number more accurately reflects Cole's actual contribution.  WAA would have you believe he's contributed nearly a win more than an average reliever would have done, and I just can't see it.

Okay, I'm belaboring Cole, simply because his season looks so easy to understand - he's had no ups and downs, merely ups, in games where the stakes were low - and the ways of measuring him differ greatly (percentage-wise anyway, given the small sample) which helps highlight how the ways are not the same. Cole's lack of meaningful use stands as contrast to how others are used.

Now compare to game logs for Griffin and Emilio.  They have pitched more, and pitched (at times) when the stakes were high.  Jax was the winning pitcher twice, in extra innings where one mistake could reverse the outcome, and he was credited with a massive .304 WPA each time, justly so.  But he's also blown leads in late innings, resulting in negative WPA of similar or greater magnitude.  Jax has actually had only 6 games out of his 25 appearances where his WPA (positive or negative) was as small as Cole Sands's largest WPA.  Ups and downs come with the territory of being a reliever.  Jax's WAR is 0.0, his WAA is thus slightly negative at -0.1, which suggests mediocrity.  The problem is that bad outings can saddle the team with a loss nearly on the spot, unless the offense stages a miracle comeback - whereas an inning or two of good work won't necessarily preserve the win unless you're the closer because it still leaves some work to the next pitcher in line.  IOW, just subjectively speaking, the downs for a reliever hurt the team more than the ups help.  And this is reflected in Jax's aggregate negative WPA of -.809.  He hasn't destroyed his team's season by any means, but all in all he may have cost us a net loss of a game compared to an average reliever.

Pagan has fared better than Jax.  Some ups, some downs.  6 of his 20 games have resulted in tiny WPA, so relatively he's been used in somewhat lower leverage than Jax, and again the negatives outweigh the positive (a blown save against the Dodgers, a disastrous outing versus Boston) for an aggregate -0.173 - basically the good and bad have cancelled each other.

Circling back to Cole Sands, would he have done any better?  I'm skeptical, but we just don't know, and his near-zero WPA more accurately reflects my uncertainty than his 0.6 Wins Above Average amassed in garbage time.  More than that, none of these stats take into account whether a pitcher was facing the heart of the lineup or the bottom of the order, and we do see Rocco picking his moments for putting in one reliever versus another.

And just for completeness, any stat is likely to confirm that Jhoan Duran has been really good, even if he has suffered a blown save and a ninth-inning loss - WAR is 1.1, WAA is 0.8, and WPA is 0885.  Being the closer is the definition of high-leverage and WPA reflects that - but so does his WAR.  He gets statistical credit with WPA for pitching in high leverage, but it's high leverage for a reason and the downfall would be large if he failed.

Not sure I've directly answered the questions you asked, but maybe this point of view helps you form your own opinions.

18 hours ago, ashbury said:

These are all fair questions and I'm not sure I'm up to the challenge of giving you a definitive answer.  I have opinions, and it's going to be hard to avoid rambling on a topic like this.  I don't know whether enough others are even reading this thread by this point, to make it worthwhile to open a dialog, but I do find it hard to resist potential rabbit holes. 😀

….

Thanks, Ash. This is helpful. It both helped confirm that my understandings weren’t all wet and also added some good nuance for my understanding of WPA. 

And I’m glad to join rabbit holes. My son just told me that his high school English teacher suggested that the title of my son’s autobiography would likely be, “But I Digress…” Said son gets  that honestly, from his father. 

With that in mind, here’s my summary of Pagan’s season. Consider it my contribution to the rabbit hole, but if you find it helpful, I might turn it into its own thread. Let me know if you think it deserves further conversation. 

A.. The debacle in L.A.

B. The game in Boston where Maeda got hurt, and Pagan came in early, down 1-0, and gave up six. As I remember, he had some bad luck, with some soft contact turning into hits. It can’t be argued that he pitched well, but it can be argued that he contributed in that he bought some outs in a game that was going to be tough to win, given the circumstances.

C. A total of 15 games where his WPA was between 0.056 and -0.041. Here’s a breakdown of these games.
C1. In nine of them, he came in in the seventh, with six of them starting the inning and the Twins down one or two and one starting the inning up six. In the other two seventh-inning appearances, he came in with a runner on first, once with one out and up six and once with two outs and down one. In the former, he faced four batters, giving up two hits, getting two outs and letting the inherited run score. In the latter, he also pitched the eighth, giving up a single hit. In those nine games, he pitched 9.0 innings, giving up six hits, a walk and a HBP. He gave up an earned run and an unearned run and struck out 10, This subset of nine games includes Sunday, where he came in, hit a batter, and was pulled because of injury. In two of these nine games, the Twins rallied, and Pagan got the win.
C2. In three of the category C games, he pitched the ninth, coming in down six, down one ahead four. In the first two, he went 1-2-3. In the latter, he gave up a run on two hits. He struck out three in the 3.0 innings.
C3. He came in in the eighth twice, once opening the inning down two, where he gave up a walk and a strikeout in completing the inning, and once with two outs and two on, down three, where he gave up a double to let one score, followed by a groundout and a one-hit ninth.
C4. In the other, he started the sixth down four and walked two and struck out two, allowing no runs.
D. A game with a WPA of .105. He entered a tie game in the fifth with two outs and a runner on first. He got the out and then pitched a one-hit sixth.
E. A game with a WPA of .120. With the Twins down 1-0, Maeda gave a lead off single in the sixth. Pagan gave up a walk, got a fly ball and a double play. He then pitched the seventh, with a fly ball and two strikeouts.
F.  A game with a WPA of .304. That was the one where he pitched the bottom of the 11th in Chicago after the Twins hadn’t scored in the top. He got a strikeout, gave up a fly ball, intentionally walked Benintendi and struck out Hamilton. The Twins scored four in the 12th and Pagan got the win. 

So as I look at this, the dominant thing I see is that that Rocco’s preferred usage is to bring him in in the seventh, with the team still in the game (C1). He doesn’t like to use Lopez or Jax without them being tied or ahead, so he goes to Pagan, and Pagan has done his job. He’s also used when down in the eighth and in blowouts in the ninth.

In summary, I see 15 games in the WPA middle. Of those, 12 have a positive WPA and three have a negative. In two of the negatives, he did his job, in that the job was “don’t screw this up.” The third was Sunday, which doesn’t really count, so I’d suggest that in 14 of 14 games, he did his job, often times perfectly, other times very well, and a couple times so-so.

And on five occasions, he’s been used in higher-leverage situations, generally out of necessity because other options had been used or were otherwise unavailability. In one (A) he had a major fail, in three (D-F), he was successful, and in one (B), he was in-between, not pitching well, but given the context, playing a needed role. I’d call that one a tie and say he was 3-1-1, but I’m fine calling it a loss as well. 

So in total, that means that in 17 or 18 out of 19 games, he’s done his job. But because of one game in LA and the memory of last year, most on TD are not ready (or more accurately, willing) to give him his due.