Rookies and the YTD Drag
This is the final Sunday before we debut a new weekly schedule report, the first of which will run in this space a week from today. I thought I'd take advantage of the open slot to talk a bit about rookies who struggle in April and the threat an early slump has on their playing time.
Just before I get to today's main topic, I want to clarify or remind readers about our minor league translations, available to certain subscribers in our members only section. A reader wrote to ask about the translation for Yankees' pitcher Chase Wright, pointing out that the translated ERA showed 0.00 but that the reader had personally watched (and as a Red Sox fan enjoyed) Wright giving up four home runs in the majors recently.
I wrote the reader back directly to clarify several items which other readers may also need to keep in mind: (a) the minor league translation is a translation of only minor league statistics and does not include nor incorporate any major league performance, (b) the minor league translation is not a projection of ability. It's simply a conversion of a player's minor league stats-to-date into a neutral big league context. A pitcher could easily be a bad pitcher but have a good translated ERA - of course, that good translated ERA won't hold up in the long run but it doesn't lessen his minor league performance to date... and (c) the minor league stats that are considered are through play completed the Sunday prior to the Friday morning publication of the translations.
I suspect for readers who have subscribed to the translations for a long time, none of this is news but for those who are new to the feature or were uncertain, I wanted to offer those additional clarifications. I made a few extra comments about the translations in the debut of our new Thursday translated leaders report, which began running this past week.
Now, in the main issue of the day, I always remind readers throughout the off-season that the most solid plans of a team can quickly be abandoned if a rookie goes out and goes 0 for April. In fact, during spring training, I happened to use Kevin Kouzmanoff as an example of a player with a good projection but the highest uncertainty ratings I assign out. As fate would have it, Kouzmanoff has proven to be just one of those types of players. While the uncertainty of a rookie forecast is somewhat rooted in the uncertainty of ability, it's more often tied to the uncertainty of whether a rookie will play well enough immediately to actually hang on to the job he won in spring training.
Many readers have noticed that I've significantly downgraded Kouzmanoff's forecast the past two weeks, more a week ago than in this weekend's edition of the projections but in both cases in a downward spiral. It's not that I've downgraded his ability much, though I had to at least somewhat downgrade him given the severity of the terrible start. In fact, the weight of evidence supporting the theory of his ability still isn't outweighed by even this disastrous April he's suffered through.
Still, we can't rule out that he could end up as a Kirk Saarloos type player, who was a pitcher who made it as high as #12 on one of my annual prospect lists and who has never approached major league hitters as he did minor league hitters, even though he's managed to stick around. That is, there are players who look different in the majors when they get their shot, who actually approach the game differently than they did in their minor league successful seasons. I don't know if it's nerves or something else but it's often difficult to reconcile how such players can look so comfortable in, say, spring training against veteran major leaguers but then go out in front of the big crowds and collapse.
In any case, it's not so much that I want to talk about rookies and disappointment. There are so many rookies off to terrible starts this year, some of whom were considered "can't miss" by many of the prognosticators.
But in Kouzmanoff's case, if I can use him as the best example of a phenomenon (even more than Alex Gordon), there is now an effect on his performance that I have in past years referred to as the "year-to-date drag" if you will. This year-to-date drag is the effect that if you slump early, it makes your season look worse than if you slump late, even if in total you do exactly the same thing. I used a graphic example of this in the archived essay On Fantasy Trading in which I showed in May of 2005 that the perception of Andruw Jones' season at that time was skewed by his slump coming in April rather than in May. In other words, if you reversed the order of a season, Jones would have seemed to be having a better year even though in sum he would have produced exactly the same totals.
Particularly when it comes to rookies, here's the problem of what the drag does - It costs them their job because even if they're good enough to play, their season feels intolerable to management no matter how hot they eventually get. Such players usually find themselves back in the minors by May and I suspect that Kouzmanoff, unfortunately, is now close to that. I hope I'm wrong because I still believe he's a better player than we've seen but I have to set aside hope and go with the evidence.
To illustrate this point, I want to again use simulation modeling, much as I did the other day in this space in reference to David Wright. The simulation example seemed to generate much positive feedback because readers said it graphically illustrated something that they were already wondering about. I'll use Kouzmanoff in today's example since he's the rookie who's arguably been getting the most opportunity and been falling the most short of anyone's expectations.
Through play completed on Saturday, Kouzmanoff is now hitting just .127 in 63 at bats on the season. More specifically, he has 8 hits in those 63 at bats. Before we talk about the drag, let's ask ourselves this question? If our Opening Day theory was correct that he is actually a .286 hitter, what are the chances that he would have only 8 hits after 63 at bats?
Sadly, the chances are extremely slim. In 10,000 simulations of a .286 hitter getting 63 at bats, just 18 of them or 1 in about 600 simulated seasons resulted in a hitter hitting only .127 or worse over that stretch. Thus, this explains at least the partial downgrade of Kouzmanoff in recent days and I've now reduced him to a .268 hitter and this number continues to drop. In other words, the poor performance is outside of the margin of error enough that we had to downgrade him. There are cases where players perform outside of the margin because of bad luck (a margin after all is only right a certain percentage of the time) and more often, because they're hiding or playing through an injury. We do know that Kouzmanoff was nursing a minor elbow injury in mid-April but it didn't seem enough to explain this extended slump.
Now here comes the worse news. Let's say that he's really a .300 hitter and just suffered through a miserable stretch of bad luck or something else. I'm only floating this because now expecting him to be a .300 hitter as his real hidden ability would seem to be on the highest end of anyone's updated expectations. Let's use a simulation but this time, let's mix in the year-to-date stats. You see, those year-to-date stats can't be erased and so even if Kouzmanoff happens to go on a tear, he may not be able to hit enough to hang on to his job for much longer.
In fact, no hitter in all of baseball last year managed to finish with more than 100 at bats and an average of under .180. It doesn't mean that no player had an average of below .180 after 100 at bats but it does mean that the longer you slump, the more likely you are to lose your job.
So, in framing this simulation, I'm going to speculate that if Kouzmanoff is hitting below .180 as late as May 5th or 10th or so, he will lose at least his starting spot at third base if not his roster spot. Now, if we pretend he's a .300 hitter to give him the most benefit of the doubt and we mix in this terrible start so far, here's the breakdown of 10,000 simulated seasons where we give him 37 more at bats for an even 100 total on the season:
Just before I get to today's main topic, I want to clarify or remind readers about our minor league translations, available to certain subscribers in our members only section. A reader wrote to ask about the translation for Yankees' pitcher Chase Wright, pointing out that the translated ERA showed 0.00 but that the reader had personally watched (and as a Red Sox fan enjoyed) Wright giving up four home runs in the majors recently.
I wrote the reader back directly to clarify several items which other readers may also need to keep in mind: (a) the minor league translation is a translation of only minor league statistics and does not include nor incorporate any major league performance, (b) the minor league translation is not a projection of ability. It's simply a conversion of a player's minor league stats-to-date into a neutral big league context. A pitcher could easily be a bad pitcher but have a good translated ERA - of course, that good translated ERA won't hold up in the long run but it doesn't lessen his minor league performance to date... and (c) the minor league stats that are considered are through play completed the Sunday prior to the Friday morning publication of the translations.
I suspect for readers who have subscribed to the translations for a long time, none of this is news but for those who are new to the feature or were uncertain, I wanted to offer those additional clarifications. I made a few extra comments about the translations in the debut of our new Thursday translated leaders report, which began running this past week.
Now, in the main issue of the day, I always remind readers throughout the off-season that the most solid plans of a team can quickly be abandoned if a rookie goes out and goes 0 for April. In fact, during spring training, I happened to use Kevin Kouzmanoff as an example of a player with a good projection but the highest uncertainty ratings I assign out. As fate would have it, Kouzmanoff has proven to be just one of those types of players. While the uncertainty of a rookie forecast is somewhat rooted in the uncertainty of ability, it's more often tied to the uncertainty of whether a rookie will play well enough immediately to actually hang on to the job he won in spring training.
Many readers have noticed that I've significantly downgraded Kouzmanoff's forecast the past two weeks, more a week ago than in this weekend's edition of the projections but in both cases in a downward spiral. It's not that I've downgraded his ability much, though I had to at least somewhat downgrade him given the severity of the terrible start. In fact, the weight of evidence supporting the theory of his ability still isn't outweighed by even this disastrous April he's suffered through.
Still, we can't rule out that he could end up as a Kirk Saarloos type player, who was a pitcher who made it as high as #12 on one of my annual prospect lists and who has never approached major league hitters as he did minor league hitters, even though he's managed to stick around. That is, there are players who look different in the majors when they get their shot, who actually approach the game differently than they did in their minor league successful seasons. I don't know if it's nerves or something else but it's often difficult to reconcile how such players can look so comfortable in, say, spring training against veteran major leaguers but then go out in front of the big crowds and collapse.
In any case, it's not so much that I want to talk about rookies and disappointment. There are so many rookies off to terrible starts this year, some of whom were considered "can't miss" by many of the prognosticators.
But in Kouzmanoff's case, if I can use him as the best example of a phenomenon (even more than Alex Gordon), there is now an effect on his performance that I have in past years referred to as the "year-to-date drag" if you will. This year-to-date drag is the effect that if you slump early, it makes your season look worse than if you slump late, even if in total you do exactly the same thing. I used a graphic example of this in the archived essay On Fantasy Trading in which I showed in May of 2005 that the perception of Andruw Jones' season at that time was skewed by his slump coming in April rather than in May. In other words, if you reversed the order of a season, Jones would have seemed to be having a better year even though in sum he would have produced exactly the same totals.
Particularly when it comes to rookies, here's the problem of what the drag does - It costs them their job because even if they're good enough to play, their season feels intolerable to management no matter how hot they eventually get. Such players usually find themselves back in the minors by May and I suspect that Kouzmanoff, unfortunately, is now close to that. I hope I'm wrong because I still believe he's a better player than we've seen but I have to set aside hope and go with the evidence.
To illustrate this point, I want to again use simulation modeling, much as I did the other day in this space in reference to David Wright. The simulation example seemed to generate much positive feedback because readers said it graphically illustrated something that they were already wondering about. I'll use Kouzmanoff in today's example since he's the rookie who's arguably been getting the most opportunity and been falling the most short of anyone's expectations.
Through play completed on Saturday, Kouzmanoff is now hitting just .127 in 63 at bats on the season. More specifically, he has 8 hits in those 63 at bats. Before we talk about the drag, let's ask ourselves this question? If our Opening Day theory was correct that he is actually a .286 hitter, what are the chances that he would have only 8 hits after 63 at bats?
Sadly, the chances are extremely slim. In 10,000 simulations of a .286 hitter getting 63 at bats, just 18 of them or 1 in about 600 simulated seasons resulted in a hitter hitting only .127 or worse over that stretch. Thus, this explains at least the partial downgrade of Kouzmanoff in recent days and I've now reduced him to a .268 hitter and this number continues to drop. In other words, the poor performance is outside of the margin of error enough that we had to downgrade him. There are cases where players perform outside of the margin because of bad luck (a margin after all is only right a certain percentage of the time) and more often, because they're hiding or playing through an injury. We do know that Kouzmanoff was nursing a minor elbow injury in mid-April but it didn't seem enough to explain this extended slump.
Now here comes the worse news. Let's say that he's really a .300 hitter and just suffered through a miserable stretch of bad luck or something else. I'm only floating this because now expecting him to be a .300 hitter as his real hidden ability would seem to be on the highest end of anyone's updated expectations. Let's use a simulation but this time, let's mix in the year-to-date stats. You see, those year-to-date stats can't be erased and so even if Kouzmanoff happens to go on a tear, he may not be able to hit enough to hang on to his job for much longer.
In fact, no hitter in all of baseball last year managed to finish with more than 100 at bats and an average of under .180. It doesn't mean that no player had an average of below .180 after 100 at bats but it does mean that the longer you slump, the more likely you are to lose your job.
So, in framing this simulation, I'm going to speculate that if Kouzmanoff is hitting below .180 as late as May 5th or 10th or so, he will lose at least his starting spot at third base if not his roster spot. Now, if we pretend he's a .300 hitter to give him the most benefit of the doubt and we mix in this terrible start so far, here's the breakdown of 10,000 simulated seasons where we give him 37 more at bats for an even 100 total on the season:
The hits column refers to how many hits he gets in the simulation in 37 more at bats. The "NEW AVG" column is what his new batting average would be at the end of the 100 at bats, including the terrible start he's had so far. So, using this simulation, even if we let Kouzmanoff be treated as a .300 hitter, almost half of the simulations result in him still having an average below .180 at the end of that stretch.
By the way, a 21 for 37 stretch like that which happened twice in our 10,000 simulations are possible. Nick Markakis last year was hitting in the low .200s into early June but managed to raise his average up in one incredible stretch in June and July that saw him turn around his season and end up with a .291 average. The Orioles stuck with him and he delivered but comebacks like that are rare, not only because most players aren't capable of them but because most players who slump as badly as Markakis did don't get the chance to turn it around.So now let's put in our latest theory that Kouzmanoff's actually a .268 hitter (a downgrade of almost 20 points from the Opening Day set). Here then is how his first 100 at bats would play out then, again including the year-to-date drag:
So, even if our latest theory of his ability is correct, almost 60% of the simulations result in him still having an average below .180 at the end of 100 at bats. This is why I often will drop a player's playing time forecast even if I haven't entirely downgraded the projection of his ability. We must consider: If Kouzmanoff really is hitting .180 or worse after another 35 or 40 at bats, is it reasonable to imagine the Padres sticking with him much longer? I suspect not and that's why playing time reductions often happen before the most severe of ability downgrades.
Ironically, the year-to-date effect can work in a rookie's favor. A rookie who goes out and hits .300 in April can suffer through a miserable May and keep his job. This is because that great April performance boosts the appearance of his season. In baseball, ironically, it's the opposite of "what have you done for me lately?" While players are all noticed when they're on a hot streak, the earlier any performance occurs, the more it confuses our perception of a player's season... and more importantly, the more it influences managerial decisions about whether to continue giving that player opportunities.
posted by David Luciani at
6:19 AM
