The Red Sox have finalized their roster for opening day, according to Gordon Edes at ESPNBoston. There aren't any big surprises, as the Red Sox have gone with maintaining depth. For the final two bullpen spots, Dennys Reyes (L) and Matt Albers (R) won out. Both pitched very well in spring training and were out of options. In order to keep Alfredo Aceves or Hideki Okajima, the Red Sox would have had to expose Albers or Reyes to waivers. Aceves pitched fairly well, and will be stretched out as a starter in AAA to provide some much needed starting pitching depth. With Felix Doubront hitting the DL with an elbow injury, Aceves is the Red Sox first option should one of their front five go down. Papelbon, Bard, Jenks, Wakefield, and Wheeler round out the rest of the bullpen.
Thanks to Jed Lowrie's flexibility, the Red Sox are only carrying one back up infielder. That should ensure that he's able to get semi-regular at bats to start the season. It also lets the Red Sox keep both Mike Cameron and Darnell McDonald as back up outfielders. I'd expect Cameron to get fairly regular AB against lefties, spelling Drew and Ellsbury, while McDonald is more of a typical backup getting only a handful of at bats in blow outs. Having two guys who can man centerfield on the bench is a nice situation to have. Ryan Kalish and Josh Reddick, McDonald's only real competition for the final outfield spot, were both sent down to minor league camp a few days ago. As with the relievers, because Kalish and Reddick have options remaining, the Red Sox are able to hang on to more assets than if they exposed McDonald to waivers. Perhaps more importantly, this will ensure that both Reddick and Kalish get regular at bats so they can continue to improve.
Monday, March 28, 2011
Saturday, March 19, 2011
What makes a prospect successful?
My brother Andrew and I were recently having a discussion about prospects, and whether or not left handed hitting prospects struggled more than right handed hitters. Mike Moustakas' difficulties against left handed hitters worried Andrew, and even managed to conjure up the ghost of Alex Gordon. His thinking was that right handed hitters have to learn how to hit right handed pitchers, otherwise they wouldn't succeed, even in the minors. Left handed hitters, on the other hand, couldget by without learning to hit left handed pitching. There is also fewer quality left handed pitching in the minors, particularly when it comes to velocity, and when suddenly exposed to major league lefties, a left handed prospect could struggle.
I established my prospect pool using Baseball America's top 25 rankings from 1995-2005. I only went up to 2005 to ensure that any top 25 prospects had a chance to become established major league players. Because our original discussion focused on top prospects like Alex Gordon and Mike Moustakas, I only selected the top 25. Once pitchers were excluded and the list was pared down to account for players appearing in multiple years, I was left with a total of 109 batters - 59 righties, 37 lefties, and 13 switch hitters.
To measure their success in the majors, I went with Wins Above Replacement (WAR), which is readily available from Fangraphs. Although these values include defense, I didn't want to exclude defense entirely, as a prospect may have made the list based on their defensive value. If, for example, right handed hitters have more of their value in defense because all shortstops are right handed, this difference could incorrectly be attributed to some difference in the hitting of prospects. I looked at WAR three different ways - the average annual WAR for their career, the WAR for their best year, and the total WAR accumulated for their career. I considered a major league career to have started the first season the player had at least 100 at bats, to reduce the impact of "cups of coffee", particularly on average WAR. Here are the average values.
As you can see, there is not much of a difference between left handed and right handed prospects. Interestingly, top prospects who are switch hitters seem to have a much higher rate of success than either right handed or left handed batters. To see if extreme examples influence the averages, we need to look at the breakdown of successful players. I categorized players by average WAR for their career, and defined anyone with an average WAR of less than zero as a washout, between zero and one as a scrub, one to three as a starter, and anyone who average WAR of greater than 3 as a star. Some examples, Nomar Garciaparra and Miguel Tejada are just on the star side of the starter/star threshold, thanks to some poor decline years for both. Players who easily end up in the star category should be familiar to any baseball fan – Carlos Beltran, Alex Rodriguez, and David Wright. Starters include players like Johnny Damon, Darren Erstad, Carlos Pena, and BJ Upton. Scrubs are players like Jay Payton, Jeff Francouer, Wilson Betemit, and Todd Walker - although they've started and succeeded for a couple years in their career, for the most part they are bench players. Washouts either never made it, like Drew Henson, or were terrible in the big leagues, like Andy Marte.
Again, we see no major differences between left handers and right handers, but top prospect switch hitters have an incredible track record. The sample size is very small (n=13), but top tier prospects that are switch hitters have an incredible track record.
Of course, I would be completely remiss if I didn't actually run any statistics to see if these perceived differences were borne out. After all, I'd definitely dock my students a few points for saying, "See! They're different!" without any statistical support.
Warning! Stats ahead!
An analysis of variance (ANOVA) is the quick and easy way to tell if the handedness of a batter impacts any of the WAR measurements we used. Handedness was not a significant predictor for either total WAR or peak WAR (p=0.16 and 0.14 respectively, where a difference is significant if p is less than 0.05). Once we know that there is a significant effect of handedness on average WAR, we can use a Tukey's Honestly Significant Difference Test, which shows that switch hitters in Baseball America's top 25 prospects have significantly better results than left handed or right handed hitters, but there is no difference between left handed and right handed hitters.
End of Stats
Now that you've made it through the statistics (or skimmed, don't worry, there won't be a quiz at the end), we can take a look at the pretty pictures instead!
A quick refresher on box plots: the line in the middle of the box is the median for the group. The top and bottom of the box are the 75th and 25th percentile, respectively (also called the first and third quartile) and the lines represent the range of the remaining data. Outliers are shown as small circles, but there is only one - Alex Rodriguez and his average of 6.75 wins above replacement. I like to imagine that little circle looking down on everyone else while being fed popcorn, just like at the Super Bowl!
So what exactly does this all mean for prospects? Are switch hitting prospects more likely to succeed? It is one possible explanation, but it certainly isn't proven with this analysis. One alternative explanation is that for a switch hitter to be listed in Baseball America's top 25, they have to be pretty damn good. Switch hitters often first start switch hitting in high school or later, so the non-dominant side can lag behind in development. Perhaps if a switch hitter is playing well enough to make Baseball America's top 25, they're already a fairly polished player.
Either way, a switch hitter on the top prospect list seems to have a great shot at becoming an impact player. In the last two years there have only been three switch hitters in the top 25 – Carlos Santana, Justin Smoak, and Aaron Hicks. I'm sure this article is a small consolation to the Mariners and their fans after the Cliff Lee trade, when they ended up with the switch hitter Justin Smoak (BA's #13 prospect for 2010 season) who struggled mightily in Seattle instead of right handed Jesus Montero (#4 in 2010, #3 in 2011).
Let me know if you have any other questions that might be answerable using this database. A comparison study using pitchers would be an interesting follow up, as would expanding the prospect pool to the top 50 or beyond.
I established my prospect pool using Baseball America's top 25 rankings from 1995-2005. I only went up to 2005 to ensure that any top 25 prospects had a chance to become established major league players. Because our original discussion focused on top prospects like Alex Gordon and Mike Moustakas, I only selected the top 25. Once pitchers were excluded and the list was pared down to account for players appearing in multiple years, I was left with a total of 109 batters - 59 righties, 37 lefties, and 13 switch hitters.
To measure their success in the majors, I went with Wins Above Replacement (WAR), which is readily available from Fangraphs. Although these values include defense, I didn't want to exclude defense entirely, as a prospect may have made the list based on their defensive value. If, for example, right handed hitters have more of their value in defense because all shortstops are right handed, this difference could incorrectly be attributed to some difference in the hitting of prospects. I looked at WAR three different ways - the average annual WAR for their career, the WAR for their best year, and the total WAR accumulated for their career. I considered a major league career to have started the first season the player had at least 100 at bats, to reduce the impact of "cups of coffee", particularly on average WAR. Here are the average values.
| Handedness |
| WAR | ||
|
| Count | Average | Peak | Total |
| Left | 37 | 1.75 | 4.13 | 15.98 |
| Right | 59 | 1.59 | 3.59 | 17.48 |
| Switch | 13 | 3.20 | 5.22 | 28.82 |
As you can see, there is not much of a difference between left handed and right handed prospects. Interestingly, top prospects who are switch hitters seem to have a much higher rate of success than either right handed or left handed batters. To see if extreme examples influence the averages, we need to look at the breakdown of successful players. I categorized players by average WAR for their career, and defined anyone with an average WAR of less than zero as a washout, between zero and one as a scrub, one to three as a starter, and anyone who average WAR of greater than 3 as a star. Some examples, Nomar Garciaparra and Miguel Tejada are just on the star side of the starter/star threshold, thanks to some poor decline years for both. Players who easily end up in the star category should be familiar to any baseball fan – Carlos Beltran, Alex Rodriguez, and David Wright. Starters include players like Johnny Damon, Darren Erstad, Carlos Pena, and BJ Upton. Scrubs are players like Jay Payton, Jeff Francouer, Wilson Betemit, and Todd Walker - although they've started and succeeded for a couple years in their career, for the most part they are bench players. Washouts either never made it, like Drew Henson, or were terrible in the big leagues, like Andy Marte.
| Washout | Scrub | Starter | Star | |
| Avg. WAR | <0 | 0-1.5 | 1.5-3 | 3+ |
| Left | 14% | 41% | 27% | 19% |
| Right | 25% | 34% | 22% | 19% |
| Switch | 0% | 23% | 23% | 54% |
Again, we see no major differences between left handers and right handers, but top prospect switch hitters have an incredible track record. The sample size is very small (n=13), but top tier prospects that are switch hitters have an incredible track record.
Of course, I would be completely remiss if I didn't actually run any statistics to see if these perceived differences were borne out. After all, I'd definitely dock my students a few points for saying, "See! They're different!" without any statistical support.
Warning! Stats ahead!
An analysis of variance (ANOVA) is the quick and easy way to tell if the handedness of a batter impacts any of the WAR measurements we used. Handedness was not a significant predictor for either total WAR or peak WAR (p=0.16 and 0.14 respectively, where a difference is significant if p is less than 0.05). Once we know that there is a significant effect of handedness on average WAR, we can use a Tukey's Honestly Significant Difference Test, which shows that switch hitters in Baseball America's top 25 prospects have significantly better results than left handed or right handed hitters, but there is no difference between left handed and right handed hitters.
End of Stats
Now that you've made it through the statistics (or skimmed, don't worry, there won't be a quiz at the end), we can take a look at the pretty pictures instead!
A quick refresher on box plots: the line in the middle of the box is the median for the group. The top and bottom of the box are the 75th and 25th percentile, respectively (also called the first and third quartile) and the lines represent the range of the remaining data. Outliers are shown as small circles, but there is only one - Alex Rodriguez and his average of 6.75 wins above replacement. I like to imagine that little circle looking down on everyone else while being fed popcorn, just like at the Super Bowl!So what exactly does this all mean for prospects? Are switch hitting prospects more likely to succeed? It is one possible explanation, but it certainly isn't proven with this analysis. One alternative explanation is that for a switch hitter to be listed in Baseball America's top 25, they have to be pretty damn good. Switch hitters often first start switch hitting in high school or later, so the non-dominant side can lag behind in development. Perhaps if a switch hitter is playing well enough to make Baseball America's top 25, they're already a fairly polished player.
Either way, a switch hitter on the top prospect list seems to have a great shot at becoming an impact player. In the last two years there have only been three switch hitters in the top 25 – Carlos Santana, Justin Smoak, and Aaron Hicks. I'm sure this article is a small consolation to the Mariners and their fans after the Cliff Lee trade, when they ended up with the switch hitter Justin Smoak (BA's #13 prospect for 2010 season) who struggled mightily in Seattle instead of right handed Jesus Montero (#4 in 2010, #3 in 2011).
Let me know if you have any other questions that might be answerable using this database. A comparison study using pitchers would be an interesting follow up, as would expanding the prospect pool to the top 50 or beyond.
Monday, March 7, 2011
The 2011 Red Sox Pitching Staff
So, yeah, the time frame for getting my weighted mean pitching projections was a little bit off - a few days, a couple months, no big difference. Fortunately, there weren't a ton of baseball stories in the meantime and it gave my gingerbread Fenway plenty of time at the top.
Just as with the hitting statistics, I’m using a 5-4-3 weighted mean for the IP, ERA, FIP, and xFIP and am not using any kind of park effects. Depending on which stat you would like to believe (ERA, FIP, xFIP) and whether or not you’d like to believe my rather basic projection system in the first place, the Red Sox pitchers should allow between 637 and 665 runs. That would place them between 5th and 8th in the AL. None of these numbers are defense adjusted. While it would be difficult to make defensive projections with Youkilis moving to 3B for the first time in several years, Ellsbury moving back to center field from left, and Carl Crawford moving to left, but with questions about how he’ll maintain his defensive value in Fenway thanks to the Green Monster, I feel comfortable saying that the defense should be above average.
Another quick note, there are replacement level innings in the mix as well. Games where you lose on the road (and therefore don’t pitch in the bottom of the 9th) and extra innings games about come out in the wash, so I used the nice round number of 1450 innings pitched. Based on who I expected to pitch (and an additional reliever to account for the impressive bullpen depth the Red Sox have accumulated), the innings totaled up to 1344. This left 106 innings of replacement level pitching, split evenly between starting and relief pitching. The replacement level ERA was based on a study by Fangraphs (AL reliever: 4.40, AL starter: 5.63), but may be a bit high if the depressed 2010 offensive levels represent a new run scoring environment as it was written a few years ago.
Without further ado, here are your Red Sox pitchers:
A few thoughts on the projections...
1. The exact identity of the left handed pitcher (or pitchers) in the bullpen isn’t a huge deal. Because they’re throwing 50-60 innings and have very similar ERAs, FIPs, and xFIPS swapping out Okajima for other pitchers doesn’t make a big difference in the overall number. The amount of variation surrounding expectations for a relievers performance is easily dwarfed by any kind of variability between pitchers in the back end of the bullpen.
2. Perhaps the biggest impact will be the depth of the bullpen arms available in AAA. Thanks to Andrew Miller’s interesting contract, which causes his option for next year to vest if he is picked up on waivers, the Red Sox can safely stash him in AAA despite the fact that he no longer has options remaining. This was possible because Miller wants to stay with the Red Sox and believe they give him the best shot at becoming a Major League caliber pitcher again.
3. Hopefully the starters can throw more than 800 innings. The weighted means penalize Buchholz for not being an established pitcher until 2010, but with an innings total of 179 last year, Buccholz should have a limit of around 200 or 210 innings, if he has one at all. Beckett and Matsuzaka are hardly sure things, but having a swing man like Wakefield who can throw 150 innings if needed is massive. However, Wakefield is getting older and can’t be relied upon quite as often as he used to. Perhaps with an eye towards a post-Wakefield ERA, the Red Sox snagged Alfredo Aceves off of waivers. While his stuff isn’t spectacular, Aceves has succeeded in both the bullpen and as a starter and could provide a nice insurance policy on Wakefield’s back. As an added bonus, the Yankees let him go and are now basically having to hold open tryouts for their 4th and 5th starting spots.
4. Overall, FIP and xFIP don’t seem to differ substantially from ERA projections, although they do for specific players. Buchholz is probably the most interesting of the bunch; he posted an ERA of 2.33, despite having underlying peripherals that indicate an ERA much closer to 4 (3.62 FIP, 4.20 xFIP). According to scouts, the stuff is still there, but in order to continue to post an ace level ERA, he’ll need to start striking out more batters. Jon Lester made a similar leap between the 2008 season where he was seemingly pitching over his head (3.32 ERA, 4.08 xFIP, 6.5 K/9) to his 2009 breakout (3.41 ERA, 3.13 xFIP, 9.96 K/9). It seems a bit odd talking about a breakout year when a player’s ERA gets worse, but if you saw Lester in 2008 and 2009 or 2010, the improvement is obvious. These types of breakthroughs don’t happen all that often and it would be a huge credit to the Red Sox coaches if they managed to convert two “lucky” pitchers into top tier starters in 5 years.
Just as with the hitting statistics, I’m using a 5-4-3 weighted mean for the IP, ERA, FIP, and xFIP and am not using any kind of park effects. Depending on which stat you would like to believe (ERA, FIP, xFIP) and whether or not you’d like to believe my rather basic projection system in the first place, the Red Sox pitchers should allow between 637 and 665 runs. That would place them between 5th and 8th in the AL. None of these numbers are defense adjusted. While it would be difficult to make defensive projections with Youkilis moving to 3B for the first time in several years, Ellsbury moving back to center field from left, and Carl Crawford moving to left, but with questions about how he’ll maintain his defensive value in Fenway thanks to the Green Monster, I feel comfortable saying that the defense should be above average.
Another quick note, there are replacement level innings in the mix as well. Games where you lose on the road (and therefore don’t pitch in the bottom of the 9th) and extra innings games about come out in the wash, so I used the nice round number of 1450 innings pitched. Based on who I expected to pitch (and an additional reliever to account for the impressive bullpen depth the Red Sox have accumulated), the innings totaled up to 1344. This left 106 innings of replacement level pitching, split evenly between starting and relief pitching. The replacement level ERA was based on a study by Fangraphs (AL reliever: 4.40, AL starter: 5.63), but may be a bit high if the depressed 2010 offensive levels represent a new run scoring environment as it was written a few years ago.
Without further ado, here are your Red Sox pitchers:
| IP | ERA | FIP | xFIP | ||
| SP 1 | Lester | 207 | 3.29 | 3.27 | 3.44 |
| SP 2 | Beckett | 167.33 | 4.51 | 3.82 | 3.53 |
| SP 3 | Buchholz | 122 | 3.49 | 4.07 | 4.18 |
| SP 4 | Lackey | 189 | 4.08 | 3.96 | 4.10 |
| SP 5 | Matsuzaka | 125.33 | 4.26 | 4.21 | 4.74 |
| Swing | Wakefield | 146.67 | 4.74 | 4.64 | 5.16 |
| Closer | Papelbon | 68 | 2.82 | 2.97 | 3.47 |
| SU | Bard | 63 | 2.53 | 3.37 | 3.47 |
| RH | Jenks | 55 | 3.7 | 3.43 | 3.31 |
| RH | Atchison | 60 | 4.29 | 4.38 | 4.01 |
| LH | Okajima | 60 | 3.56 | 4.19 | 4.55 |
| RH | Albers | 25 | 4.66 | 4.2 | 4.65 |
| RH | Wheeler | 55.67 | 3.26 | 4.35 | 4.18 |
| Fill | Replacement | 106 | 5.15 | 5.15 | 5.15 |
A few thoughts on the projections...
1. The exact identity of the left handed pitcher (or pitchers) in the bullpen isn’t a huge deal. Because they’re throwing 50-60 innings and have very similar ERAs, FIPs, and xFIPS swapping out Okajima for other pitchers doesn’t make a big difference in the overall number. The amount of variation surrounding expectations for a relievers performance is easily dwarfed by any kind of variability between pitchers in the back end of the bullpen.
2. Perhaps the biggest impact will be the depth of the bullpen arms available in AAA. Thanks to Andrew Miller’s interesting contract, which causes his option for next year to vest if he is picked up on waivers, the Red Sox can safely stash him in AAA despite the fact that he no longer has options remaining. This was possible because Miller wants to stay with the Red Sox and believe they give him the best shot at becoming a Major League caliber pitcher again.
3. Hopefully the starters can throw more than 800 innings. The weighted means penalize Buchholz for not being an established pitcher until 2010, but with an innings total of 179 last year, Buccholz should have a limit of around 200 or 210 innings, if he has one at all. Beckett and Matsuzaka are hardly sure things, but having a swing man like Wakefield who can throw 150 innings if needed is massive. However, Wakefield is getting older and can’t be relied upon quite as often as he used to. Perhaps with an eye towards a post-Wakefield ERA, the Red Sox snagged Alfredo Aceves off of waivers. While his stuff isn’t spectacular, Aceves has succeeded in both the bullpen and as a starter and could provide a nice insurance policy on Wakefield’s back. As an added bonus, the Yankees let him go and are now basically having to hold open tryouts for their 4th and 5th starting spots.
4. Overall, FIP and xFIP don’t seem to differ substantially from ERA projections, although they do for specific players. Buchholz is probably the most interesting of the bunch; he posted an ERA of 2.33, despite having underlying peripherals that indicate an ERA much closer to 4 (3.62 FIP, 4.20 xFIP). According to scouts, the stuff is still there, but in order to continue to post an ace level ERA, he’ll need to start striking out more batters. Jon Lester made a similar leap between the 2008 season where he was seemingly pitching over his head (3.32 ERA, 4.08 xFIP, 6.5 K/9) to his 2009 breakout (3.41 ERA, 3.13 xFIP, 9.96 K/9). It seems a bit odd talking about a breakout year when a player’s ERA gets worse, but if you saw Lester in 2008 and 2009 or 2010, the improvement is obvious. These types of breakthroughs don’t happen all that often and it would be a huge credit to the Red Sox coaches if they managed to convert two “lucky” pitchers into top tier starters in 5 years.
Subscribe to:
Posts (Atom)