Cannons in the bushes: rating minor league outfielders’ arms

Down in the low minors there is a 23-year-old center fielder who guns down every baserunner trying to take an extra base on his arm. No kidding. Had his 2009 throwing performance come in the majors, it would rank as one of the best in history.

Unfortunately it is unlikely we’ll ever see him play in a big league park. You don’t expect an outfielder with a .304 OBP and an even lower slugging percentage in A-ball to make it to the Show.

More on him later.

Thanks to Sean Smith’s TotalZone metric, individual players have advanced (i.e., something more interesting than fielding percentage) fielding stats listed at Minor League Splits. However, as Sean explained in his first article on TotalZone here at THT, he was “attempting to measure the outfielder’s range, not his arm.”

Minor league outfielders’ arms are exactly what we’ll try to measure in the series of articles beginning with this one.

Outfield arms have been rated since the early months of 2006 here at THT, thanks to the work of John Walsh. We had it in our stats pages and the 2007 Annual featured an article (again by John Walsh) on the best arms in history. Roberto Clemente came out on top, so you can bet the metric is quite good.

I invite you to follow the links I provided a few lines back, but here’s a quick summary of what John did in the past for MLB outfielders—and what I have tried to replicate for minor leaguers.

Five situations have been considered:

{exp:list_maker}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 base unoccupied). {/exp:list_maker}

They are not all the situations in which an outfielder’s arm is tested, but they cover most of the cases. In each situation, you consider whether the lead runner took the extra base (for example third base for the first situation), or held, or was thrown out. You compare the result to what happens in MLB on average. The comparison is performed with the run expectancy matrix, so the resulting difference in expected runs is assigned to the outfielder making the play. Add up the results of every play by an outfielder and you have his arm rating for the season.

If this concise explanation doesn’t satisfy you, once again I invite you to read the work of the giants on whose shoulders I’m sitting.

Back to the minors.

For the majors, we have Retrosheet. You can’t imagine how important and time-saving it is to have proof-read data until you have to put your hands on something not so clean. I parsed minor league play-by-play data from MLBAM’s Gameday (a big thank you to Kyle Willkomm, author of Baseball On A Stick), then I used whatever SQL and Regular Expressions tricks I was aware of to determine the destination of the baserunners. In the end I had to manually recode a few dozen remaining plays.

Now that you have read of my pains, you are entitled to some results. Following are the players I have found to be the 10 best throwing right fielders in the minors in 2009.

But not quite yet. There’s still the warnings/disclaimer part.

First, I have not calculated park factors – I have left them for another day.

Second, these are the 2009 arm ratings: I have performed no regression toward the mean. Thus, what I’ll show are numbers for a one-year performance, not player values.

A Hardball Times Update
Goodbye for now.

Third, ratings have been calculated as if the outfielder had delivered his performance on a major league field. This means I have not attempted a translation to an MLB equivalent nor calculated a league-specific run environment, so an outfield assist in Triple-A has been considered equal to an assist in A-ball.

These warnings mean that
{exp:list_maker}If you don’t like the choices I have made, either you do your own arm rating or wait until I improve mine. (But you are free and encouraged to comment on them);
You should not take my arm rating numbers as true run-prevented figures, though I believe the ranking won’t change too much once park factors, translations and other tweakings are implemented. {/exp:list_maker}

The best minor league right fielder arms of 2009.

Each player has a stat line attached to his name: the first number (bold) is the arm rating, the second is the TotalZone defensive run value, then the triplet that follows is the classic batting average/on-base percentage/slugging percentage line. An outfielder must have 75 opportunities (plays that satisfy one of the five conditions previously listed) to qualify.

10. Terry Evans (4; -9; .291/.341/.520)
Right handed, at 28 (in a few days) he has had a few cups of coffee with the Angels. He comes from the Cardinals farm system; he was dealt to the Halos in 2006 for Jeff Weaver. He played most of his season in Salt Lake City, sometimes patrolling center field too. He increased his strikeouts to 146.

9. Ryan Harvey (4; 4; .246/.309/.519)
Right handed, 25 years old. A first-round pick by the Cubs (No. 6 in 2003), he was released last spring and played the season at Tulsa, the Double-A affiliate of the Rockies. He strikes out a lot.

8. Jason Heyward (4; 19; .323/.408/.555)
Baseball America has him as the Braves’ top prospect. He is lefty and just 20. Today we have found that he has a very good arm, but his TotalZone numbers are really off the charts (+10 is a very good fielder, +15 or more is outstanding). Add to the mix a good plate discipline (51 strikeouts, 51 walks).

7. Roberto Valencia (5; 15; .311/.350/.357)
He is 21, bats left and throws right. He plays for the Piratas de Campeche in the Triple-A Mexican League, so his team has no affiliation to any MLB club. He’s a groundball singles hitter with no power—and a great defender, looking at Sean’s and my numbers.

6. Omar De La Torre (5; -13; .339/.386/.552)
I thought about leaving minor league veterans out of my list, but in the end I decided to give credit to whomever had a good arm. He is 30 and right handed, and doesn’t seem to have value with the glove. I wonder how much the 7,300-foot altitude (Puebla, Mexico) is helping his throws.

5. Roger Kieschnick (5; 12; .296/.345/.532)
Right-handed thrower, bats from the left side. At 22 (23 in a few days) he debuted in pro ball last season for the San Jose Giants (Single-A advanced). One season of numbers shows many strikeouts, extra base power (37 doubles, eight triples, 23 homers) and very good defense. Keep an eye on his development.

4. Christian Quintero (5; 2; .334/.391/.526)
Another Mexican League veteran, he’ll be 34 by the start of next season. Right handed for the Guerrieros de Oaxaca, he was on the Tampa Yankees of the Gulf Coast League back in 1996.

3. Edwar Gonzalez (5; 14; .232/.296/.329)
Right handed, also hailing from the GCL Yankees, at 27 he has not made past to Double-A ball yet: numbers show his bat is responsible for that.

2. Melky Mesa (5; 11; .225/.309/.423)
Going for 23, righty, and—again—a Yankee. His offensive numbers clearly show that he’s not suited for the bigs (168 Ks in the Sally League). Too bad because, unless we are dealing with a one-year fluke, his defense looks stellar. On a per-play base his arm rating would put him on top, ahead of…

1. Denis Phipps (8; 15; .237/.288/.385)
And here’s another player who would be very useful if baseball, like football, had a team for defense only. He was born right handed 24 years ago in the great forge of baseball talent of San Pedro de Marcoris and belongs to the Cincinnati Reds organization.

Next time: The center fielders and the kid who guns down baserunners by the dozen.

Wait, I forgot the Venus de Milos of the bushes: Wilfredo Arano (-7), Maurice Gartrell (-6), Mitch Einertson (-5), Robert Perez (-4) and Nicholas Francis (-4).

References & Resources
Sean Smith’s TotalZone first work and its adaptation to minor league baseball.

John Walsh’s outfielders arm rating.

Minor League Stats from, FanGraphs, Baseball-Reference, Minor League Splits.

Newest Most Voted
Inline Feedbacks
View all comments
Matt Lentzner
14 years ago

In a way, this is peculiar. You would think that if a player was athletic and had a good arm, but couldn’t hit he’d be converted to a pitcher.

I guess I don’t know when it’s too late to make that move.

Peter Jensen
14 years ago

Max – You know that I am a big fan of the analysis that you have written for THT, but I have a lot of questions about this post.  How did you calculate the runs saved for each play?  I know you said this:

The comparison is performed with the run expectancy matrix, so the resulting difference in expected runs is assigned to the outfielder making the play.

but that leaves several choices.  You can either subtract the RE of the resulting BaseOut situation from the RE of the initial BaseOut state or you could do what Walsh did and calculate an league average value for advancing the additional base, or being thrown out independent of the actual BaseOut situation.  I believe Walsh’s method gives a more accurate measure of the true value of the skill.

Also, you define the 5 situations that you are evaluating, but Walsh correctly showed that advancement on the first three of your 5 situations is highly dependent on the number of outs.  Did you evaluate each of the out situations separately?

John didn’t mention it in his article, but runner advancement is also dependent on whether the hit was a ground ball, fly ball, or line drive.  Did you look at each of these situations separately? 

Even if you created these additional bins you still have two problems that are impossible to solve for the minor leagues.  First, unpublished analysis that I have done has shown that runner advancement is also dependent on whether the runner was running on the pitch, either as a hit an run or as part of a steal.  This information is available through Retrosheet for the major leagues, but is not in the MLBAM minor league PBP information.  Second, Walsh mentions that the same methodology he uses to measure outfield arms is also used to measure runs generated by runners in taking an extra base.  But each metric gives all the credit for the runs either to the outfielder as runs saved by his good arm or to the runner as runs produced by his good baserunning.  This is double counting.  The correct answer lies somewhere in between and no analysis that I know of has figure out an equitable way to apportion the runs between runner and outfielder and the batter for creating the type of hit that he produced.  The probability is that the run values that you have attributed to the outfielders arm should actually be closer to 1/3 to 1/2 that amount, with the other 2/3 to 1/2 going to either the batter or the runner.

Max Marchi
14 years ago


1. I did as John did.

2. Yes, I evaluated each of the out situations separately.

3. No, I did not separate FBs / GBs / LDs for the hits; this is something I’m definitely going to do in a future version, in which I’m also planning to factor the position-where-the-ball-was-fielded in. I don’t think it’s right behind the corner, because the work of normalizing the gameday coordinates must be tackled first (as you have shown some time ago).

4. Double counting is an issue that occurs many times in baseball stats, and I believe it’s quite hard to quantify the part to apportion to one side and the part to apportion to the other. In this specific example, I would even throw the third base coach into the mix.

And finally, thank you for your support!