A Deeper Look at Baserunning, Part 2: Players

by Adam Peterson on Mar 21, 2009 9:48 PM EDT in By the Numbers Comment 10 comments

Last week, I began a three part series focusing on baserunning:

Team Baserunning (here)
Individual Baserunning
Comparison to Baseball Prospectus Baserunning Statistics

Part 2 focuses on individual baserunning performances during the 2008 season. I have posted the full player spreadsheet on Google Docs here.

Same as I did with the team totals, I calculated total expected runs for five areas of baserunning, SB/CS, extra bases, staying on the current base, and outs on the base paths (OOBP).

Based on my calculations, here are my MLB baserunning awards for the 2008 season:

The "Jackrabbit Slim Baserunner of the Year" Award: Willy Taveras (+13.59 ER_TOTAL) (runner up: Ichiro Suzuki: +12.73)
The "Time Him on a Sundial Basewalker of the Year" Award: Dioner Navarro (-8.98 ER_TOTAL) (runner up: Magglio Ordonez: -8.75)
The "Hamburgler Base Stealer of the Year" Award: Willy Taveras (+7.35 ER_ALLSB) (runner up: Jimmy Rollins +4.13)
The "Red Light" Award: Nick Markakis (-3.98 ER_ALLSB) (runner up: Hunter Pence: -3.68)
The "Silent But Deadly Baserunner of the Year" Award: Shane Victorino (+9.76 ER_NONSB) (runner up: Chone Figgins: +9.37)
The "Anti-Paul Molitor" Award: Prince Fielder (-7.88 ER_NONSB) (runner up: Bengie Molina: -7.59)
The "Perfect Baserunning" Award (most extra bases plus steals without being thrown out a single time): Jason Bay (49) (runner up: Mark Teixeira: 36).
The "Thanks for the Assist" Award (most times thrown out stealing or trying to advance an extra base): Jose Reyes (30) (runner up: B.J. Upton: 29)
The "Pit Bull" Most Aggressive Baserunner Award (at least 40 extra bases): Corey Patterson (59.5%) (runner up: Maicer Izturis: 58.2%).

Additional analysis and details on the Twins baserunners follows after the jump.

Minnesota Twins Base Stealing

As a team, the Twins ranked #24 in MLB stealing bases, with -9.94 expected runs combined due to stolen bases and caught stealing. I won't go into detail about base stealing strategy here, but I find it striking that no single Twins baserunner added more than one run due to stealing bases:

Player	ER_ALLSB	SB	ER_SB	CS	ER_CS	SB%
Matt Tolbert	+0.64	6	+0.91	1	-0.28	86%
Michael Cuddyer	+0.10	5	+0.58	1	-0.48	83%
Delmon Young	-0.68	14	+1.84	5	-2.52	74%
Joe Mauer	-0.76	1	+0.13	2	-0.88	33%
Alexi Casilla	-0.98	6	+0.63	3	-1.61	67%
Denard Span	-0.99	17	+2.84	7	-3.83	71%
Nick Punto	-1.07	14	+1.95	6	-3.02	70%
Brian Buscher	-1.13	0	0.00	2	-1.13	0%
Carlos Gomez	-2.08	33	+5.29	12	-7.37	73%

Key:

ER_SB: Total expected runs gained due to successful stolen bases
ER_CS: Total expected runs lots due to times caught stealing
ER_ALLSB: Sum ot ER_SB and ER_CS. Total expected runs due to stolen base attempts
SB, CS, SB%: Total stolen bases, times caught stealing and success percentage

Not surprisingly, the only Twins with overall positive expected runs had success rates above 80%. Even though Carlos Gomez stole 33 bases, his overall contribution was a little over -2 runs. Being caught stealing so many times (tied for 8th most in the majors) hurt the team more than the stolen bases helped, at least from a direct expected runs standpoint. As I noted in Part 1, I have not investigated second order effects of stolen bases, such as more fastballs, etc.

Minnesota Twins Extra Bases and Aggressiveness

We've looked at stolen bases, and the Twins were below average in 2008. But as I've shown elsewhere (here), the Twins were the best in the majors at the "little things", including baserunning. What gives? Aggressiveness on the base paths, that's what gives. If you remember from Part 1, the Twins were second best in the majors at "non-SB" baserunning, creating an additional +13.57 runs on the basepaths compared to the average 2008 team. The Twins were also the most aggressive baserunning team in the majors, attempting to take an extra base 44.7% of the time (league average 40.0%) and adding an additional +52.83 extra bases (compared to the average team) over the course of the season.

Player	ER_NONSB	EB_EXP	EB	EB_ADD	AGG%	OOBP
Joe Mauer	+5.43	83.32	88	+4.68	45.0%	3
Matt Tolbert	+4.29	14.71	22	+7.29	66.7%	0
Alexi Casilla	+3.01	43.77	54	+10.23	47.6%	5
Carlos Gomez	+2.59	54.22	64	+9.78	52.9%	10
Denard Span	+2.20	45.87	57	+11.13	46.7%	6
Randy Ruiz	+1.52	10.14	13	+2.86	56.0%	1
Adam Everett	+1.22	11.04	12	+0.96	43.3%	1
Mike Lamb	+0.90	16.47	17	+0.53	37.5%	1
Justin Morneau	+0.44	64.71	69	+4.29	47.4%	3
Nick Punto	+0.11	47.62	49	+1.38	46.3%	7
Brendan Harris	-0.01	41.39	39	-2.39	36.4%	4
Brian Buscher	-0.08	17.19	21	+3.81	41.8%	2
Mike Redmond	-0.89	9.63	7	-2.63	21.2%	0
Michael Cuddyer	-0.96	21.99	22	+0.01	39.7%	3
Jason Kubel	-1.04	47.03	46	-1.03	39.5%	3
Delmon Young	-1.51	59.39	65	+5.61	45.5%	10
Craig Monroe	-1.70	12.39	10	-2.99	35.6%	3

Key:

ER_NONSB: Expected runs outside of stolen base attempts.
EB_EXP: Expected extra bases taken by a league average baserunner in the situations faced by this baserunner.
EB: Total extra bases taken.
EB_ADD: Extra bases added. Total extra bases taken minus expected extra bases.
AGG%: Aggressiveness. Percentage of time that the baserunner attempts to take an extra base, rather than staying on his current base.
OOBP: Total outs made on the basepaths, not including caught stealing.

Joe Mauer grades out as the best baserunner due to a combination of aggressiveness (45.0%) and making few outs (3). In a relatively small sample size, Tolbert's aggressiveness was off the charts (66.7%) without making a single out. Among regulars, Carlos Gomez was easily the most aggressive baserunner (52.9%, nearly 10 extra bases added), but his 10 outs knocked down the total expected runs a bit. On the other side, Mike Redmond was the least aggressive (21.2%), and Delmon Young was Gomez without as many extra bases.

Joe Mauer vs. Jose Reyes

Who was the better baserunner in 2008? Jose Reyes and his 54 stolen bases, or Joe Mauer? In terms of total expected runs, the winner was Mauer.

Player	ER_TOTAL	SB	CS	ER_ALLSB	EB	EB_ADD	AGG	OOBP	ER_NONSB
Joe Mauer	+4.67	1	2	-0.76	88	+4.68	45.0%	3	+5.43
Jose Reyes	+4.62	54	19	+1.38	93	+11.82	51.5%	11	+3.24

One would expect that Reyes' 54 steals and much more aggressive baserunning would have put him ahead of Mauer. Wrong. Because of Reyes' 19 times caught stealing, he only had an advantage of a shade over two runs stealing bases compared to Mauer. In non-SB baserunning, Reyes' additional 8 outs on the basepaths more than cancel out aggressiveness and 7+ more extra bases added. In the end, this shows that aggressive, but smart baserunning can be more effective to a team than running wild. Not exactly news, but interesting that the two came out so close in overall expected runs.

Albert Pujols

I was just reading Joe Posnanski's (excellent writer, by the way) article in Sports Illustrated a week ago. About halfway through, right after a paragraph that mentions Pujols being the best rated defensive first baseman in John Dewan's +/- fielding ratings each of the past three years, Posnanski says:

And it's more than his offense and defense. He runs the bases aggressively and successfully, especially for a man with below-average speed.

Since we have all the data at our fingertips, let's look at that statement in a bit more detail.

	ER_SB	ER_NONSB	ER_TOTAL	EB	STAY	OOBP	AGG	EXP_EB	EB_ADD
Albert Pujols	-0.07	-0.12	-0.19	63	91	8	43.8%	56.88	6.12

Was Albert Pujols an aggressive runner in 2008? I rate his aggressiveness at 43.8%, in the 69th percentile among all players, and a bit above the overall league average of 40.0%. Yes, the data shows that Pujols is aggressive, but nearly in the Carlos Gomez 52.9% territory. I'll give Posnanski this one. Not as aggressive as Mauer, but above average.

Was Albert Pujols a successful baserunner in 2008? Pujols was thrown out on the base paths 8 times, tied with 23 other players for 23rd in the league behind B.J. Upton's league leading 12 OOBP. He was thrown out 4.94% of all extra base opportunities, above the league average of 3.82%. Stealing bases, Pujols had 6 SB out of 10 attempts, below the league average of 69.5%. Unfortunately, based on this data, I have to disagree with Posnanski, at least in 2008.

Next Steps

Part 3 will compare these baserunning statistics to other methods, including Baseball Prospectus. To the extent that I understand the other methods, I will compare not only the methods, but also the resulting numbers. Where do we agree? Where do we disagree?

1 recs | Comment 10 comments

Story-email Email Printer Print

Comments

Just thinking out loud here....

how is it that Joe Mauer has more added runs than he does added bases? Is it because he got thrown out less than would be expected given his attempts at advancement?

If it is…aren’t you double counting somewhere in here? Giving him credit both for the extra base and the non-out? I think that’s a problem somehow…still thinking it through.

by Eric in Madison on Mar 21, 2009 10:08 PM EDT reply actions 0 recs

I'm not really giving someone credit

for both the extra base and the non-out. It’s really just a spectrum, with three possibilities for each extra base opportunity. The context differs for each specific situation, but considering league averages, if a runner is thrown out: -0.67 runs. If a runner stays: -0.066 runs. If a runner takes the extra base: +0.181 runs.

By avoiding the outs, a runner avoids the larger losses of runs in much the same way that a pitcher avoids larger numbers of runs allowed by avoiding home runs. By the same method, I would not be double counting runs by allocating for the lack of HR allowed and for the actual result, single, strikeout, whatever.

by Adam Peterson on Mar 22, 2009 7:24 AM EDT up reply actions 0 recs

Let me rephrase the above question

It seems that you are giving Mauer credit for being fairly aggressive and not making many outs on the bases. But isn’t that already reflected in the extra-bases vs, expected extra-bases measurement?

by Eric in Madison on Mar 21, 2009 10:59 PM EDT reply actions 0 recs

Runs consider not only the extra bases

but also the outs made and the context. Mauer got thrown out much less often than league average over his total number of opportunities. 3 outs in 202 opportunities, or 1.49. League average was 3.82, or 7.7 outs over Mauer’s opps. In other words, Mauer avoided 4.7 outs on the basepaths. Back of the envelope, at a league average of about -0.67 expected runs per OOBP (see Part 1), that corresponds to about 3.15 of Mauer’s total non-SB expected runs. The rest is largely due to the aggressiveness and extra bases.

by Adam Peterson on Mar 22, 2009 7:18 AM EDT up reply actions 0 recs

Expected extra bases

One more thing, my calculation of “expected” extra bases does not consider outs made. It’s simply another way of looking at the data. If, for a given extra base situation, the league advances the extra base 40% of the time, then for that situation I assign +0.4 “expected” extra bases for a specific runner when he faces that situation. Outs are not already reflected in the measurement.

by Adam Peterson on Mar 22, 2009 7:28 AM EDT up reply actions 0 recs

Data Request

First of all, kudos (again) on your work here – great stuff, as usual.

One data point that I’d find interesting – what is the average success rate for taking extra bases? That would provide a bit better idea of how many runs players are costing above or below average when they get thrown out on the basepaths.

Other questions:
Just making sure, but can we assume the extra base percentages take into account the area the ball was hit to (and maybe hit type), since this is based on Gameday data?

Is the non-SB data the same thing as the baserunning portion of your “little things” analysis, or is there something new I’m missing? It’s still useful in this article, since it’s about baserunning as a whole; just checking whether it’s new data or not.

"There are only two things that are infinite, the universe and human stupidity, and I'm not sure about the former." - Albert Einstein

by BeefMaster on Mar 23, 2009 8:24 AM EDT reply actions 0 recs

Average success rate

Considering only the league total extra bases and outs on the base paths, it appears that 9.6% (1637 or 15502) of all extra base attempts (EB+OOBP) result in an out. I think that’s your average success rate.

Answer #1: Yes. This whole analysis is based on my baseline analysis of baserunning. I analyzed teh MLB Gameday data and determined the total number of advance, stay and outs for each baserunning situation. Situations are defined based on hit type (GB, LD, FB, PF, BU), result (hit or out), location (position, will be fielding zone in a future iteration). The baseline data is stored in an XML file as a list of these:

This entry means that on a fly ball out to LF, a runner on first advanced to second base 42 times, made an out 27 times, and stayed on first base 1074 times.

Answer #2: In my original “little things” analysis, I only used total baserunning expected runs, noted as “ER_TOTAL” in this story. When I decided to look at baserunning in more detail, I broke out expected runs based on stolen bases, non-SB, etc as separate data.

by Adam Peterson on Mar 23, 2009 2:02 PM EDT up reply actions 0 recs

Cool

Both “Answers” were pretty much what I expected (although I guess I was thinking the “little things” baserunning was the non-SB data, not the total); thanks for the clarifications. Also, it appears that your XML got ripped out (I’m guessing unrecognized stuff in greater-than-less-than signs gets stripped out as bad HTML tags), but I got the gist of your comment anyway.

The 90% success rate is lower than I would have expected – I don’t generally think of extra base attempts as being foiled that often, despite Michael Cuddyer’s best efforts. Are pickoffs following line outs included in the outs on base? I guess that would inflate the “out” numbers a bit.

Were there many team-specific trends in the non-SB data? I recall some discussion about trying to evaluate third base coaches, although I realize that at the moment there isn’t really much way to separate the players’ effects from the coaches’. If you manage to get the historical data working, and the information shows some consistency, it would then be possible to study effects of coaching by analyzing players who changed teams and teams that changed coaches (the latter should be more effective due to sample size, I would think).

Do force outs factor into this analysis at all? I doubt the sample sizes are big enough for many players to have a repeatable pattern for this; it’s more just curiosity on my part.

One more methodology question – were the expected rates at which a player was likely to take the extra base adjusted to take the number of outs into effect? Coaches are generally more likely to send a runner from third with two outs than with zero or one out, and I’m wondering if a runner in that situation gets less credit for taking the extra base than the same runner scoring with no outs.

"There are only two things that are infinite, the universe and human stupidity, and I'm not sure about the former." - Albert Einstein

by BeefMaster on Mar 23, 2009 2:32 PM EDT up reply actions 0 recs

Additional stuff

Guess I should have previewed the comment…wasn’t expecting the XML to be stripped out.

90% was lower than I expected as well. Then again, each situation is different. For the one I cited, 27 outs and only 42 extra bases lowers the overall success rate. By comparison, for a runner on third and a fly ball out to LF, the runner advanced 416 times versus 17 outs.

Yes, pickoffs after line drives are included as outs on the base paths.

I haven’t analyzed team-specific trends by situation in the non-SB data. The framework is available to break down further to analyze aggressiveness of base coaches in specific situations. Can’t easily separate the coach from the player though.

No. Force outs are not included. For a force out, the “default” end state that ends up allocated to the batter is the runner being out, batter reaches first, and other runners advance if forced. I haven’t looked at individual runners being forced out less often.

No. Baserunning situation and expected out rates do not consider the number of outs. You’re right, a baserunner would be penalized for staying on third with zero outs versus advancing with two outs. I guess this makes sense to add to each situation, my only concern is further reducing the sample size.

by Adam Peterson on Mar 23, 2009 3:22 PM EDT up reply actions 0 recs

I can't come up with a good subject line for this post

Yes, pickoffs after line drives are included as outs on the base paths.

I was going to comment that this seems to penalize the runner quite a bit for something that’s usually not all that much his fault, but on the other hand, since your analysis is weighted between batter and runner, this is probably already taken into account as much as it needs to be (yet another thing I like about this analysis).

I haven’t analyzed team-specific trends by situation in the non-SB data. The framework is available to break down further to analyze aggressiveness of base coaches in specific situations. Can’t easily separate the coach from the player though.

Yeah… separating the coach from the player is something that can’t be done without multi-year data. You could try it for players who switched teams midyear, but that’s going to have some serious sample-size issues.

You’re right, a baserunner would be penalized for staying on third with zero outs versus advancing with two outs. I guess this makes sense to add to each situation, my only concern is further reducing the sample size.

I’m not sure whether reducing the sample size is worth the potential gain in accuracy. It will likely help when multiple years are analyzed, as that should offset some of the sample size issues.

"There are only two things that are infinite, the universe and human stupidity, and I'm not sure about the former." - Albert Einstein

by BeefMaster on Mar 23, 2009 4:10 PM EDT up reply actions 0 recs