I made a dumb mistake in my last post; for some reason, I decided that the inverse of K/9 was 9/(K/9), instead of 1/(K/9). Thanks to Stephen Nelson for pointing this out, and I've replaced that statistic in the post with IP/K, because it is also best when low (you'll notice that the IP/K numbers are the same as they were before; the calculations didn't change, but the explanation did).
Update: Yet another stupid mistake fixed. Ah, well...
Friday, March 26, 2004
Congratulations, you won't have to suffer through any more posts concerning Franklin's solo home run rate, the correllation between solo home runs and at bats with nobody on base, or divisional rate stat adjustment. These are each relatively complicated topics that deserve more attention and discussion than they receive on a low-grade blog. Topics such as these are best left to the experts who know what to do with them, and, as such, I will no longer spend hours creating Excel spreadsheets that provide data I don't fully understand.
...you didn't think I was serious, did you?
Quite the contrary! In fact, prepare yourselves for yet another column dealing with quantitative analysis. However, this time I think the points of discussion will be much easier to understand, and I'm cautiously optimistic that what I'm about to write will generate some kind of positive feedback.
I assume that everyone reading this is familiar with OPS and the perils of reading too much into it. As a quick-and-dirty metric, it's fairly serviceable, and it has made its way into various sports columns in newspapers around the country as journalists attempt to keep up with the ever-changing field of quantitative analysis while keeping things simple enough for their readership. However, the metric's simplicity is part of the problem. Check out this hypothetical example:
Player A: .260/.400/.450
Player B: .260/.330/.520
Both players will have identical OPS' (.850), even though Player A was more valuable than Player B. This is the main problem with OPS, but don't fret! If you're determined to use a similar statistic for player comparison, there's a relatively simple, nice way around the issue. The league-average line last year was .264/.333/.422, with an OPS of .755. But you can adjust this figure: the average SLG was 1.27 times the average OBP, so you can perform the following calculation:
SLG + (1.27 * OBP)
...to come out with an adjOPS figure of .844.
Now, if you do this for the hypothetical players listed above, then you come out with:
Player A: .957 adjOPS
Player B: .938 adjOPS
And you notice that a gap forms between the two players. Adjusting the statistic in this way doesn't account for the fact that a point of OBP is more valuable than a point of SLG, but you can get around this problem by increasing the coefficient by which you multiply OBP by a couple tenths of a point. This will further widen the gap between the two players, and it becomes evident that, all other things equal, Player A was the more productive player than Player B.
For the last few months, I've fiddled with the idea of creating a kind of OPS statistic for pitchers, that could be used to compare a group of pitchers by easy addition and simple multiplication. I never really gave it much more than a passing thought, but just by coincidence, it hit me this morning:
Look at their peripherals!
Indeed, rate stats such as BB/9, K/9, HR/9, and H/9 can be used in a similar way to OBP and SLG in calculating OPS. How is that, you ask? Well, let me tell you.
First things first. The ML league-average peripherals last year were: (AL) [NL]
BB/9: 3.3 (3.16) [3.42]
K/9: 6.4 (6.11) [6.65]
HR/9: 1.08 (1.11) [1.05]
H/9: 9.15 (9.30) [9.01]
Now, the idea that came to me was to simply add up the numbers and see what kind of number you get; this would represent a league-average pitcher (as the components of the statistic were based on the statistical means). However, there are two problems inherent in this equation:
1) Ideal level. A lower BB/9, HR/9, and H/9 is good for a pitcher, as opposed to K/9, which is better when high.
2) Balance. Simple addition would weigh H/9 as more important than any other statistic, while giving HR/9 the least influence.
I managed to tackle each of these issues while sitting in lecture this morning. BB/9, HR/9, and H/9 are all better off with lower values, whereas K/9 is better when it has a high value. How can you adjust this to make it uniform? I decided that, to solve the problem, you could find out the pitcher’s IP/K ratio; this way, a good pitcher will have four low values across the board, and a bad pitcher will have four high values, as opposed to the three low/one high or one low/three high complication of the original equation. To get IP/K, you just calculate 9/(K/9).
What about the second problem? Well, you solve it in the exactly the same way we adjusted OPS earlier: divide the highest peripheral by each of the other stats, and you come up with coefficients by which to multiply the statistics in order to balance them out. Here's a numerical explanation, which I hope will clear some things up:
The average ML pitcher posted a BB/9 of 3.3, a IP/K of 1.41, a HR/9 of 1.08 and a H/9 of 9.15. The H/9 stat - 9.15 - is the highest out of the four, so in order to balance things out, you divide 9.15 by 3.3 to find the coefficient for BB/9, divide 9.15 by 1.41 to find the coefficient for IP/K, and divide 9.15 by 1.08 to find the coefficient for HR/9.
After you do all this, you come up with the following equation:
(2.772727273*BB/9) + (6.506666667 * 9/K) + (8.472222222 * HR/9) + H/9 = League Average RATING
After plugging in the numbers for each variable in the equation (represented by the various peripheral stats), you come out with a figure of 36.6. Now, I don't know what to call my statistic yet, but whatever it is, the average major league pitcher had 36.6 points of it.
I went on to separate by league, and came up with the following equations:
AL average RATING = (2.943037975*BB/9) + (6.313666667* 9/K) + (8.378378378* HR/9) + H/9
NL average RATING = (2.634502924*BB/9) + (6.657388889* 9/K) + (8.580952381* HR/9) + H/9
AL average RATING: 37.2
NL average RATING: 36.04
So, now that that's out of the way, how about a little hypothetical comparison again? The following are two National League pitchers:
Pitcher A: 4.00 BB/9, 1.5 IP/K, 1 HR/9, 9 H/9
Pitcher B: 2.00 BB/9, 2.00 IP/K, 1.5 HR/9, 9 H/9
If you plug these values into the equation, then you come up with:
Pitcher A: 38.10
Pitcher B: 40.46
...and you may infer that Pitcher A, while 2.07 RATING points below league-average (remember, in this case a high RATING is bad, while a low RATING is good), was still more than two points better than Pitcher B.
And so, without further ado, the following is a table for your 2003 Seattle Mariners:
I adjusted the H/9 and HR/9 ratings for ballpark; there were actually five more home runs hit at Safeco than on the road (89 to 84), so I divided each player's HR/9 by 1.06. However, there were only 90% as many hits in Seattle than on the road, so each player's H/9 ratio was divided by 0.9.
What does the table show us?
Now that the lengthy explanation is out of the way, it's really not that bad of a statistic, and it provides a decent, relatively quick idea of a pitcher's performance when you don't have access to pERA or cERA (peripheral and component ERA, respectively). All that's left is naming the statistic, which I'll leave up to you guys. Give me some ideas, here.
...you didn't think I was serious, did you?
Quite the contrary! In fact, prepare yourselves for yet another column dealing with quantitative analysis. However, this time I think the points of discussion will be much easier to understand, and I'm cautiously optimistic that what I'm about to write will generate some kind of positive feedback.
I assume that everyone reading this is familiar with OPS and the perils of reading too much into it. As a quick-and-dirty metric, it's fairly serviceable, and it has made its way into various sports columns in newspapers around the country as journalists attempt to keep up with the ever-changing field of quantitative analysis while keeping things simple enough for their readership. However, the metric's simplicity is part of the problem. Check out this hypothetical example:
Player A: .260/.400/.450
Player B: .260/.330/.520
Both players will have identical OPS' (.850), even though Player A was more valuable than Player B. This is the main problem with OPS, but don't fret! If you're determined to use a similar statistic for player comparison, there's a relatively simple, nice way around the issue. The league-average line last year was .264/.333/.422, with an OPS of .755. But you can adjust this figure: the average SLG was 1.27 times the average OBP, so you can perform the following calculation:
SLG + (1.27 * OBP)
...to come out with an adjOPS figure of .844.
Now, if you do this for the hypothetical players listed above, then you come out with:
Player A: .957 adjOPS
Player B: .938 adjOPS
And you notice that a gap forms between the two players. Adjusting the statistic in this way doesn't account for the fact that a point of OBP is more valuable than a point of SLG, but you can get around this problem by increasing the coefficient by which you multiply OBP by a couple tenths of a point. This will further widen the gap between the two players, and it becomes evident that, all other things equal, Player A was the more productive player than Player B.
For the last few months, I've fiddled with the idea of creating a kind of OPS statistic for pitchers, that could be used to compare a group of pitchers by easy addition and simple multiplication. I never really gave it much more than a passing thought, but just by coincidence, it hit me this morning:
Look at their peripherals!
Indeed, rate stats such as BB/9, K/9, HR/9, and H/9 can be used in a similar way to OBP and SLG in calculating OPS. How is that, you ask? Well, let me tell you.
First things first. The ML league-average peripherals last year were: (AL) [NL]
BB/9: 3.3 (3.16) [3.42]
K/9: 6.4 (6.11) [6.65]
HR/9: 1.08 (1.11) [1.05]
H/9: 9.15 (9.30) [9.01]
Now, the idea that came to me was to simply add up the numbers and see what kind of number you get; this would represent a league-average pitcher (as the components of the statistic were based on the statistical means). However, there are two problems inherent in this equation:
1) Ideal level. A lower BB/9, HR/9, and H/9 is good for a pitcher, as opposed to K/9, which is better when high.
2) Balance. Simple addition would weigh H/9 as more important than any other statistic, while giving HR/9 the least influence.
I managed to tackle each of these issues while sitting in lecture this morning. BB/9, HR/9, and H/9 are all better off with lower values, whereas K/9 is better when it has a high value. How can you adjust this to make it uniform? I decided that, to solve the problem, you could find out the pitcher’s IP/K ratio; this way, a good pitcher will have four low values across the board, and a bad pitcher will have four high values, as opposed to the three low/one high or one low/three high complication of the original equation. To get IP/K, you just calculate 9/(K/9).
What about the second problem? Well, you solve it in the exactly the same way we adjusted OPS earlier: divide the highest peripheral by each of the other stats, and you come up with coefficients by which to multiply the statistics in order to balance them out. Here's a numerical explanation, which I hope will clear some things up:
The average ML pitcher posted a BB/9 of 3.3, a IP/K of 1.41, a HR/9 of 1.08 and a H/9 of 9.15. The H/9 stat - 9.15 - is the highest out of the four, so in order to balance things out, you divide 9.15 by 3.3 to find the coefficient for BB/9, divide 9.15 by 1.41 to find the coefficient for IP/K, and divide 9.15 by 1.08 to find the coefficient for HR/9.
After you do all this, you come up with the following equation:
(2.772727273*BB/9) + (6.506666667 * 9/K) + (8.472222222 * HR/9) + H/9 = League Average RATING
After plugging in the numbers for each variable in the equation (represented by the various peripheral stats), you come out with a figure of 36.6. Now, I don't know what to call my statistic yet, but whatever it is, the average major league pitcher had 36.6 points of it.
I went on to separate by league, and came up with the following equations:
AL average RATING = (2.943037975*BB/9) + (6.313666667* 9/K) + (8.378378378* HR/9) + H/9
NL average RATING = (2.634502924*BB/9) + (6.657388889* 9/K) + (8.580952381* HR/9) + H/9
AL average RATING: 37.2
NL average RATING: 36.04
So, now that that's out of the way, how about a little hypothetical comparison again? The following are two National League pitchers:
Pitcher A: 4.00 BB/9, 1.5 IP/K, 1 HR/9, 9 H/9
Pitcher B: 2.00 BB/9, 2.00 IP/K, 1.5 HR/9, 9 H/9
If you plug these values into the equation, then you come up with:
Pitcher A: 38.10
Pitcher B: 40.46
...and you may infer that Pitcher A, while 2.07 RATING points below league-average (remember, in this case a high RATING is bad, while a low RATING is good), was still more than two points better than Pitcher B.
And so, without further ado, the following is a table for your 2003 Seattle Mariners:
| Player | BB/9 | IP/K | HR/9 | H/9 | RATING |
| AL | 3.16 | 1.47 | 1.11 | 9.3 | 37.2 |
| Soriano | 2.04 | 0.78 | 0.32 | 5.65 | 19.25 |
| Nelson | 3.34 | 0.80 | 0.68 | 9.00 | 29.57 |
| Benitez | 6.91 | 0.96 | 0.59 | 6.96 | 38.29 |
| Rhodes | 3.00 | 1.13 | 0.63 | 9.79 | 32.72 |
| Sasaki | 4.05 | 1.15 | 0.51 | 9.28 | 32.72 |
| Looper | 2.57 | 1.17 | 1.21 | 9.97 | 35.07 |
| Mateo | 1.37 | 1.21 | 1.39 | 8.03 | 31.29 |
| Putz | 7.36 | 1.22 | 0.00 | 10.87 | 40.24 |
| Sweeney | 0.96 | 1.33 | 0.00 | 7.48 | 18.74 |
| Garcia | 3.17 | 1.40 | 1.31 | 9.71 | 38.83 |
| Pineiro | 3.23 | 1.40 | 0.76 | 9.05 | 33.79 |
| Taylor | 4.26 | 1.41 | 0.00 | 13.38 | 34.82 |
| Meche | 3.04 | 1.43 | 1.37 | 10.01 | 39.47 |
| Moyer | 2.76 | 1.67 | 0.75 | 9.23 | 34.17 |
| Franklin | 2.59 | 2.14 | 1.36 | 9.36 | 41.92 |
| Carrara | 4.34 | 2.23 | 1.76 | 13.76 | 55.36 |
| Hasegawa | 2.22 | 2.28 | 0.58 | 8.47 | 34.26 |
| White | 9.00 | 0.00 | 8.49 | 14.96 | 112.58 |
| TEAM | 2.91 | 1.44 | 1.02 | 9.27 | 35.47 |
I adjusted the H/9 and HR/9 ratings for ballpark; there were actually five more home runs hit at Safeco than on the road (89 to 84), so I divided each player's HR/9 by 1.06. However, there were only 90% as many hits in Seattle than on the road, so each player's H/9 ratio was divided by 0.9.
What does the table show us?
- The Mariners' staff, as a whole, were more than 1.7 points better than AL average
- Freddy Garcia was officially below-average last year
- Rafael Soriano is creepy-good
- Shigetoshi Hasegawa was much closer to the league mean than his ERA would indicate
- Ryan Franklin really wasn't that great
- Matt White sucks
Now that the lengthy explanation is out of the way, it's really not that bad of a statistic, and it provides a decent, relatively quick idea of a pitcher's performance when you don't have access to pERA or cERA (peripheral and component ERA, respectively). All that's left is naming the statistic, which I'll leave up to you guys. Give me some ideas, here.
Another Braves trade today:
Chris Reitsma for Jung Bong and Bubba Nelson.
I don't really understand this, from Atlanta's perspective.
Chris Reitsma for Jung Bong and Bubba Nelson.
I don't really understand this, from Atlanta's perspective.
Thursday, March 25, 2004
Tonight's Anger-Inducing Picture of the Day:
However, Lalime's 29-save shutout helps to ease the pain.
However, Lalime's 29-save shutout helps to ease the pain.
Two trades went down today:
Randy Choate to Arizona in exchange for John Patterson.
Juan Cruz and Steve Smyth to Atlanta for Andy Pratt and Richard Lewis.
The latter looks like a legitimate steal for the Braves, as Cruz should flourish away from Baker's poor influence.
Randy Choate to Arizona in exchange for John Patterson.
Juan Cruz and Steve Smyth to Atlanta for Andy Pratt and Richard Lewis.
The latter looks like a legitimate steal for the Braves, as Cruz should flourish away from Baker's poor influence.
I couldn't help it; I just had to bring back a Golden Oldie, courtesy of Chris at At Least the Red Sox Have 1918.
Here's a fun little toy: Alta Vista's Babel Fish Translation Service. Just because, I entered the title of this blog, "At Least The Red Sox Have 1918," which though a series of translations into Spanish, French, German, Chinese, Portugese, Japanese, Korean, and Italian, came back as:
To the little red o of sulfuration that it is become fullfilled in 1918
Here's another: "Sodo Mojo" comes back as:
Under inside desiring anxiously the origin of spremuta 1
Looks like we can close the book on that one!
Here's some of the titles of the other M's blogs run through the translations:
U.S.S. Mariner - "Great sheep of U.S.S."
Mariner Musings - "It disappears the inactive fancy of the group"
Sports And Bremertonians - "Sports And Bremertonians"
Mariners Wheelhouse - "House of the civil official of the great sheep"
Mariner Optimist - "Opportunist of the great sheep"
Cracking The Safe - "It apprehends the crack of the box"
Trident Fever - "Inside of ten 3 of the defect"
Fire Bavasi - "Suddenness Bavasi"
The Safest Blog On The Web - "The majority is pure in the Blog that is safe"
From Basketball To Baseball... - "In baseball that it cultivates the tool..."
Just Another Mariners Blog - "Almost different great ambition of Blog"
I could spend hours doing this. It's funny, the translations at least convey some of the original meaning until you run them through Korean; for some reason "Mariner," "Sailor," or "Marine" seems to end up as "Great Sheep" at that step. I'll leave that one alone for now.
Well, we've had some new members join the blogosphere in the last two months, so here's how the following names translate:
Mariner Minors: Worker of the mountain of the great ambition
Mariners Weekly: Each interest of fan width unit ship
Edgar is God: Edgar is the shoes
Bobby's Sports and News: Sports and message of official policies
Grand Salami: It it inner agencies of the victim of the taste of the garlic, than that one abundant available it knows it
Dead Reckoning: The movable law was conceited
Hope Springs Eternal: The effect that is eternal in place nine inside if it maintains the desire to it of the danger of the resistance work
Bavasi Stinks: I smell of Bavasi
Rumblings and Grumblings: The internal order of this place heavyheartedly thinks inside about of the Rumblings,
Mariner Bullpen: With respect to a great ambition of the cow it will sound,
SS Mariner: Great ambition of the body that melts
Olympia Mariner: Of the great ambition of Olympia
Here's a fun little toy: Alta Vista's Babel Fish Translation Service. Just because, I entered the title of this blog, "At Least The Red Sox Have 1918," which though a series of translations into Spanish, French, German, Chinese, Portugese, Japanese, Korean, and Italian, came back as:
To the little red o of sulfuration that it is become fullfilled in 1918
Here's another: "Sodo Mojo" comes back as:
Under inside desiring anxiously the origin of spremuta 1
Looks like we can close the book on that one!
Here's some of the titles of the other M's blogs run through the translations:
U.S.S. Mariner - "Great sheep of U.S.S."
Mariner Musings - "It disappears the inactive fancy of the group"
Sports And Bremertonians - "Sports And Bremertonians"
Mariners Wheelhouse - "House of the civil official of the great sheep"
Mariner Optimist - "Opportunist of the great sheep"
Cracking The Safe - "It apprehends the crack of the box"
Trident Fever - "Inside of ten 3 of the defect"
Fire Bavasi - "Suddenness Bavasi"
The Safest Blog On The Web - "The majority is pure in the Blog that is safe"
From Basketball To Baseball... - "In baseball that it cultivates the tool..."
Just Another Mariners Blog - "Almost different great ambition of Blog"
I could spend hours doing this. It's funny, the translations at least convey some of the original meaning until you run them through Korean; for some reason "Mariner," "Sailor," or "Marine" seems to end up as "Great Sheep" at that step. I'll leave that one alone for now.
Well, we've had some new members join the blogosphere in the last two months, so here's how the following names translate:
Mariner Minors: Worker of the mountain of the great ambition
Mariners Weekly: Each interest of fan width unit ship
Edgar is God: Edgar is the shoes
Bobby's Sports and News: Sports and message of official policies
Grand Salami: It it inner agencies of the victim of the taste of the garlic, than that one abundant available it knows it
Dead Reckoning: The movable law was conceited
Hope Springs Eternal: The effect that is eternal in place nine inside if it maintains the desire to it of the danger of the resistance work
Bavasi Stinks: I smell of Bavasi
Rumblings and Grumblings: The internal order of this place heavyheartedly thinks inside about of the Rumblings,
Mariner Bullpen: With respect to a great ambition of the cow it will sound,
SS Mariner: Great ambition of the body that melts
Olympia Mariner: Of the great ambition of Olympia
From today's Miracle News Section:
Pete Vuckovich, who scouts for the Brewers after spending much of his big league career with Milwaukee, has been a constant sight around the Seattle camp the past 10 days or so.
Reports are he might be scouting right-handed pitcher Kevin Jarvis for the Brewers. Milwaukee is short of starting pitching and Jarvis might be a fit there, particularly if the Brewers could get the Mariners to pick up part of the $4.5 million Jarvis is owed for the 2004 season.
Pete Vuckovich, who scouts for the Brewers after spending much of his big league career with Milwaukee, has been a constant sight around the Seattle camp the past 10 days or so.
Reports are he might be scouting right-handed pitcher Kevin Jarvis for the Brewers. Milwaukee is short of starting pitching and Jarvis might be a fit there, particularly if the Brewers could get the Mariners to pick up part of the $4.5 million Jarvis is owed for the 2004 season.
In the spirit of rate stats, which have been the topic of late, I decided to slap together a massive spreadsheet detailing the frequency of certain events among teams in the majors, broken down by league and then division. I have simplified the spreadsheet - which included numbers for every team in the majors - into a 5 * 10 table that shows a few statistics in an easy-to-read fashion. Observe the chart below, and note that when factoring in Plate Appearances and Walks, intentional walks were excluded, as the odds of a hit/home run/strikeout are 0 (because it's a guaranteed base on balls).
What does this table tell us?
All right. Didn't we pretty much know all this *before* we saw the chart? I mean, 8 or 9% of all NL at bats go to pitchers, who are notoriously lousy hitters. Therefore, the National League should post lower home run and hit rates, while striking out more often. According to the table, this is indeed the case. Something we didn't know before the table, however, is that the National League generally draws more walks than the AL.
So where is this leading, then? Well, most of us are familiar with "independent" stats; by this I mean statistics that are independent of exterior influence. For example, K/9 is a "defense-independent" statistic, because the defense plays no part in determining how often a pitcher records strikeouts. These kinds of numbers can become more and more complex. Something you'll stumble across pretty often is the "ballpark-independent" variety of stats, such as EqA, which adjust for how favorable a stadium was to hitters/pitchers in order to produce a final figure displaying how a player's performance would have been in a neutral ballpark.
Well, something that I think could be worth exploring is the notion of "division-independent" or "league-independent" statistics. Johan Santana posted a BB/9 ratio of 2.61 last year, whereas Tom Glavine's was 2.90. How much of this difference came because Santana pitched in the division that walked the least, while Glavine spent a bunch of innings working against teams who walked more than anyone else?
There are a lot of things that need perfecting before this kind of thing can be considered a legitimate manipulation of the stats, but it's something to think about. 83.1 of Santana's innings pitched (out of 158.1 total IP) came against AL Central teams, who (excluding the Twins) walked once every 12.67 plate appearances. If we were to replace those innings against the ALC with innings against an ML-average team (drawing a walk once every 11.53 PA's), then Santana's BB/9 rises to 2.79. For Glavine, 88.2 of his innings came against NLE foes, who (excluding the Mets) drew a walk once every 10.9 plate appearances. If you go through the same process - replacing those innings with innings against league-average teams - then Glavine's BB/9 ratio drops from 2.90 to 2.80. A difference of 29 ratio points has essentially been erased by adjusting for the divisions in which each pitcher played.
Let's give it another spin, shall we? 86.2 of Victor Zambrano's innings came against AL East opponents, who (excluding Tampa Bay) homered once every 30.95 plate appearances. Zambrano's normal HR/9 ratio was 1.00. If you adjust for the division by replacing that 1-every-30.95 with 1-every-35.17, his HR/9 dips to 0.93. This is a small difference, to be sure, but it's a difference nonetheless, and could serve to partly explain such things as, say, a lower strikeout rate for a pitcher when he changes leagues.
What's that, you say? You want to see what difference it makes for Ryan Franklin? 83 of his 212 innings came against AL West opponents, who struck out at a rate of once every 6.46 PA's, the least frequently of any division in the majors. Adjusting for division raises Franklin's K/9 by two-tenths of a point to 4.40. Still sucks, but it's incrementally better (I guess).
Keep in mind that these examples have only been adjusting for divisional play. Why stop there - why not adjust for the league? You could conceivably figure out how a pitcher would have performed had he played in the other league, or something along those lines. I don't want to run through any more calculations tonight, but if you're so inclined, just contact me and I'll send you a spreadsheet with all the necessary data.
Now, before we go too far with these kinds of admittedly minor manipulations, we must first ask a question: are the divisional and league-wide rate stats a function of the hitters or the pitchers (or some combination of both)? After all, if the AL Central has the lowest walk rate because the hitters face a disproportionate amount of AL Central pitchers, who may have just happened to not allow many walks, then none of this does us any good, does it? Well, as it turns out, AL Central pitchers have a virtually identical BB/9 ratio as NL East pitchers (who, if you recall, pitched in the division which drew walks most often). Thus, we may infer that each division faces a significant enough sample of various pitchers so that the collection of said pitchers faced may be considered league-average. It's the same way we consider only a Rockies player's road stats when looking at his line; the group of ballparks in which the player played away from Colorado, although slightly disproportionate, is large enough so that it essentially represents the player's performance in a neutral ballpark.
The funny thing is, the adjustments made to any given player's statistics using the aforementioned method are quite small. You aren't going to see a guy's K/9 ratio jump a full point when moving him from, say, the NL Central to the NL West. It's more likely that you'll see a change of a couple tenths of a point, or even more likely, a few hundredths. Even so, the art of quantitative analysis is in constant pursuit of accuracy and precision, and if you are able to make a proper adjustment of a few hundredths of a point in order to better understand a player's true performance, why not do it?
| Division/League | PA/BB | PA/K | PA/HR | PA/H |
| ALE | 11.53 | 6.23 | 32.90 | 4.00 |
| ALC | 12.55 | 6.03 | 35.76 | 4.18 |
| ALW | 11.59 | 6.46 | 34.66 | 4.15 |
| NLE | 11.10 | 5.94 | 36.58 | 4.18 |
| NLC | 11.40 | 5.47 | 33.47 | 4.21 |
| NLW | 11.14 | 5.84 | 36.68 | 4.23 |
| AL | 11.89 | 6.22 | 34.37 | 4.10 |
| NL | 11.23 | 5.72 | 35.92 | 4.21 |
| ML | 11.53 | 5.95 | 35.18 | 4.16 |
What does this table tell us?
- Teams in the AL Central draw significantly fewer walks than any other division
- National League teams strike out more frequently than AL teams, by a fair margin
- Teams in the NL Central strike out quite often, whereas the AL West is much better at making contact
- The American League was more powerful than the NL, with the AL East leading the way
- The AL East was far and away the best division at recording hits
All right. Didn't we pretty much know all this *before* we saw the chart? I mean, 8 or 9% of all NL at bats go to pitchers, who are notoriously lousy hitters. Therefore, the National League should post lower home run and hit rates, while striking out more often. According to the table, this is indeed the case. Something we didn't know before the table, however, is that the National League generally draws more walks than the AL.
So where is this leading, then? Well, most of us are familiar with "independent" stats; by this I mean statistics that are independent of exterior influence. For example, K/9 is a "defense-independent" statistic, because the defense plays no part in determining how often a pitcher records strikeouts. These kinds of numbers can become more and more complex. Something you'll stumble across pretty often is the "ballpark-independent" variety of stats, such as EqA, which adjust for how favorable a stadium was to hitters/pitchers in order to produce a final figure displaying how a player's performance would have been in a neutral ballpark.
Well, something that I think could be worth exploring is the notion of "division-independent" or "league-independent" statistics. Johan Santana posted a BB/9 ratio of 2.61 last year, whereas Tom Glavine's was 2.90. How much of this difference came because Santana pitched in the division that walked the least, while Glavine spent a bunch of innings working against teams who walked more than anyone else?
There are a lot of things that need perfecting before this kind of thing can be considered a legitimate manipulation of the stats, but it's something to think about. 83.1 of Santana's innings pitched (out of 158.1 total IP) came against AL Central teams, who (excluding the Twins) walked once every 12.67 plate appearances. If we were to replace those innings against the ALC with innings against an ML-average team (drawing a walk once every 11.53 PA's), then Santana's BB/9 rises to 2.79. For Glavine, 88.2 of his innings came against NLE foes, who (excluding the Mets) drew a walk once every 10.9 plate appearances. If you go through the same process - replacing those innings with innings against league-average teams - then Glavine's BB/9 ratio drops from 2.90 to 2.80. A difference of 29 ratio points has essentially been erased by adjusting for the divisions in which each pitcher played.
Let's give it another spin, shall we? 86.2 of Victor Zambrano's innings came against AL East opponents, who (excluding Tampa Bay) homered once every 30.95 plate appearances. Zambrano's normal HR/9 ratio was 1.00. If you adjust for the division by replacing that 1-every-30.95 with 1-every-35.17, his HR/9 dips to 0.93. This is a small difference, to be sure, but it's a difference nonetheless, and could serve to partly explain such things as, say, a lower strikeout rate for a pitcher when he changes leagues.
What's that, you say? You want to see what difference it makes for Ryan Franklin? 83 of his 212 innings came against AL West opponents, who struck out at a rate of once every 6.46 PA's, the least frequently of any division in the majors. Adjusting for division raises Franklin's K/9 by two-tenths of a point to 4.40. Still sucks, but it's incrementally better (I guess).
Keep in mind that these examples have only been adjusting for divisional play. Why stop there - why not adjust for the league? You could conceivably figure out how a pitcher would have performed had he played in the other league, or something along those lines. I don't want to run through any more calculations tonight, but if you're so inclined, just contact me and I'll send you a spreadsheet with all the necessary data.
Now, before we go too far with these kinds of admittedly minor manipulations, we must first ask a question: are the divisional and league-wide rate stats a function of the hitters or the pitchers (or some combination of both)? After all, if the AL Central has the lowest walk rate because the hitters face a disproportionate amount of AL Central pitchers, who may have just happened to not allow many walks, then none of this does us any good, does it? Well, as it turns out, AL Central pitchers have a virtually identical BB/9 ratio as NL East pitchers (who, if you recall, pitched in the division which drew walks most often). Thus, we may infer that each division faces a significant enough sample of various pitchers so that the collection of said pitchers faced may be considered league-average. It's the same way we consider only a Rockies player's road stats when looking at his line; the group of ballparks in which the player played away from Colorado, although slightly disproportionate, is large enough so that it essentially represents the player's performance in a neutral ballpark.
The funny thing is, the adjustments made to any given player's statistics using the aforementioned method are quite small. You aren't going to see a guy's K/9 ratio jump a full point when moving him from, say, the NL Central to the NL West. It's more likely that you'll see a change of a couple tenths of a point, or even more likely, a few hundredths. Even so, the art of quantitative analysis is in constant pursuit of accuracy and precision, and if you are able to make a proper adjustment of a few hundredths of a point in order to better understand a player's true performance, why not do it?
Wednesday, March 24, 2004
Looks like Dan Evans is going to work for the Mariners.
Lose one Gillick, bring in his lesser-experienced equivalent, I guess.
Lose one Gillick, bring in his lesser-experienced equivalent, I guess.
Reader Gareth Owen brought up an interesting point in an email. Here's what he had to say:
*****This is backwards from how I'd expect: batting line-ups being organised so your big home run threats are up disproportionately often with men on base.
One thing to consider is that with men on base, there are other things that batters are trying to do -- bunt runners over, sacrifice a guy in from third.
How do the stats look if you take the sacrifices (and failed sacrifices) out of the men-on-base numbers. i.e. look only at the numbers where it can be assumed the batter is allowed to swing for the fences. Could it be enough to remove / reverse the anomaly?*****
It's a valid question, and just one of the reasons why I love feedback. Now, I don't know where to find out the number of times hitters failed while trying to bunt last year, nor do I have access to the number of times batters, say, tried to put the ball on the ground behind the runner, so as to advance him to the next base (such a statistic doesn't exist). However, what I do have is the total number of sacrifice hits last year: 1626.
Now, clearly each of those sacrifices came with runners on base. After subtracting this figure from the total number of at bats last season, we come up with the following:
57.3% of all at bats *in which a hitter had the potential to hit a home run* came with nobody on base. This is 0.6% higher than the 56.7 figure from yesterday.
Now, 57.3% is closer to the 58.5% probability of a home run being a solo shot, but it's still more than a full point apart. Thus, what we find is that solo home runs are still hit more often than you'd expect, given the relative situational rates, but the gap is one-third smaller than we thought yesterday. If you account for those other at bats where the hitter wasn't looking to hit a home run (failed bunts, hit-and-runs), the percentage will rise another few tenths of a point, narrowing the gap further.
...of course, there are also those at bats with nobody on in which a hitter tried to reach by bunting (here's looking at you, Castillo and Ichiro), which would *widen* the gap a little bit. So the gap is still there, and it's still a difference of at least a full percentage point (I don't count sacrifice flies, because I highly doubt that hitters approach the plate thinking "all right, I'm going to hit this 350 feet to dead center and advance that guy on third." Sac flies just *happen*).
Later today, I intend to study some more rate stats (not necessarily related to home runs, though!), so be sure and stop by. I have no idea what the results of my research are going to be, but hey, that's part of the fun, right?
*****This is backwards from how I'd expect: batting line-ups being organised so your big home run threats are up disproportionately often with men on base.
One thing to consider is that with men on base, there are other things that batters are trying to do -- bunt runners over, sacrifice a guy in from third.
How do the stats look if you take the sacrifices (and failed sacrifices) out of the men-on-base numbers. i.e. look only at the numbers where it can be assumed the batter is allowed to swing for the fences. Could it be enough to remove / reverse the anomaly?*****
It's a valid question, and just one of the reasons why I love feedback. Now, I don't know where to find out the number of times hitters failed while trying to bunt last year, nor do I have access to the number of times batters, say, tried to put the ball on the ground behind the runner, so as to advance him to the next base (such a statistic doesn't exist). However, what I do have is the total number of sacrifice hits last year: 1626.
Now, clearly each of those sacrifices came with runners on base. After subtracting this figure from the total number of at bats last season, we come up with the following:
57.3% of all at bats *in which a hitter had the potential to hit a home run* came with nobody on base. This is 0.6% higher than the 56.7 figure from yesterday.
Now, 57.3% is closer to the 58.5% probability of a home run being a solo shot, but it's still more than a full point apart. Thus, what we find is that solo home runs are still hit more often than you'd expect, given the relative situational rates, but the gap is one-third smaller than we thought yesterday. If you account for those other at bats where the hitter wasn't looking to hit a home run (failed bunts, hit-and-runs), the percentage will rise another few tenths of a point, narrowing the gap further.
...of course, there are also those at bats with nobody on in which a hitter tried to reach by bunting (here's looking at you, Castillo and Ichiro), which would *widen* the gap a little bit. So the gap is still there, and it's still a difference of at least a full percentage point (I don't count sacrifice flies, because I highly doubt that hitters approach the plate thinking "all right, I'm going to hit this 350 feet to dead center and advance that guy on third." Sac flies just *happen*).
Later today, I intend to study some more rate stats (not necessarily related to home runs, though!), so be sure and stop by. I have no idea what the results of my research are going to be, but hey, that's part of the fun, right?
Tuesday, March 23, 2004
So I've been doing some more research about this morning's topic. I put together a spreadsheet with every team's ratio of solo-to-total home runs, as well as the ratio of at bats with none on-to-total at bats against. I decided that it's way too large to put into a table (that would also be a very tedious process for yours truly), so if you're interested, just contact me and I'll send you the Excel file.
Anyway, here's what I found:
Just for kicks (and for the visual learners among you), here's a graph, with each data point representing a team:
It looks like a somewhat random arrangement of points, but the trend line shows an evident correllation between the frequency with which a team pitches with nobody on base and the frequency with which said team allows solo home runs. This serves to support the idea that the two conditions are directly related to each other; adding data from other years would likely confirm this hypothesis.
One of the more interesting things to note is that 58.5% of all home runs are solo shots, even though 56.7% of all at bats come with nobody on base. Now, whether or not this represents a significant difference is up to the reader - it may or may not show up in 2002 and 2001 stats, but I'll leave that research to somebody else. The difference is essentially this:
With nobody on, a pitcher allows a home run once every 31 at bats.
With runners on base, a pitcher allows a home run once every 33.4 at bats.
Under the assumption that there *is* a legitimate difference between those two numbers, the separation could be caused by any number of things. Maybe pitchers are more aggressive when the worst possible outcome of a certain at bat is one run; it's possible that your average pitcher is more likely to go right after a hitter with nobody on base, which would result in a few more balls flying over the fence. Maybe all the talk about how difficult it is to pitch out of the stretch is incorrect, that Joe Average is actually more successful when he's not going out of the windup. Maybe when there are runners on, batters are just looking to make contact with the ball, whereas they try for the homer more often when there's nobody on base to strand. There are potentially limitless reasons for this statistical discrepancy, and it would make for an interesting study.
Something else to note is that 22 of the 30 teams allowed solo home runs more often than you'd expect, given the proportion of their at bats against that came with nobody on; only eight pitching staffs had a (% Solo HR)/(% AB None On) ratio below 1.00. If this is indeed a true statistical trend, and not just a coincidence/anomaly, why is it so? Is it part of an organizational philosophy to attack the batter more when there's no danger of a multi-run homer? If this is the case, it could tie in with the brief discussion in the last paragraph. What about those teams who *don't* allow as many solo homers as you'd expect? Do they have different pitching strategies? There's a whole world to explore here, and I don't know where to begin looking for answers.
The idea that solo home runs are directly proportional to the number of times a hitter or team goes to the plate with nobody on base is a logical conclusion, and really pretty simple; if Situation A, which lends itself to Outcome Z, occurs 60% of the time, and Situation B, which lends itself to Outcome Y, occurs 40% of the time, one may rationally conclude that Outcome Z will show up about 60% of the time. But why is it that teams allow slightly more solo home runs than you'd expect, given their situational splits? This is a relatively unexplored field, and one can only hope that Keith Woolner sheds some light on it.
Anyway, here's what I found:
- Of the 5207 home runs hit last year, 3048 were solo shots, for a ratio of 58.5%
- 56.7% of all at bats came with nobody on base
- Only one team had a difference of more than 0.074 percentage points between % solo home runs allowed and % at bats against with nobody on
- 22 out of all 30 teams had a higher (solo HR)/(total HR) ratio than (AB with none on)/(total AB) ratio
- Only three teams had a higher percentage of solo home runs allowed than the Mariners
- The Cubs had the highest ratio of (% Solo)/(% AB None On), while the Braves had the lowest
- In general, teams in pitcher's parks allowed a higher percentage of solo homers than teams in hitter's parks
Just for kicks (and for the visual learners among you), here's a graph, with each data point representing a team:
It looks like a somewhat random arrangement of points, but the trend line shows an evident correllation between the frequency with which a team pitches with nobody on base and the frequency with which said team allows solo home runs. This serves to support the idea that the two conditions are directly related to each other; adding data from other years would likely confirm this hypothesis.
One of the more interesting things to note is that 58.5% of all home runs are solo shots, even though 56.7% of all at bats come with nobody on base. Now, whether or not this represents a significant difference is up to the reader - it may or may not show up in 2002 and 2001 stats, but I'll leave that research to somebody else. The difference is essentially this:
With nobody on, a pitcher allows a home run once every 31 at bats.
With runners on base, a pitcher allows a home run once every 33.4 at bats.
Under the assumption that there *is* a legitimate difference between those two numbers, the separation could be caused by any number of things. Maybe pitchers are more aggressive when the worst possible outcome of a certain at bat is one run; it's possible that your average pitcher is more likely to go right after a hitter with nobody on base, which would result in a few more balls flying over the fence. Maybe all the talk about how difficult it is to pitch out of the stretch is incorrect, that Joe Average is actually more successful when he's not going out of the windup. Maybe when there are runners on, batters are just looking to make contact with the ball, whereas they try for the homer more often when there's nobody on base to strand. There are potentially limitless reasons for this statistical discrepancy, and it would make for an interesting study.
Something else to note is that 22 of the 30 teams allowed solo home runs more often than you'd expect, given the proportion of their at bats against that came with nobody on; only eight pitching staffs had a (% Solo HR)/(% AB None On) ratio below 1.00. If this is indeed a true statistical trend, and not just a coincidence/anomaly, why is it so? Is it part of an organizational philosophy to attack the batter more when there's no danger of a multi-run homer? If this is the case, it could tie in with the brief discussion in the last paragraph. What about those teams who *don't* allow as many solo homers as you'd expect? Do they have different pitching strategies? There's a whole world to explore here, and I don't know where to begin looking for answers.
The idea that solo home runs are directly proportional to the number of times a hitter or team goes to the plate with nobody on base is a logical conclusion, and really pretty simple; if Situation A, which lends itself to Outcome Z, occurs 60% of the time, and Situation B, which lends itself to Outcome Y, occurs 40% of the time, one may rationally conclude that Outcome Z will show up about 60% of the time. But why is it that teams allow slightly more solo home runs than you'd expect, given their situational splits? This is a relatively unexplored field, and one can only hope that Keith Woolner sheds some light on it.
Well, Mariano Rivera's sticking around. $21m/2yr extension, with a $10.5m option. Once again, the point must be made that the Yankees aren't playing with normal money, but that's still quite a bounty for a 34 year old relief pitcher.
On a tangentially related note, did ESPN.com do something to piss Yahoo! off? It used to be that you could just throw a player name into the query bar and the first link you'd get would be for the ESPN player profile. Now said results are hidden in the third and fourth pages of links, which is irritating.
On a tangentially related note, did ESPN.com do something to piss Yahoo! off? It used to be that you could just throw a player name into the query bar and the first link you'd get would be for the ESPN player profile. Now said results are hidden in the third and fourth pages of links, which is irritating.
Something I noticed in the table is that both Garcia and Meche managed to limit the amount of multi-run homers allowed. Each allowed a greater ratio of solo home runs than their percentage of time spent pitching with men on base would suggest, which could be inferred as showing that both pitchers were more aggressive with nobody on (resulting in more balls flying over the fence), whereas they were more careful with ducks on the pond.
Actually, it could be inferred as showing lots of things. That's just one of them.
Actually, it could be inferred as showing lots of things. That's just one of them.
Something interesting came up over at USS Mariner this morning. 61.8% of Ryan Franklin's home runs allowed were solo shots, and this seemed far too high to David Cameron, who assumed that a normal ratio would be about 1/4 (as a solo homer is one of four possible types of home runs). I've done me a little research to find out whether or not Franklin's percentage is that abnormal. The following is a table that shows the ratio of (solo homers)/(total homers) for each pitcher, followed by the ratio of (at bats with no runners on base)/(total at bats). It's my hypothesis that the ratio of solo/total home runs allowed by any given pitcher or team will directly reflect the percentage of time an allowed home run would be a solo shot. For example, I believe that a guy who pitches with nobody on base 50% of the time will allow solo home runs 50% of the time as well. So here's the table:
As you can see, Franklin's 61.8% ratio is actually 1.2 points below the team average; 63% of all home runs allowed by Mariners last year were solo shots.
What's more, you notice that there is a discrepancy between the ratios of the starting rotation and the bullpen. This is because relievers pitch more often with runners on base (44.1% of the time, as opposed to 39.7% for starters). As expected, relief pitchers allowed a greater percentage of non-solo home runs.
Perhaps later I'll expect the ratio of solo/non-solo home runs for other teams, but this alone provides more than adequate proof that Franklin's percentage really didn't stray from the mean.
| Player | Solo HR | Non-solo HR | % Solo | None On AB | Total AB | % None On AB |
| Franklin | 21 | 13 | 61.8 | 480 | 794 | 60.5 |
| Garcia | 22 | 9 | 71.0 | 467 | 770 | 60.6 |
| Meche | 21 | 9 | 70.0 | 438 | 711 | 61.6 |
| Moyer | 11 | 8 | 57.9 | 475 | 810 | 58.6 |
| Pineiro | 12 | 7 | 63.2 | 479 | 796 | 60.2 |
| Mateo | 10 | 4 | 71.4 | 194 | 314 | 61.8 |
| Carrara | 2 | 4 | 33.3 | 58 | 120 | 48.3 |
| Hasegawa | 4 | 1 | 80.0 | 145 | 264 | 54.9 |
| Rhodes | 0 | 4 | 0.0 | 100 | 207 | 48.3 |
| Nelson | 2 | 1 | 66.7 | 73 | 137 | 53.3 |
| Sasaki | 1 | 1 | 50.0 | 66 | 130 | 50.8 |
| Soriano | 2 | 0 | 100 | 124 | 185 | 67.0 |
| White | 1 | 1 | 50.0 | 4 | 8 | 50.0 |
| Benitez | 0 | 1 | 0.0 | 30 | 53 | 56.6 |
| Looper | 0 | 1 | 0.0 | 13 | 26 | 50.0 |
| Rotation | 86 | 46 | 65.2 | 2339 | 3881 | 60.3 |
| Bullpen | 22 | 18 | 55.0 | 807 | 1444 | 55.9 |
| Team Totals | 109 | 64 | 63.0 | 3146 | 5325 | 59.1 |
As you can see, Franklin's 61.8% ratio is actually 1.2 points below the team average; 63% of all home runs allowed by Mariners last year were solo shots.
What's more, you notice that there is a discrepancy between the ratios of the starting rotation and the bullpen. This is because relievers pitch more often with runners on base (44.1% of the time, as opposed to 39.7% for starters). As expected, relief pitchers allowed a greater percentage of non-solo home runs.
Perhaps later I'll expect the ratio of solo/non-solo home runs for other teams, but this alone provides more than adequate proof that Franklin's percentage really didn't stray from the mean.
Monday, March 22, 2004
The Angels are apparently thinking about paying Garrett Anderson 48 million dollars over the next four years.
For those of you who are curious, GA's 40.6 2003 VORP made him worth about $8.3m last year. Given that he's likely to decine over the next four seasons, I'd think that anything north of $30m/4yr would be a gross overpayment.
Update: The Angels are refusing to go higher than $40m/4yr, while it's speculated that GA is seeking something in the neighborhood of $56m over the same amount of time.
For those of you who are curious, GA's 40.6 2003 VORP made him worth about $8.3m last year. Given that he's likely to decine over the next four seasons, I'd think that anything north of $30m/4yr would be a gross overpayment.
Update: The Angels are refusing to go higher than $40m/4yr, while it's speculated that GA is seeking something in the neighborhood of $56m over the same amount of time.
A few fun things from this Rosenthal article:
The four years and $42 million remaining on Jason Kendall's contract isn't the only deterrent to the Mariners' pursuit of the Pirates catcher. The Mariners, a team that emphasizes defense, might not be comfortable with Kendall taking playing time from Dan Wilson, a superior defender. Scouts, however, are divided on how much Kendall has slipped defensively; some say he is still adequate. Kendall also could play outfield and serve as a DH. . . .
The Kendall topic has been discussed ad nauseum; just how many teams are going to offer Wiki Gonzalez and Kevin Jarvis to the Pirates in exchange for the catcher? So far, two...
Given the way our team has been put together, the only place Kendall could play is catcher (he'd make a lousy DH, anyway). He's been a good defensive catcher once on his career - 1999 - when he only played 75 games at the position. Since 2001, an average season has been sandwiched between two subpar years in which Kendall was a combined 13 runs below league average defensively (give or take a few, given the nature of defensive metrics).
Dan Wilson was a good defensive catcher for a while. Terrific, in fact, until 1998, when his defense fell off the map. From 1994-97, Wilson was 56 runs above league average, but since that year (arguably the best overall season of his career) he's hovered around the line of inadequacy. As a 35 year old catcher, we can expect that Wilson's defense will continue to slip, to the point at which the difference between his and Kendall's glovework is nothing more than two or three runs.
Yet, we know there's more to being a catcher than receiving and throwing the baseball. There's also pitch-calling, something of which Bob Melvin is most critical. So, what of Kendall's ability to call a good game? Well, take it for what it's worth, but his CERA - the ERA of the pitching staff while Kendall was behind the plate - was 4.46, as opposed to 6.04 for Craig Wilson. Call it a problem of sample size (Wilson only caught 15 games) or just a poor statistic, but it might be the best indication that we have of a player's ability to call a good game.
Hey, look at that, Wilson's CERA was only 0.08 points better than Davis'...
Kendall also caught base-stealers 26.7% of the time, whereas Wilson was successful on 30% of his attempts. Dan Wilson might be the better defensive catcher of the two, but the gap isn't very large, and it really shouldn't be considered an obstacle in pursuing Kendall.
...his contract is an obstacle, though, so let's just stay away from that can o' worms.
Here's another part of the article that caught my eye:
The Mariners' Class AAA rotation will be stocked with prospects, and one scout says it might be better than the Rangers' major league starting staff. The scout is not alone in his opinion: A Cactus League manager says Mariners Class AAA RHP Clint Nageotte is more advanced than Rangers RHP Colby Lewis. . . .
Here's how the rotations will probably stack up:
Tacoma
Nageotte
Blackley
Johnson
Baek
Madritsch (Heaverlo?)
Texas
Rogers
Lewis
Dickey
Park
Rusch/Rodriguez/Callaway
Tacoma will have a stable of decent-to-excellent prospects in the rotation, but are they really better than the Rangers' group of starters? Fortunately for us, PECOTA has a little thing called EqERA that will come in handy. We'll use the weighted mean projection instead of the 50% line.
Nageotte: 5.17
Blackley: N/A (probably around 5)
Johnson: 4.80
Baek: 4.84
Madritsch: 5.88
Heaverlo: 5.10
Rogers: N/A (let's use something around 4.80)
Lewis: 4.97
Dickey: 4.63
Park: 5.02
Rusch: 4.94
Rodriguez: 5.05
Callaway: N/A (probably in the neighborhood of 5.25)
If you feel like disobeying all kinds of mathematical principles and decide to average out the ERA's, you come up with a 4.95 figure for Texas, as opposed to 5.13 for Tacoma.
Pretty close, isn't it?
The four years and $42 million remaining on Jason Kendall's contract isn't the only deterrent to the Mariners' pursuit of the Pirates catcher. The Mariners, a team that emphasizes defense, might not be comfortable with Kendall taking playing time from Dan Wilson, a superior defender. Scouts, however, are divided on how much Kendall has slipped defensively; some say he is still adequate. Kendall also could play outfield and serve as a DH. . . .
The Kendall topic has been discussed ad nauseum; just how many teams are going to offer Wiki Gonzalez and Kevin Jarvis to the Pirates in exchange for the catcher? So far, two...
Given the way our team has been put together, the only place Kendall could play is catcher (he'd make a lousy DH, anyway). He's been a good defensive catcher once on his career - 1999 - when he only played 75 games at the position. Since 2001, an average season has been sandwiched between two subpar years in which Kendall was a combined 13 runs below league average defensively (give or take a few, given the nature of defensive metrics).
Dan Wilson was a good defensive catcher for a while. Terrific, in fact, until 1998, when his defense fell off the map. From 1994-97, Wilson was 56 runs above league average, but since that year (arguably the best overall season of his career) he's hovered around the line of inadequacy. As a 35 year old catcher, we can expect that Wilson's defense will continue to slip, to the point at which the difference between his and Kendall's glovework is nothing more than two or three runs.
Yet, we know there's more to being a catcher than receiving and throwing the baseball. There's also pitch-calling, something of which Bob Melvin is most critical. So, what of Kendall's ability to call a good game? Well, take it for what it's worth, but his CERA - the ERA of the pitching staff while Kendall was behind the plate - was 4.46, as opposed to 6.04 for Craig Wilson. Call it a problem of sample size (Wilson only caught 15 games) or just a poor statistic, but it might be the best indication that we have of a player's ability to call a good game.
Hey, look at that, Wilson's CERA was only 0.08 points better than Davis'...
Kendall also caught base-stealers 26.7% of the time, whereas Wilson was successful on 30% of his attempts. Dan Wilson might be the better defensive catcher of the two, but the gap isn't very large, and it really shouldn't be considered an obstacle in pursuing Kendall.
...his contract is an obstacle, though, so let's just stay away from that can o' worms.
Here's another part of the article that caught my eye:
The Mariners' Class AAA rotation will be stocked with prospects, and one scout says it might be better than the Rangers' major league starting staff. The scout is not alone in his opinion: A Cactus League manager says Mariners Class AAA RHP Clint Nageotte is more advanced than Rangers RHP Colby Lewis. . . .
Here's how the rotations will probably stack up:
Tacoma
Nageotte
Blackley
Johnson
Baek
Madritsch (Heaverlo?)
Texas
Rogers
Lewis
Dickey
Park
Rusch/Rodriguez/Callaway
Tacoma will have a stable of decent-to-excellent prospects in the rotation, but are they really better than the Rangers' group of starters? Fortunately for us, PECOTA has a little thing called EqERA that will come in handy. We'll use the weighted mean projection instead of the 50% line.
Nageotte: 5.17
Blackley: N/A (probably around 5)
Johnson: 4.80
Baek: 4.84
Madritsch: 5.88
Heaverlo: 5.10
Rogers: N/A (let's use something around 4.80)
Lewis: 4.97
Dickey: 4.63
Park: 5.02
Rusch: 4.94
Rodriguez: 5.05
Callaway: N/A (probably in the neighborhood of 5.25)
If you feel like disobeying all kinds of mathematical principles and decide to average out the ERA's, you come up with a 4.95 figure for Texas, as opposed to 5.13 for Tacoma.
Pretty close, isn't it?
Sunday, March 21, 2004
Subscribe to:
Posts (Atom)