Note: I started writing this in June, I think, revisited it in August, then did a final edit in November. At the time of the final edit, I did not think my overall point needed to change, so I stuck with the pre- SL 2001 numbers. That may have been unfair to Chipper in the grand scheme, because he did not end up with numbers as strong as what I projected. But in the context of what a 7th round pick should produce, I think it was reasonable to keep the comparison to Tatis that I based a lot of my assumptions on.
On final edit, and probably each time I looked at this previous to that, it is very hard to read. Lots of numbers, lots of assumptions I want the reader to buy into, some made up terminology, lots of vague references to the master game -- something I haven't played with great frequency in my lifetime and a game that I have not looked at much since about 1994. But because I think the process is as valid as anything else I can think of, I stand by my conclusions.
And finally, as it's November and Steve has announced his intention to leave our league, understand that I wouldn't have spoken about playing him as being painful. I respect him (especially what he's going through given the changes in his personal life and given what Taz presented him) too much to mindlessly mock him, but I kept that line in there to give context to what some of us felt early in the season about his stance on playability.
Competing with Steve Swinea is one of the true joys of the Summer League . . . and occasionally one of the biggest pains. Steve is always looking for an edge, which is exactly what makes the SL a success year after year. How he deploys those edges, therein lies the pain.
In 2001, he found three edges of consequence. The first, and the one I'll go into the most detail on, was his using Chipper Jones at SS. The others were his use of an early round pick on Mac and his decision to switch Bonds and Sheffield in the outfield when platoon rated PR's were at the plate.
In the 9 previous SL's, to my knowledge no one had used a 6 at SS with any great frequency. In 2000, Graham played Jay Bell (7) more than half the time, in '98 Dan played Jeff Blauser (7) regularly while Chris played Billy Spiers (7) occasionally, in '97 Chipper (7) saw occasional time for Joe and Dunston (7) the same for Steve, in '95 Will Cordero played regularly for Jim (95A) and Chris (95B), and in '94 Tony Phillips (I think a 7) play occasionally for Greg.
Based on what I know about the Master game, the supposed engine that drives BBW, 6's will a great majority of the time be a "fielding 3" in situations where the SS's defense has an effect on the outcome of the play. That doesn't always mean the 6 makes an error -- sometimes it means a 6 doesn't turn the double play that another SS would, or occasionally it means a hit where an out could be made, although this is pretty rare. And in comparing the 6 versus say, a 9, it doesn't mean the 9 is immune to occasionally being a fielding 3 ... it's a roll of the dice kind of thing. But it's not pure luck, it's probability -- the dice will much more often roll toward fielding 3 numbers for the 6. Mind you that over the course of season, luck is mostly evened out.
6's have a couple other adverse effects on defense. For some batting outcomes, the sum of the infield defense is measured. In these cases, obviously, the 6 drags down the sum toward a lesser total, a total that could cost an occasional error or yield an extra hit. Similarly, there are occasions where the team total defense is measured by batting outcomes. The same adverse effects can happen when a 6 is at SS.
So what does it boil down to? How many runs are lost with the 6 at SS? How many runs does Steve gain by playing Chipper at SS? If Steve's options are Mark Loretta (the 9 who filled in occasionally for Chipper and replaced him in late game defensive manuevers before being traded) and Chipper, it's pretty obvious that Chipper's run production will far exceed Steve's other options. I'll try to boil down the offensive and defensive run differentials, with some pseudo-scientific assumptions relative to Steve's situation.
Assumption 1 -- my knowledge of the Master game and BBW is relevant to this discussion
Assumption 2 -- a 9 replaces the 6 in late game substitutions on occasion
Assumption 3 -- the most similar season to SL 2001 was SL 2000
Assumption 4 -- there are about 40 batters faced per game in the SL
Assumption 5 -- the most similar regular to Mark Loretta SL 2001 in SL 2000 was Mark Loretta
Assumption 6 -- the most similar regular to Chipper Jones SL 2001 in SL 2000 was Fernando Tatis
Regarding #'s 5 and 6, as Casey Stengel said, you can look it up:
Loretta - http://espn.go.com/mlb/profiles/stats/batting/5504.html
Chipper - http://espn.go.com/mlb/profiles/stats/batting/5164.html
Tatis - http://espn.go.com/mlb/profiles/stats/batting/5864.html
I'm sure I'll make more assumptions later, and hopefully they are all reasonable.
So how many more runs would Chipper account for offensively vs. Loretta? In the SL 2000 season, Loretta in 559 plate appearances created 72.2 runs, while Tatis created 100.8 in 648 plate appearances. One thing I'd like to note is that in relation to his offensive numbers of MLB 99 vs. SL 2000, Loretta did better than expected ... also relatively better than expected compared to Tatis. But this is pseudo-science, and I can't recreate the scenario with a different subject, unless I waver a little and say he more resembled the Neifi Perez of SL 2000, who had appropriately lesser numbers. Because I can, I'll look at both those scenarios.
One area where Loretta was dissimiliar in the two seasons was his stolen base percentage. To normalize his numbers, let's use the stolen base runs calculation to reduce his total runs created to a number more congruous of what his SL 2001 expectation would be. This will reduce the number to about 71 runs.
Let's then say that we are basing a full season on 660 plate appearances. If we multiply Loretta's 71 runs by 1.18 (660/559), we get a full season total of 83.8. Using Tatis's 117.1 multiplied by 1.01 (660/652), Tatis's full season total becomes 118.3. So in 2001, we'll say that Chipper would be expected to have produced 34.5 more runs than what Loretta would have produced. Of course, if we used Neifi 2000 as our similar player to Loretta, who had 46.5 runs created in 539 pa (reduced to 42 accounting for Neifi's steal success), the expected full season total for Loretta drops to 51.4.
Defensively, this is where my pseudo-science breaks down, but I'll keep going with my assumptions. Given that there are about 40 batters per game, we'll say that this means there are about 50 batting outcomes. That is to say, for every defensive "pitch" there are 1.25 outcomes per batter. Not sure if that's really right, but it sure seems reasonable and 50 is a nice clean number.
How many times does the SS's defensive rating affect those 50 outcomes in a game? The APBA numbers to consider would be 24, 28, 18, 23 and 39, and other rare plays number like 36, 37, 38, 40 and 41. Cumulatively, these numbers may come up about one of out every 10 outcomes. The SS-6 vs. the SS-9 will have the same fielding range (1, 2 or 3) on a portion of these specific outcomes, maybe 40% of the time. Of those outcomes, many times the difference between fielding 1 and 2 and fielding 2 and 3 is negligible -- so that reduces it to about 3 times in a game where the SS-6 has a negative effect. Additionally, the SS-6 will drag down the team and infield sum, I'll say on 1 of the total outcomes in a game. So the total adverse affects in a game happen during 4 of the 50 outcomes.
Adverse affects from the SS position include more errors happening, less double plays turned, and less force plays (in favor of outs at first). If there are 4 adverse affects by the SS-6, let's assume this means an extra out is lost to errors (both team and individually) and three "baserunner advancement prevents" are lost.
If in a standard game there are 40 plate appearances, then total bases may average about 16 and bases on balls plus hit by pitches will average 4. With a 6 at SS, plate appearances will increase to 41 with the lost out, meaning total bases should increase to 16.4 (16 x 1.025) and walks and hbp's increase to 4.1.
In the standard SL game, one team scores about 5.5 runs. They do this with the 16 total bases plus the 4 walks plus hbp's -- what I'll call 20 "sum of bases." For the sake of convenience and solely basing team runs on this data, we'll say then that for every 3.64 bases from a team's "sum of bases," a run is scored (this would be the sum of bases / average runs scored).
With a 6 at SS instead of a 9 and the extra out there becomes an extra plate appearance, the 20 "sum of bases" becomes 20.5 (as with 40 pa's there were 20 bases). The three "baserunner advancement prevents" lost plus the base lost to an error (or two-base error) maybe mean 1.5 more bases per game in the "sum of bases" equation. So with a 6 at SS, the "sum of bases" is increased to 22. Reversing the sum of bases equation, we see that a 6 at SS will yield more runs per game, up to 6.05 runs per game according to the equation (now dividing 22 instead of 20 "sum of bases" by the 3.64 team bases needed for a run).
But a 6 at SS like Chipper will not stay in every game to reach his 660 plate appearances and play his full season. After he hits in an inning, it's reasonable to assume that often with a lead, a manager will give some of his defensive innings to a 9 such as Loretta. But the plate appearances that the backup gets will not affect Chipper's full season, as the 6 with the greater offense will allow for more team plate appearances. The SS-6 may only participate in 9 out of 10 innings, or 90%. So the total runs allowed per game thus decreases to 6 (as the sum of bases is reduced to 21.85).
Similarly, the total team run production will increase because of the 6 at SS. Though the SS-6 stays at 660 plate appearances, the backup will help increase the total number of plate appearances at the SS position to maybe 690. Say that Loretta is the 9 who backs up the 6 in Chipper, Loretta creates 0.127 runs per plate appearance (this number is smaller if we consider Neifi -- only 0.078), meaning with the extra 30 plate appearances, he will add to the team 3.8 runs. The team with Chipper & Loretta then produces 122.1 (w/ Neifi, 120.4), the team with Loretta produces only Loretta's totals, the 83.8 (or Neifi alone produces 51.4). So the numbers are probably between 38 and 71 -- that is to say that Chipper and his backup in a full season will produce between 38 and 71 more runs.
Defensively, the numbers are heavily in favor of the 9. The team with the 6 will lose .5 runs per game, or 81 over the course of a season.
If my assumptions, pseudo-scientific methods, and conclusions are correct, the defensive contributions of a 9 at SS are better, maybe substantially, in terms of run prevention than the offensive contributions of a 6 at SS are. Consider too the fact that the opportunity cost of Steve's equation, Chipper as a 7th round pick vs. Loretta a 21st round pick, and I conclude that the overall value of Chipper or any other good hitting 6 SS does not outweigh the overall value of an average offensive 9 SS.
Note: As of August 9, 2001, Chipper was ripping into this theory by fielding at about a .950 clip. I've used 6's as part-time replacements at SS for the past three seasons, and they've been closer to .900 than .950. This is not to say that Chipper isn't costing Steve's team in other areas defensively, but his fielding percentage (and Steve's relative defensive success) suggests Chipper is not costing Steve nearly as many runs as expected.
Note #2: A Rob Neyer column from August 10, 2001 rebukes a Larry Bowa comment claiming Scott Rolen saves "75 to 100 runs per year." Neyer says that the defensive difference between an average 3B and a "fine" 3B is about 12 runs. In our equation, Neifi and Loretta are probably more than "fine." Also, Chipper is well-below average. And the difference between SS and 3B is significant. But again, the SABRmetrics that Neyer uses suggests that Chipper is not costing Steve as many runs as I suggest. But in defense of my numbers, MLB and BBW are different enough to suggest that the quantifiable difference could be as large as what I suggest.
Back to Draft Notes