This is now the third year that I’ve evaluated outfield arms here at
the Hardball Times and I must say it’s one of my favorite analyses to
do and write up. For one, I myself cannot wait to find out which
outfielders did the best job of controlling the running game during
the previous season. Who excelled in gunning down adversaries as
they tried to gain an extra base, or who, based just on reputation,
most often caused opposing runners to think, “No, I think I’ll hold
second after all.”
It’s hard to get a handle on this during the season, because (as far
as I know), nobody does an arm analysis while the games are still
going on. And, as I’ve discussed before, assists, which are of course available in-season, are
not a good way to judge outfield arms. Naturally, if you don’t have
any additional information, assists are better than nothing. But, and
this is the great thing, we do have more information, namely the
fabulous play-by-play data made available by Retrosheet each winter.
I’ve described the method in detail in my first outfield
arms piece and then last year I gave a concise
version of the analysis. Now, I know nobody clicks on the links to
old articles, but you really should have some idea of how I do all
this, so let me just repeat what I wrote last year:
Using play-by-play data, I consider five different
situations when a throw from the outfield is important:
- Single with runner on first base (second base unoccupied).
- Double with runner on first base.
- Single with runner on second base.
- Fly out with runner on third base, fewer than two outs.
- Fly out with runner on second base, fewer than two outs (third
plays, I add up how often the runner is thrown out (kill) or how often the
runner is “held,” i.e. prevented from taking an extra base (hold). A
comparison with league average allows me to rate the outfielder’s arm.
Wow, that is concise, isn’t it? Here’s a bit more: For each
outfielder, I determine what I call Hold+ and Kill+, which are simply
a player’s hold (or kill) rate divided by that which would be expected
from an average outfielder given the same opportunities. So, for
example, a Kill+ of 120 means the player in question threw out 20%
more baserunners than average. I also calculate runs saved above
average (which I just call Runs), by assigning an average run value,
based on the run expectancy matrix, to each outcome (kill, hold,
advance) in each situation.
Actually, I added a new piece to the analysis this year. While the
five categories listed above account for a large part of the
outfielder’s throwing duties, sometimes a player does some excellent
work in other situations, which I haven’t been considering. I try to
account for these “extra assists” as described in the next section.
Coco Crisp threw out seven runners in 2007, but only one of them
occurred in one of our five categories. What were these six
extra assists? Well, four of them were runners doubled off first
base after Crisp had caught a fly ball or line drive. (This probably
has more to do with Crisp’s acrobatic catches and less about his arm,
but a runner eliminated is a runner eliminated.) Two other
times, Crisp threw out a runner trying to stretch a single into a
Ignoring these six assists by the Boston center fielder
hardly seems fair, does it? Now, Crisp is an unusual case, but a
number of players had three or four of these extra assists, and I
thought the analysis would be improved if I could take these into
Fitting these extra assists into the system outlined above isn’t
really straightforward. It’s not easy to define the
opportunity for doubling up a runner at first base, or throwing out a
batter who tries to stretch a single. So, I had to do something a
little less rigorous for these extra assists. What I did was determine for
each fielder the number of extra assists over and above the average
number, based on defensive innings played. The average number of
extra assists is one every 540 innings for corner outfielders and one
per 600 innings for center fielders. I then multiply by the average
run value of these assists, which I found to be about .57 runs, to get runs saved above average.
To give you an idea of how big the effect is, Crisp picks up 2.3 runs
from extra assists (the most of any outfielder in 2007), while the large
majority of players pick up (or lose) less than one run.
This year, for the first time, I have studied the effects of different
ball parks on outfield throwing. Now, I don’t believe that most parks will
have any real effect on throwing out base runners, although a few
might. For example, the ability to throw out or hold base runners
might be different in parks with artificial turf. Last year Alex
Rios and Mike Cuddyer, both of whose teams (Jays and Twins)
play their home games on fake grass, ranked first and fourth,
respectively, among right fielders. Several readers wrote to suggest
that Rios or Cuddyer may have been helped by playing on turf.
Two years ago, Manny Ramirez led the American League in outfield
assists, and since Manny is considered (not necessarily by me) rotten at everything he does
without a bat in his hand, many peopled wondered if the Green Monster in left
field at Fenway was somehow helping Manny’s throwing. Perhaps he plays
shallower than the average left fielder and hence has a shorter
throw. So, Fenway needs to be looked at, definitely.
On the other hand, the majority of parks don’t have abnormally short
or deep outfield fences or artificial turf and I expect that there
shouldn’t be any material differences in throwing at those parks.
I determined park factors both for throwing out and for holding
base runners (the Kill+ and Hold+ variables). I did this in the usual
way by comparing all throwing situations in a team’s home games with
those in the away games. Before even doing this, I realized that any
park factor analysis was going to be plagued by small sample sizes. A
typical number of throwing chances in a season (in my system, of
course) is around 200 per team per position. Even if we use multi-year
samples, we need to be aware of sample size issues.
The first thing I did was to focus on the astroturf parks: here are
the Kill+ and Hold+ park factors, calculated combining all three
outfield positions (values over 100 favor the outfielder):
Park PF(Hold+) PF(Kill+) TOR 100 101 MIN 98 104 TBA 100 104 ------------------------------- Turf 100 103
So, if there is any benefit from playing on turf, it’s quite small.
These numbers are based on data from 2000 to 2007, but despite using
eight seasons, I’m not convinced that the 103 factor for Kill+ is
significant. A detailed statistical analysis, which I haven’t done
yet, is needed to tell us if that 103 is meaningfully different from
What about left field at Fenway? There I find a PF(Hold+) of 107 and
PF(Kill+) of 104, again using data from 2000-2007. That 107 seems to
be saying that runners tend to hold up because they know the ball
comes quickly off the wall. However, I found lots of other parks
where at least one field had a Hold+ park factor of 107 or more: RF in
Philadelphia, for example, and CF in San Francisco and Cleveland, and
several others. In all, left field in Fenway was one of 11 fields with
a Hold+ PF of 107 or greater. This makes me very skeptical—I
cannot help but think that the noise in these numbers is drowning out
Note that I’m not saying that we absolutely don’t need park factors
for outfield arms. I’m just saying the issue needs some careful study
and statistical analysis and I haven’t had the chance to do that yet.
My intuition tells me that any effects that we may eventually turn up
will be small, so I think it’s worthwhile publishing the results as I
always have, without corrections for home park.
Ok, that’s enough theory, let’s get to the rankings. Read on.
The gold medal for the best right field arm of 2007 goes to Mike
Cuddyer. Let’s break down Cuddyer’s opportunities to see how he did
Cuddyer, Mike - MIN - RF Situatation | Opps | H K | Hexp Kexp S-1B | 58 | 33 6 | 33.9 0.7 S-2B | 37 | 16 5 | 12.1 2.6 D-1B | 34 | 20 4 | 20.1 1.5 OF-3B | 16 | 5 0 | 3.8 0.8 OF-2B | 19 | 10 0 | 10.9 0.5 Overall | 164 | 84 15 | 80.8 6.1 Notation: Opps - number of opportunities H - Hold: runner did not take extra base. K - Kill: runner (or batter) thrown out on bases. Hexp - Expected Holds based on league average play. Kexp - Expected Kills based on league average play.
Here’s the thing about saving your team runs with your arm: it sure
helps if you can gun down more than your share of base runners. Holding
a runner is great, but after the play he’s still out there on base,
isn’t he? Throw him out and you can be sure of one thing: that runner
is not going to score. Cuddyer nailed 15 base runners when his
expectation was only six. He also picked up one run from extra
assists. Cuddyer was also above average in holding runners, but it was
the kills that helped the most.
The following table shows all right fielders who had at least 80
opportunities in 2007. The ranking is by Runs/200, or runs saved per
200 opportunities (which is about a full season’s worth):
Name Opps Kill+ Hold+ Runs Runs/200 Cuddyer_Mike 164 245 110 12.0 14.6 Francoeur_Jeff 177 241 112 11.9 13.4 Victorino_Shane 135 170 128 8.2 12.2 Young_Delmon 171 168 121 9.2 10.8 Rios_Alexis 160 124 112 4.5 5.6 Teahen_Mark 183 172 95 4.8 5.3 Abreu_Bobby 177 59 116 3.2 3.6 Hermida_Jeremy 146 114 102 2.1 2.8 Markakis_Nick 203 112 104 1.9 1.9 Guerrero_Vladimir 121 92 113 1.1 1.8 Guillen_Jose 212 83 107 1.2 1.1 Kearns_Austin 200 98 98 -0.3 -0.3 Scott_Luke 95 116 92 -0.6 -1.2 Ethier_Andre 112 175 78 -0.8 -1.5 Cruz_Nelson 90 128 88 -0.7 -1.6 Ordonez_Magglio 142 58 108 -1.3 -1.8 Nixon_Trot 82 127 101 -1.1 -2.6 Winn_Randy 115 0 101 -1.8 -3.1 Drew_J.D. 114 73 94 -1.8 -3.1 Dye_Jermaine 185 84 89 -3.3 -3.6 Encarnacion_Juan 82 33 105 -1.8 -4.3 Hawpe_Brad 144 68 90 -3.2 -4.4 Nady_Xavier 94 89 84 -2.3 -4.9 Hart_Corey 144 56 95 -5.6 -7.7 Green_Shawn 97 27 97 -3.8 -7.9 Griffey_Ken 162 31 84 -6.5 -8.1 Giles_Brian 130 22 83 -7.6 -11.6
Braves right fielder Jeff Francoeur was great again this year, after excellent showings in
2005 and 2006.
Shane Victorino was top notch in right field for the Phillies, his Hold+
of 128 was the highest of any qualifying outfielder in 2007. Delmon
Young rounds out the top four, helped out by five extra assists.
Rios, who had the best arm in 2006, was off a bit in 2007, but he
still showed very well. Vlad Guerrero, although he has the great
reputation as evidenced by his high Hold+ of 113, has only been around
average for each of three years that I’ve been doing the analysis. (He
was much better earlier in his career, as I wrote in last year’s
Hardball Times Baseball Annual).
Shawn Green, who was the worst in the majors last year, was terrible
again, although this year he managed to scramble off the lowest rung,
thanks to poor throwing of Griffey and, especially, Brian Giles.
B.J. Upton had the best center field arm of 2007—what is it
with these ex-second basemen? (See Soriano, below.) I haven’t seen
much of Upton, so I checked a couple of his throws on video and he
does seem to have a strong arm. It will be interesting to see if he
can keep up the strong performance moving forward.
Name Opps Kill+ Hold+ Runs Runs/200 Upton_B.J. 108 224 101 4.6 8.5 Taveras_Willy 99 240 102 3.0 6.1 Edmonds_Jim 125 166 111 3.5 5.6 Suzuki_Ichiro 199 143 116 5.2 5.2 Freel_Ryan 81 120 104 2.0 5.0 Cabrera_Melky 185 207 99 4.2 4.5 Amezaga_Alfredo 103 266 86 2.0 3.9 Patterson_Corey 138 132 101 2.6 3.8 Roberts_Dave 100 168 103 1.9 3.7 Rowand_Aaron 204 79 115 3.7 3.6 Jones_Andruw 185 56 126 2.4 2.6 Pence_Hunter 117 141 106 0.9 1.5 Lofton_Kenny 107 97 104 0.8 1.5 Hall_Bill 154 119 105 1.0 1.3 Granderson_Curtis 222 99 100 1.2 1.1 Young_Chris 168 61 104 0.2 0.2 Cameron_Mike 178 114 108 0.1 0.1 Crisp_Coco 148 24 105 -0.1 -0.2 Beltran_Carlos 155 120 90 -0.2 -0.2 Duffy_Chris 87 39 106 -0.7 -1.7 Wells_Vernon 140 97 93 -1.6 -2.3 Sizemore_Grady 193 73 105 -2.6 -2.7 DeJesus_David 182 90 103 -2.7 -3.0 Matthews_Gary 154 95 86 -2.4 -3.1 Hunter_Torii 174 64 97 -3.1 -3.6 McLouth_Nate 92 41 88 -2.7 -5.8 Swisher_Nick 84 0 94 -2.5 -6.1 Owens_Jerry 101 36 90 -3.5 -6.9 Logan_Nook 111 31 89 -4.1 -7.4 Pierre_Juan 184 56 83 -6.9 -7.5
I usually, for purely selfish reasons, hope that Ichiro will take the
top ranking in my analysis. In 2005 Suzuki came out about average (in right
field) and in 2006 he ranked fifth among right fielders (5.1 Runs/200), but that
didn’t meet the expectations of his many fans, a goodly number of whom
wrote to say that my rankings were, ahem, less than accurate. His
2007 performance, now in center, was almost a carbon copy of 2006. I
suppose I should be expecting some email.
Have a look at Nick Swisher‘s Kill+ — actually, he doesn’t have
one. To be fair, Swisher did have one extra assist in his 481 center
field innings. The Oakland outfielder also played around 400 innings
in right field, where he recorded a grand total of two assists (both
extra) and had a Runs/200 of -0.4. I’m not sure how the A’s plan to
deploy Swisher and Travis Buck in left and right field, but it actually (from
the throwing viewpoint) doesn’t matter much. Buck was pretty bad, too,
in limited action (-9 Runs/200 in 57 opps). [Update: scratch that last thought, Swish is now a White Sox.]
Note that you have to interpret Kill+ and Hold+ with some caution.
For example, compare Hunter Pence and Corey Patterson in the table
above. (Whoa! I just spotted Torii Hunter below Crisp, DeJesus and
Matthews. Who’d've thunk it?) Anyway, Pence was better than Patterson
in both Hold+ and Kill+, but ranks lower in Runs/200. The first thing
you should know is that both of these guys had five kills against 3.6 expected
for Pence and 3.8 expected for Patterson. In other words, they were
virtually equal in throwing out runners. Furthermore, Patterson’s
kills (and holds) occurred in slightly more critical situations, in
terms of run value, than Hunter’s did, hence the slightly higher
ranking for Patterson. Lastly, Patterson gained a little in extra
assists, while Pence lost a fraction of a run. The lesson here: don’t
split hairs over rankings that are fairly close.
Juan Pierre had the worst arm of any full-time center fielder, but
that is no surprise to anybody. No need to split hairs with Pierre,
his is truly a noodle arm, and it’s definitely not al dente (sorry).
People don’t usually talk about the arms of left fielders, but Alfonso
Soriano is definitely generating some buzz. This past season, Soriano tied for
the major league lead in OF assists with 19 and was the only left
fielder with more than 13 assists. In 2006, he led all outfielders
with 22 assists, no other left fielder had more than 12. This year, in terms of
Runs/200, our ranking criterion, the ex-second baseman completely
lapped his left field mates, lapped them two or three times, as you
can see in the table below.
Name Opps Kill+ Hold+ Runs Runs/200 Soriano_Alfonso 152 266 108 12.9 17.0 Monroe_Craig 103 154 106 2.2 4.3 Willingham_Josh 182 124 113 3.1 3.4 Anderson_Garret 100 147 100 1.6 3.2 Alou_Moises 101 151 98 1.6 3.2 Matsui_Hideki 133 109 103 1.9 2.9 Michaels_Jason 82 64 110 1.0 2.4 Bay_Jason 222 136 100 1.5 1.4 Duncan_Chris 130 84 101 0.5 0.7 Holliday_Matt 182 101 101 0.5 0.5 Ibanez_Raul 176 121 96 0.4 0.4 Byrnes_Eric 116 133 85 -0.2 -0.4 Jenkins_Geoff 151 93 97 -0.5 -0.6 Lee_Carlos 188 89 96 -0.7 -0.7 Burrell_Pat 142 125 96 -0.5 -0.7 Brown_Emil 81 190 87 -0.3 -0.7 Ramirez_Manny 129 110 90 -0.9 -1.5 Payton_Jay 157 18 106 -1.2 -1.5 Willits_Reggie 84 66 104 -0.7 -1.8 Crawford_Carl 176 32 108 -1.7 -2.0 Gathright_Joey 85 111 94 -1.7 -3.9 Harris_Willie 84 71 93 -1.8 -4.3 Church_Ryan 109 79 101 -2.4 -4.3 Dunn_Adam 161 57 95 -4.0 -5.0 Diaz_Matt 85 95 89 -2.1 -5.0 Kubel_Jason 84 32 98 -2.7 -6.4 Bonds_Barry 89 32 99 -2.9 -6.6 Gonzalez_Luis 120 101 81 -5.8 -9.6 Stewart_Shannon 153 55 85 -8.8 -11.6
The left field arms fall off pretty quickly after Soriano, although Craig Monroe seems
quite solid: he ranked fourth among left fielders last year.
Ex-catcher Josh Willingham, after showing around average last year,
was more impressive in 2007.
Barry Bonds, who earlier in his career had a very good arm, thanks
more to accuracy and a quick release than to brute strength, was
near the bottom of the heap in 2007. Obviously, it’s not surprising
that a 42-year-old with bum knees had trouble throwing runners out.
Shannon Stewart was the worst of the lot in left, costing his team
almost one game in the standings with his weak throwing (not that it
mattered for the A’s in 2007).
Graph with colored dots
A good way to visualize outfield arms is to make to plot Hold+ versus
Kill+, as you see in the graph on the right. Each point represents a
single player and the color tells you if he played right, center or
left. Players above the horizontal dotted line were above-average in
throwing out runners, while players to the right of the vertical line
were above-average in holding runners. Naturally, the best overall
arms are in the northeast quadrant, while the candy arms can be found
in the southwest.
Andre Ethier, right there in the middle of the northwest quadrant, is
a curious case as he is very good at throwing out batters, but they
still run wild on him. Runners should know better, since he was also
very good, albeit in left field, in 2006. Perhaps this past season they figured they
could run on a mere “left fielder.” I expect his hold rate to rise
next year (assuming he gets playing time, either in LA or elsewhere).
Ethier’s opposite number is Andruw Jones, who sits alone out in the
middle of the southeast quadrant. Andruw’s Hold+ of 126 is among the
very best in 2007, while he only killed about half of the expected
number. Hmmm, is Andruw coasting on his reputation? Jones was the
top throwing center fielder in 2005, but slipped to about average in
2006, with a poor Kill+ and average Hold+.
Another interesting case is Alfredo Amezaga (look near the very top of
the graph) who threw out more than twice the number of expected
runners (eight killed, three expected), but suffered from a
below-average hold rate. I’ll be looking forward to see if runners get
more cautious against Amezaga in the future.
Do these outfield arm evaluations make sense? I am not an impartial observer, of course, but I think they do. It’s interesting to compare
these ranking with the results from Tom Tango’s Fans’ Scouting Report 2007 (a fabulous resource, by the way). The fans seem to be in agreement on most of our top outfield arms, and the worst ones, too. Of course, there are plenty of disagreements, as well: to take just one, Carlos Beltran rates very highly with the fans, but only appears average in this analysis. I think a detailed comparison of the results from the fans’ poll and this study would be very interesting. Since the fans rate outfield arms in three separate categories: “Release”, “Strength” and “Accuracy”, it might be useful in identifying which of the three characteristics is most important in controlling the running game.