Wrap Up, Part One
Well, 2007 is finally in the books. The Mariners finish the year 88-74 despite being outscored by 19 runs on the year, and, of course, we like to break things down, and occassionally, we’ve been known to use some numbers.
We’ll have more post-season analysis in the days to follow, but for now, here’s a quick look at the rankings of the offensive players in terms of runs added above average. This is not position adjusted – we’re simply look at how much value each player added as an offensive player, accumulating runs through hitting and baserunning, compard to a league average hitter. This is based on linear weights, which essentially gives each play an average run value, then sums them up the total individual contributions.
Ichiro: +24 runs
Ibanez: +19 runs
Guillen +16 runs
Vidro: +12 runs
Beltre: +10 runs
Johjima: +1 run
Betancourt: -5 runs
Sexson: -9 runs
Lopez: -22 runs
No big surprises here – Ichiro’s the best hitter on a team of above average hitters, Lopez had a disaster of a season at the plate, and Sexson was terrible as well.
Now, just for fun, let’s do some positional adjustments. Generally, statistical analysis will do position adjustments based on the average performance of hitters at that position, but the more I’ve looked at this, the less I think that’s the right way to do it. Let’s use 2007 National League middle infielders as an example, for instance.
Average NL SS: .279/.337/.420
Average NL 2B: .272/.341/.418
Second baseman and shortstops in the National League put up basically the same line this year. A normal positional adjustment would compare Dan Uggla and Hanley Ramirez to the same baseline. But think about this for a second – does that make any sense at all? We know that almost every shortstop on the planet is capable of playing a quality second base, but very few second baseman can play a credible major league shortstop. Just by the fact that they’ve been deemed able to play shortstop in the major leagues, the SS group is showing a demonstrable skill that the 2B group is not. That has real value, and comparing a batting line to the average of his peers will miss that.
So, instead, I’ve come around to the idea that we should do a position adjustment based on the defensive spectrum. We know that shortstop is harder to play than second base and center field is harder to play than left field, and there’s no reason to expect the relative difficulty of those positions to change on a year-to-year basis, so we can use fixed values to determine the positional adjustments. Tango’s done most of the legwork, and if you’re really interested in the nuts and bolts, I’d recommend checking out his blog, but here’s the overview for the rest of us:
C: +10 runs
SS/CF: +5 runs
2B/3B: +0 runs
LF/RF: -5 runs
1B: -10 runs
DH: -15 runs
This fits pretty well with what we understand about the game. Catcher’s an island unto itself, but the premium defenders play the tougher up the middle positions, the bad defenders get hid at designated hitter or first base, and some of them sneak into corner outfield spots, and the in between guys end up at second or third base. So, applying the defensive spectrum to a positional adjustments, we get the following values for the Mariners’ starting position players.
Ichiro: +29 runs
Ibanez: +14 runs
Guillen: +11 runs
Johjima: +11 runs
Beltre: +10 runs
Betancourt: +0 runs
Vidro: -3 runs
Sexson: -19 runs
Lopez: -22 runs
This gives us a pretty accurate description of how valuable the Mariners line-up was offensively. Ichiro’s the shining star, with Ibanez/Guillen/Johjima/Beltre all being about the same, Betancourt and Vidro hanging around average, and Sexson and Lopez killing rallies every day.
So, there you go – there’s the rankings of the position players as offensive contributors. Tomorrow, we’ll look at the defensive contributions and sum up the position players value in total. And yes, defense changes things. A lot.
Okay. I guess I’m just going to have to accept being wrong without really getting it. (It did occur to me, Dave, that young players like Jones would be more willing to swallow hard and accept the role.)
It is true, though, isn’t it, that the DH position generally does not have the best hitting totals, as logic would lead us to expect from the pool of available players?
That, of course, I understand.
How is the -15 number arrived at? Is it an estimate, or is it derived from something?
148 – Yeah I guess we’re talking apples to oranges here then. These #s are a very basic way of accounting for the fact there is added value in having someone that COULD do something other than DH, just because they can in fact do that additional thing.
But as you state in any scenario based in reality where you need to account for salaries, player egos, lack of talent at other positions, etc., the actual amount of players you WOULD play at DH full time becomes much smaller. So you’re point is somewhat applicable to the reality of the game. But just in assigning values I’m sure you can understand why it is more valuable to have someone that COULD DH (ie anyone) as opposed to someone that COULD DH or play C. (ie smallest group of avail players) And of course this is not looking at the actual skills of the players yet.
Well, you’re talking about something different. Let’s take Willie Bloomquist as an example, because I haven’t bashed on Willie in a while (tee hee!).
No team would realistically put Willie Bloomquist in as their everyday DH; there is always going to be a better option available as long as there’s someone else on your team that hasn’t lost both their arms to gangrene.
However, he still could potentially fill that position “defensively”, since there is no defense to play at DH, so he is part of the available pool of players at DH; by way of contrast, David Ortiz couldn’t play center field defensively (although it’d be amusing to watch him try), so he is not part of the available pool of players at that position.
It doesn’t really have anything to do with whether a player would be realistically used at a position, just whether they can fill its defensive requirements.
Mistake – Paragraph 2 in 154 should be your, not you’re
Apparently I need sleep, 154 should also say more valuable to have someone that could DH or play C as opposed to someone that can only DH
It is true, though, isn’t it, that the DH position generally does not have the best hitting totals, as logic would lead us to expect from the pool of available players?
I think the devil is in the details here. As I mentioned, a lot of teams just don’t even attempt to fill the DH spot with one player, instead using it for other purposes. If we looked at only full-time designated hitters, we’d see that they are, indeed, the best hitters of any position.
Because the position offers unique advantages in terms of creativity, adjusting back to a league average of players chosen by the teams won’t give us a realistic assessment of the available options. Remember, Reggie Willits DH’d 14 times this year.
As for the -15 run number, there’s no rock solid math behind it, so you can move it to -13 or -14 if you want. But, there’s no way to argue that it can be higher than -11, as the penalty has to be higher than that of playing first base, and I’d argue that a 1 or 2 run penalty isn’t stiff enough. If you think its 3 or 4 runs, thats fine – you may be right, and I’m okay with it being -13 or -14 or -15. The idea is still the same.
As you’ve probably noticed, I’m not trying to do super precise work here. I’m not handing out MVPs based on the differences between fractions of a run with this tool. For the purposes we’re using it for, the conclusions we’re drawing won’t change.
Thanks.
If Ortiz is a -10 fielder at 1B relative to all 1B, and he gets the -10 1B penalty, if you move him to DH, shouldn’t he get something close to a -20 penalty? That’s WHY he is a DH, because his fielding is below the -10 acceptable level.
(It comes out to -15 instead of -20, because it’s harder to DH. You give them a +5 more “hitting off the bench” runs, so that the overall penalty is -15.)
***
If this was high school, your best fielders are at SS, CF, maybe 3B and LF. Now, if you were to compare to the average player at each position, the average SS = average 2B = average DH. But, that’s ludicrous isn’t it? If every team’s best player is at SS or CF, how can you then rate the average SS as being equal to the average CF?
The average QB is not equal to the average OT is he?
That’s the point. You have to approach this from reality. And the reality is that not every position is necessarily equal, even if you need to have someone man each position.
Did you know that in the early50s that the offensive output of the CF was HIGHER than 1B? Would any sane person conclude that the average 1B at that time was equal to the average CF? Clearly the average CF is a better fielder than the average 1B. If he’s also a better hitter, then the average CF at that time must have been better than the average 1B.
Approach this from a reality perspective, not from a “let me fit everything into a nice matrix” perspective. Mathematical gymnastics we don’t need.
“then rate the average SS as being equal to the average CF?” should say
“then rate the average SS as being equal to the average 2B?”
If we looked at only full-time designated hitters, we’d see that they are, indeed, the best hitters of any position.
That’s pretty easily done. Cumulative line for all DHs this year who qualified for the batting title: .285/.390/.487.
That group includes Ortiz, Vidro, Thomas, Thome, Hafner, Sheffield, Cust, and Huff. I included the latter two because they played DH more frequently than they played any position in the field, and consistent with the interpretation, they drag the averages down. Notable full-time DHs who did not qualify: Piazza (injury, shared time with Cust) and Sosa.
Still, that wouldn’t necessarily be the case every year.
As well, if you only look at the top 8 DH in playing time, you need to look at the top 8 at each position to make the comparison fair.
Here’s a couple of therads that you might find interesting:
http://www.insidethebook.com/ee/index.php/site/comments/defense_spectrum/
http://www.insidethebook.com/ee/index.php/site/comments/replacement_level_fielding/
Dave, you’re saying that because of the filtering you describe in comment 45, it doesn’t make sense to normalize, for example, SS and 2B to the same baseline, since the SS pool is more restricted. Yet in the original post you mention that in 2007 the offensive numbers were about the same for SS and 2B. Was 2007 just a fluke in that regard? If there are offensive differences between positions or groups of positions, normalizing to the position-specific or group-specific averages themselves should be okay, rather than doing a more manual weighting of +5, -5, etc.
Also, this normalizing scheme does seem to be injecting defensive considerations into an offensive measure, which I’m not sure is necessary. It’s basically adding a caveat related to defense, saying “yes, players of position X have lower offensive numbers than position Y, but that’s because they have stricter criteria for defense at that position.” Why not keep offensive and defensive stats separate and have the player’s overall value determined by the combination? I see the point of your normalization, I’m just not sure it’s necessary for an offensive stat.
It doesn’t have to be necessary for an offensive stat. In fact, you don’t want to do it.
However, if you are trying to “describe” something, you want the positional adjustment, since the reader will be thinking it anyway. Jeter’s offense is highly regarded because he is a SS, and wouldn’t be so highly regarded if he was a LF.
In any case, you want all three components (offense, defense, position). Part of the confusion seems to be that Dave only presented two of the three here.
“normalizing to the position-specific or group-specific averages themselves should be okay”… as long as you don’t just use a 1-yr average. It’s nonsensical to think that the average RF = average DH, if we know that the RF hit better and fields better. Same with saying SS = 2B. (Over the long haul, the average 2B will hit better than the average SS.)