Batter's Box Interactive Magazine Batter's Box Interactive Magazine Batter's Box Interactive Magazine
A Happy Thanksgiving to all our readers in the United States.

Manny years ago, Bill James looked at various indicators of speed in a typical stat line and decided to combine them to derive a number which would roughly indicate the speed of the player on a scale of 1 to 10.

I've done something similar with minor league stats, but I've left out the position component that James used (centerfielders and middle infielders were given high ratings in the category). I've tweaked some of the criteria and weighted them a bit differently than James did. Here are the 4 categories I used:

1) Triples as a percentage of double and triples - the idea being that a fast player can turn a double into a triple more often than an average one, while a slow player will almost always stop at second;

2) Stolen Base attempts (weighted 2*SB + 1*CS) per estimated times on first base - the idea being that fast runners are allowed to attemp a steal and so even a caught stealing is an indication of speed;

3) GIDP per out in play (AB-H-K), adjusted for estimated runners on base. To adjust for runners on base, I look at actual RBIs in relation to expected RBI, given the particular combination of hits the player produced. A batter batting in the middle of the order will likely get more actual RBI than would be warranted by his combination of hits, thus he ought to hit into more DPs than his speed warrants;

4) Runs scored in relation to expected runs scored (based on hits and walks etc), excluding homeruns;

The 4 categories do not have an equal weight in the composite measure: the SB Attempts category is weighted at roughly twice the others (which are nearly equally weighted).

The resulting composite speed score is divided by the league median to produce the final speed ratio. It is a rare player exceeds 2.0 or goes below 0.5.

Here are 2003 speed scores for all Toronto farmhands with 200+ PA for a given team:

Syracuse (AAA): Wise 1.68, Werth 1.67, Alvarez 1.36, Sequea 1.36, G Williams 1.14, Moriarty 1.11, Ryan 1.11, Colangelo 0.94, Pond 0.85, Aven 0.85, Clark 0.81, Gross 0.76, Burnham 0.73, Huckaby 0.70, Cash 0.67, Zuniga 0.66

New Haven (AA): Thompson 1.62, Adams 1.20, Singleton 1.19, Rios 1.08, Fagan 1.01, Logan 0.92, Rich 0.86, Solano 0.84, Chiaffredo 0.84, Pond 0.83, Gross 0.82, Quiroz 0.64, Griffin 0.64

Dunedin (A+): Godwin 1.54, Adams 1.30, Mayorsan 1.11, Tablado 1.01, Jova 0.96, Davenport 0.82, Snyder 0.77, Waugh 0.74, Whittaker 0.74, Whittaker 0.74, Cosby 0.73, McEachern 0.66

Charleston (A): Medina 0.96, Jova 0.95, Smith 0.91, Tablado 0.91, Dragicevich 0.87, Rivera 0.82, Owens 0.78, Hassey 0.75, Corrente 0.70, Schneider 0.64, Zinsman 0.54

A few tentative conclusions may be drawn from these numbers. While Gross' speed is a little below a typical corner outfielder's, Griffin's very low score suggests that he won't be quick enough to play a corner OF slot in the majors and will likely be moved to 1B or DH. Rios' score is near average, which indicates that he's in danger of being shifted to a corner slot when he makes the majors (with Vernon Wells in Toronto, that seemed likely anyway). Russ Adams is quick enough to be a middle infielder in the majors, but so are Jimmy Alvarez and Jorge Sequea. Jayson Werth ought to be shopped to organisations that value speed-power outfielders.

Some readers may wonder why I ignored position in calculating speed scores. The reason is that I use speed scores to adjust the defensive position factor used to adjust batting component stats. It's difficult and time-consuming to evaluate minor league fielding stats, so I decided to incorporate speed as a factor in determining future defensive value. Minor league stats list the number of games played in the outfield, but do not break that down by left, centre and right. Speed scores can remedy that by adjusting the overall outfield position factor: the fast guys are more likely to play centrefield now and in the future and therefore are expected to have more defensive value.

Speed scores can help with infielders too. A middle infielder has to be agile, and while speed and agility are not the same thing, one may note that most good defensive middle infielders stole lots of bases when they were in the minors. Even at first and third base, a higher speed score might indicate quicker reactions.

Speed Scores for Minor Leaguers | 21 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.
Gerry - Thursday, November 27 2003 @ 11:51 AM EST (#84909) #
Interesting work Robert.

I am wondering about a few players not on your list. I note you did not go down to the Auburn level. Aaron Hill split his time between Auburn and Dunedin. How did he score, either combined or at Dunedin? It would be interesting to compare him to Adams.

If you could look at a couple of players from the short season teams it would be interesting to see scores for Vito, Tingler and Roberts.

I would have guessed Rios's score would have been higher.
_S.K. - Thursday, November 27 2003 @ 11:55 AM EST (#84910) #
I'm surprised at Rios' low score as well. Robert, do these scores tend to have predictive value in your experience? In other words, will Rios probably stay at this level or fall as he grows older, or is it merely a measure of PAST performance (like Win Shares defence) which doesn't tend to accurately predict the future?
Interesting work during a slow baseball week.
Mike Green - Thursday, November 27 2003 @ 12:06 PM EST (#84911) #
Very interesting work, Robert. In the particular case of the speed of Rios vs. Werth, I have a quibble. Rios hit 11 triples vs. Werth's 1 (albeit in 1/2 the PAs). On the other hand, Werth went 11-1 as a basestealer, whereas Rios went 11-3. Werth hit into 7 DPs, whereas Rios hit into 22 (in twice the PAs).

Did Rios hit into 22 DPs because he's as slow as Jim Rice? Somehow I doubt it. My theory for the reasons that Rios hit into 22 DPs, significantly more per PA than Werth:

1. he's a right-handed hitter who hits the ball very hard, and doesn't strike out much, and may not get a quick first step out of the box, and
2. the New Haven ballclub didn't very little stealing and hitting and running, so the DP totals are quite high throughout the lineup.

I am quite sure that Werth is not faster than Rios, bearing in mind the above and the subjective reports. It may very well be, bearing in mind Werth's stolen base success rate and Rios' DP rate, that Werth has the faster first step. I'd bet that if you put a stopwatch on them for the 40, that Rios would win.
_whizland2000 - Thursday, November 27 2003 @ 12:27 PM EST (#84912) #
Is Rios the # 1 ranked outfielder in the minors according josh boyd's baseball america rankings and are there any other jays prospects on that list.
_Jabonoso - Thursday, November 27 2003 @ 12:54 PM EST (#84913) #
Thanks for your work Robert!
I'm surprised too about how fast is Werth and how slow is Gross. Is this measurement league related? As if the lower A leagues are filled with speedsters and in higher levels they are slowed down.
I wonder also if an organization that preaches station by station base on balls and homeruns will have its players with lower rates?
On the Werth shopping thing, I have not understood the following: JP do not like tall catchers, I do not think this is a universal approach. i would guess that if you give Werth a fair number of games as catcher many ( say half a dozen ) GMs will bite. Speedy slugger than can play OF and catch= Hot commodity.
With regard to Rios, i do not get it how can you hit 11 triples and have average speed. What rate would VWells get?
robertdudek - Thursday, November 27 2003 @ 01:10 PM EST (#84914) #
I agree that Werth is probably not faster than Rios. Rios' DP was quite high, even given his RBI total, and he did not attempt many steals, even compared to his teammates. Since SB and CS carries a heavy weight in speed score, Rios speed ratio was not going to be great.

Here are the stolen base component ratings for some New Haven players: Thompson 0.48, Godwin 0.40, Adams 0.20, Rios 0.15. Rios was 4th on the team. Perhaps it was simply a question of team orders that limited Rios' attempts.

With respect to GIDPs, righthanders do ground into more of them (but they will in the majors too, hence decreasing their value relative to lefthanded batters). It's very possible that a disproportionate number of Rios' outs were groundballs, which would unfairly lower his score. In a future version, I will try to estimate flyball/groundball ratio and adjust for handedness.
robertdudek - Thursday, November 27 2003 @ 01:15 PM EST (#84915) #

It's hard to say how predictive this measure is. It estimates speed as evidenced by the batting record in a given year. Players slow down after age 23 or 24, so most young players who come into the league are faster than average. The rating system says that Rios isn't particularly fast for a AA player, but we must remember that most AA players are near their speed peak.
Coach - Thursday, November 27 2003 @ 01:26 PM EST (#84916) #
Interesting stuff, Robert, as always. Perhaps I'm wrong and the difference is all quickness, but Cash seems quite a bit faster than Huckaby to me. If Griffin is slower than both of them, you're right -- he's a 1B/DH.
robertdudek - Thursday, November 27 2003 @ 01:30 PM EST (#84917) #
Keeping in mind that the with fewer PA, all measures become less reliable, here are those players with 100+ PA who were not previously listed:

Pulaski(R+): Acey 1.05, Vancamper 0.99, Braun 0.98, Hetherington 0.96, Ponce 0.95, Tingler 0.95, Reiman 0.92, Sena 0.91, Diaz 0.90, Y Rodriguez 0.86, Esposito 0.79, Thomas 0.75, Wolfe 0.74

Auburn (A-): Peralta 1.45, Roberts 1.16, Porfirio 1,06, Galloway 1.04, Snavely 1.00, Richmond 0.98, Cota 0.91, A Hill 0.81, Mangioni 0.76, Kratz 0.71, Chiaravalloti 0.64

Charleston (A): Negron 1.23, Waugh 1.03, Rico 1.01, Arnold 0.93

Dunedin (A+): Carter 1.40, A. Hill 1.14, Keene 0.98, Delfino 0.86, Perry 0.85, Yepez 0.81, Umbria 0.53

New Haven (AA): Godwin 1.39, Sanders 0.99, Sequea 0.96, Keene 0.91

Syracuse (AAA): Thompson 1.44, R Johnson 1.11, Keene 1.06
robertdudek - Thursday, November 27 2003 @ 01:42 PM EST (#84918) #

You make some good points. Rios also hit lots of doubles, and he does score high in the triples category, but not exceptionally high. Perhaps I should increase the weight of that category.

The measure is league adjsuted, and it is quite possible that the average A ball and rookie league player is faster than the average AA or AAA player in an absolute terms. The numbers say Rios is a bit faster than the average Eastern League non-pitcher. For reasons outlined in an earlier post, they are probably wrong about him.

It is also very possible that there is an organisation-wide policy to tone down base stealing (as there is at the big club). If this is true, we can increase the ratings by 5-10% for every player in the organisation.

I agree with you completely about Werth. If the Jays don't plan to make him a regular in the next 2 years, it's probably a good idea to put him in AAA and let him catch once or twice a week.
robertdudek - Thursday, November 27 2003 @ 01:44 PM EST (#84919) #
I think Cash is definitely more agile than Huckaby, but he's built a bit like a fireplug and I don't know if he's all that fast on the basepaths.
Mike Green - Thursday, November 27 2003 @ 01:46 PM EST (#84920) #
Robert, here's a suggestion. I've always found that the number of triples/at bat, compared with the team triples/ab, is a good indicator of speed (at least in the 40 yard sense).

I had a quick look at the consensus best defensive centerfielders- Andruw Jones, Jim Edmonds, Torii Hunter, Mike Cameron, Carlos Beltran and Vernon Wells. The only one who would have a very good speed score is Beltran. Jones' speed score 4 years ago would have been very good, but then his defence 4 years ago was pretty amazing. There does seem to be a correlation between raw speed and ability to play defensive centerfield, but it isn't as tight as one might have guessed.
_Jonny German - Thursday, November 27 2003 @ 01:54 PM EST (#84921) #
Another great study, Robert. I'd be interested in seeing the same numbers for some major leaguers, if you have them, to provide reference points.
_tangotiger - Thursday, November 27 2003 @ 01:58 PM EST (#84922) #
I agree with the 3b/(2b+3b). If you look at the age progression of this, compared to SB/(.8*1b+.6*bb), they go hand-in-hand at virtually the same rate (over a large enough sample).
_Oggman - Friday, November 28 2003 @ 09:47 AM EST (#84923) #
Nice study, with respect to triples. Some players just don't hit them at the rate they "should", Rickey Henderson comes to mind. Even though he had (has?) trememdous basestealling instincts and you can attibute his SBs mostly to that, you'd be hard pressed to call him slow. But the guy just didn't hit triples.

I'm afraid when it comes to speed, especially defencive speed, I think most metrics just won't cut it. For example, I bet Cal Ripken would fare very low in Speed Ratio, but he had used very good positioning to make himself a GG shortstop. Bordick, to me at least, was very similar. Also both of them had great first steps. I wonder how well Jeter would perform in speed ratio?

But over all, a very good study. I like anything where I can see something about the Kids in the minors.
_Oggman - Friday, November 28 2003 @ 09:49 AM EST (#84924) #
Hmm...the one time I don't use preview.

The first line should read: "Nice study. With respect to triples, some players..."
robertdudek - Friday, November 28 2003 @ 10:04 AM EST (#84925) #

I don't like triples per AB, because it discriminates against a few different types of players:

1) players who hit lots of homeruns (they either hit the ball out of the park or so hard that a triple is difficult)

2) Players who strikeout a lot.

3) Singles hitters who rarely hit extra-base hits
Mike Green - Friday, November 28 2003 @ 04:38 PM EST (#84926) #
You have a point, Robert, but 3b/2b+3b has some problems too. Let's divide the doubles into the following categories:

1. no contest doubles (ground rule doubles, balls down the left-field line that go to the wall except in Fenway perhaps)
2. doubles that could be triples, and
3. singles that could be doubles (the ball down either line which the outfielder cuts off before it gets to the wall)

The number of doubles a player has tends to decrease as he ages, presumably because doubles of type 3 exceed doubles of type 2. So what can easily happen, is that the ratio will be of small numbers, subject to very volatile fluctuation. For example in 1974, at age 24 Mike Schmidt hit 7 triples and 28 doubles. In 1983, at age 33, he hit 4 triples and 16 doubles. The ratio is the same, but it was quite plain from other measures that Schmidt had lost most of his speed by then.

I do think you need to look at triples/AB, in some form, perhaps in combination with 3b/2b+3b.

Keep up the good work.
_Matthew Elmslie - Friday, November 28 2003 @ 04:43 PM EST (#84927) #
Triples rate also 'discriminates' against righthanded hitters, doesn't it? Or do I have that backwards? Anyway, it's probably not that hard to adjust for.
robertdudek - Friday, November 28 2003 @ 06:22 PM EST (#84928) #
I adjusted my speed score ratings, with the following constituting major changes:

1) I increased the weight of the triples category to almost match the SBA category - the two together constitute 80% of the speed score.

2) I increased the benefit to actually hitting more triples, so 7 triples/21 doubles had a higher score than 3 triples/9 doubles, which was greater than 1 triple/3 doubles. Hitting 0 triples/15 doubles was worse than 0 triples/5 doubles etc.

3) I regressed the GIDP towards the mean and made an adjustment for left/right/switch hitters.

Rios' speed ratio goes up to about 1.27 (in 2002, he had about a 1.41).
Mike Green - Friday, November 28 2003 @ 09:55 PM EST (#84929) #
Sounds good, Robert.
Speed Scores for Minor Leaguers | 21 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.