Runs/Bases Remaining

[Jackaroo Dave]

What are your thoughts about measuring baserunning effectiveness by dividing the number of runs scored by the number of bases the runner has to go to reach home? In other words, (Runs - HR)/(3B + 2 * 2B + 3 * (1B + BB + HB))?

Aside from the usual problems in evaluating runs--they are a team stat, very context dependent, etc.--do you see any specific problems of bias or other foolishness that a sufficiently large sample size wouldn't take care of?

Of course with play by play data you can do much better, but it's not always available. As a quick and dirty stat does it measure something of value?

I got the idea from csh's observation that Rickey Henderson had a very low proportion of doubles and triples, which I attributed to an approach to baserunning. I was interested to see whether his advancement after reaching base was similar or compensatory and this seemed like one way to get at it.

Thanks for any input.

[leewileyfan]

I've been working, on and off, on a spreadsheet to equate base running [as an isolated skill set] to actual run creation above average. My approach, in steps has been:

1. Get individual player's batting line and THROW OUT doubles, triples and HR's.

2. DO NOT consider Stolen Base or Caught Stealing numbers at all.

3. Add Singles, BB's. and HBP and get the sum of those three components.

4. Since, in the history of MLB, runs scored as a percentage of Total Bases has been around .325 - .330, AND since we've taken all extra base hits off the table, run expectancy should be at or less than .3 of the bases-sum of 1B, BB and HBP. The discount is compounded by the fact that OTHER PLAYER IN LINEUP HR's, Triples and Doubles can also be discounted in isolating base runner extra base taking as a run creator.

5. A player formula MIGHT look like: (1B+BB+HBP)*.3 = Base Runner Runs. To be of value, for comparisons to a norm, you might take LG average [1B+BB+HBP*.3] for the norm.

[leewileyfan]

I ran two examples to show the way the numbers shake out for two disparate players, one famous for base-stealing, the other with NO perceptible base stealing skills - but allegedly fast and smart in taking extra bases on batted balls [his and those of team mates who followed in the batting order]. The first is Rickey Henderson. The second is Rogers Hornsby.

1. Rickey Henderson: Singles [2182] + BB [2190] + HBP [98] = 4,470 bases.
4,470 bases * .3 [expected scoring rate, dependent on batters who follow] = 1,341 Expected Runs.
Actual Runs Scored [2,295] - Henderson HR [297] = 1,998 Net Runs Scored, per model.
1,998 - 1,341 = 657 Runs Scored Above Expectation.

The spreadsheet, which uses stolen base data and 3B data [after the fact, incidental] projects that Rickey Henderson Created Base Runner Runs above average amounting to 379.2 Runs. The working conclusion is that of 657 total runs above average expectation that Henderson accomplished, 379.2 were PURELY from Base Runner Skills.

2. Rogers Hornsby: Singles [1919] + BB [1038] + HBP [48] = 3,005 bases.
3,005 bases * .3 = 902 expected runs scored.
Actual Runs [1579] - HR [301] = 1,278 net runs scored.
1,278-902 = 376 runs scored above expectation.

The spreadsheet, which has "0" net stolen bases for Hornsby [thus attributing ALL bases to speed and smarts on the base paths] projects 247.3 base running runs to Hornsby.

Working conclusion: Hornsby scored 376 runs above expectation, 247.3 of them purely from taking the extra base on batted balls in play.

[Dude Paskert]

Interesting stuff, and I like the thought process.
A good test would be to run it for Cecil Fielder and see how close you get to zero base runner skill runs!

[leewileyfan]

I just ran the numbers on Cecil Fielder. Following the procedure I outlined earlier, Fielder would have been expected to score 425 runs based on his stripped-down base running opportunities [stealing and/or taking extra bases on batted balls:

Singles: 787
BB: 693:
HBP: 43

Bases: 1523

Applying a scoring expectation rate @ .3 we would have 457 runs scored, as a basic expectation. NET of HR's, Fielder scored 744 Runs - 319 HRs = 425 runs scored from base runner extra base opportunities.

The numbers suggest that Fielder COST his teams 32 runs due to sluggishness on the base paths.

I am pleased with the concept; but I am becoming increasingly convinced that the run expectation factor should be closer to .27 or .28 than .3. If we used .273 [3 for 11] our run expectation for Fielder would drop to 1523*.273 = 416. With Fielder scoring 425, that nets a slight positive.

However, within the numbers, there is a residual implication that the two basic base-running components [steals and taking extra bases] must get separate contribution scores before drawing the composite:

1. A runner like Fielder may well have a negative number in the Base Stealing portion, actually being caught more often than he succeeded in a tiny number of attempts.

2. In the taking extra bases components, a sluggish runner seems often to be pushed beyond expectation by faster teammates running bases behind them. This is reflected in the spreadsheet by a persistent residual level of +0 to +4.9 per season for players whose base running can be estimated as neutral.

Fielder, in conclusion, was a less-than-average base runner, costing runs on the bases, except for those residual cases where he was pushed forward from behind OR batted balls challenged fielders for those extra seconds that allowed him to advance a "extra" base.

There's no real disparity here. Bottom line is that some players with extended careers MIGHT end up with a modest positive number for base running "runs." I am inclined to let it lie at that, kind of figuring "If it ain't broke - don't fix it!"

Suggestions welcome.

[Dude Paskert]

This is cool stuff!
One baserunner who used to have a rep for being one of the worst ever was Wade Boggs...many analysts said he would never take any kind of chance on the bases, and I believed it from watching him. There were a few comments about his run totals being low for a guy who was on base SO much (with a good number of doubles) and was usually backed by a powerful offense. To be fair, he was probably getting erased on DPs a lot by guys like Rice and Armas, although he likely didn't do much to help avoid the DPs. Boggs wasn't THAT slow, he just tried to make people think he was...Fielder couldn't really help being a bad baserunner, Wade probably could have avoided it with a different attitude.
Looking at '87...200 hits, 105 BB, .461 OBP, even 24HRs...and he scored 108 runs total. He did a lot better the next two years, which I think is when the Bosox finally started batting him leadoff, but the totals still weren't incredible.

[leewileyfan]

I revisited Fielder; and the residual element that seemed to impart base running runs to mediocre runners was nagging at me.

I looked at the model player I had created to ensure a basis for comparison that was "neutral;" and that model also had residual numbers.
I then decided to consider that model equivalent to neutrality [or 0]. It turns out that .0289 base running "runs" default to a player, any player, over 162 games of play. That's understandable, since it equates to one baserunner "run" from taking the extra base every 36 games.

Making .0289 equal to neutrality, Fielder actually comes up at -22 runs over his career, which is very close to the number forecast in the original model. I believe this is a good working premise.

I'll check ou Boggs later.

[leewileyfan]

Checking out the numbers on Wade Boggs, I figured he'd be a good sample for comparison with other players, one seen as a good [if not outstanding base runner] and the other a sluggish-prro base runner.

I whittled all the contributing data inputs into three final products:

Games
Opportunities to Take Extra Bases [distilled from hits, distilled and excluding 2B, 3B, HR - explained in earlier post above]
Opportunities Taken [by model estimate; based on suggested speed and aggressiveness factors]
Successful Bases Taken Converted to Runs [dependent on performance of hitters following in the batting order.

Boggs in the basis of comparison here. Fielder is the "0" [or lower] base product assuming NO to miniscule base running contribution ["noise"].
Rod Carew, selected for his very similar "Games" comps with Boggs, is the productive base runner exemplar.

Player..........................Games............. Opportunities..............Bases Taken.............Runs Value, Bases Taken

Boggs..........................2,440.............. ..1,062........................225................ ...........71

Carew..........................2,469.............. ..1,068........................348................ ..........110

Fielder..........................1,490............ ......370..........................34............. ..............11

There appears to be a "runs" default level that will accrue base running runs for the worst of base runners, a "noise" level in which batted ball and fielding dynamics, without charged errors or extra base hits, will move the sluggish runner a base further than could be anticipated. If that reality is accepted, the improved on-base status of the player will ultimately enhance his scoring opportunities into actual runs scored.

[Jackaroo Dave]

What are your thoughts about measuring baserunning effectiveness by dividing the number of runs scored by the number of bases the runner has to go to reach home? In other words, (Runs - HR)/(3B + 2 * 2B + 3 * (1B + BB + HB))? . . .

Thanks for any input.

My own second thoughts are that this is completely misconceived, but I'm glad I started the thread.

[Jackaroo Dave]

Checking out the numbers on Wade Boggs, I figured he'd be a good sample for comparison with other players, one seen as a good [if not outstanding base runner] and the other a sluggish-prro baserunner. . . .

Maybe not everyone knows what I just found out, that baseball-reference will give you the extra base taken percentages for players back to around 1950, I believe.

Carew shows up at 58% (compare Henderson @ 55, Morgan @ 51), Fielder at 26% (Vaughn 29, Ortiz 27), and Boggs 37%.

I also checked for CS > SB and found about 1600 players, but this dropped to 207 using a minimum of 20 CS.

don't know if it's feasible, leewileyfan, but what if you brought your figures into line with BBref's for the cases where there is play by play data? Then you could find comparable figures for cases where there weren't (mutatis mutandem). Just a thought.

[leewileyfan]

Maybe not everyone knows what I just found out, that baseball-reference will give you the extra base taken percentages for players back to around 1950, I believe.

Carew shows up at 58% (compare Henderson @ 55, Morgan @ 51), Fielder at 26% (Vaughn 29, Ortiz 27), and Boggs 37%.

I also checked for CS > SB and found about 1600 players, but this dropped to 207 using a minimum of 20 CS.

don't know if it's feasible, leewileyfan, but what if you brought your figures into line with BBref's for the cases where there is play by play data? Then you could find comparable figures for cases where there weren't (mutatis mutandem). Just a thought.

I often refer to Baseball-Reference for data; but I have no idea how to access the extra base percentages you cite for the players you named. I am aware of their "R-baser" entries [under Player Value] with credit to Sean Smith.

I am also at a loss in understanding the MEANINGS of the percentages given ... what is the denominator and what events are included in the numerator? For example, I have stripped player OPPORTUNITIES [denominator] to what's left after 2B, 3B and HRs have been removed. The reason for this is that 3B and HR are EXPECTED to clear the bases in front of them; and, as I see it, even doubles may be problematical in that many might have been triples with fleeter base runners.

Then too, does the numerator include tag ups and advances [after the catch] moving from 1st to 2nd, 2nd to 3rd, as well and 3B to score?

Sean Smith's R-BASER model on Baseball-Reference stipulates SB, CS, PB, WP and Defensive Indifference. with extra base advances on batted balls not being mentioned.

I have no problem exploring this; but without access to the source [how to get it] or knowing how the source defines its terms, I have no place to start.

If you provide the HOW TO to the link. I'll be happy to look at it.

It is interesting stuff. My main interest is all the history [1901 to the present] - my particular emphasis of defense [another thread here]; but the added values of base running is intriguing. I'd love to have a working model, with some credibility, all the way back through Hornsby, Cobb, Lajoie and Collins, which is why I was so glad to get involved in the exercise. It's good stuff.

[Jackaroo Dave]

My bad, leewileyfan.

On a player page, at the top of the first table of batting statistics, there's a blue "More Stats" tab. Clicking that gets you lots of interesting stuff. To get to extra baserunnung, scroll down almost to the end of the "More Stats" section, I think just before you get to the yearly leader section. There is a baserunning and miscellaneous stats section.

It gives the opportunities for taking an extra base on a single or double, as well as the times an extra base is taken, and the percentage. As you probably know, clicking or hovering over the label at the top of the column gives you a thumbnail explanation. Iirc, the stat does not include any other advances. Hope you find it of use and interest.

[leewileyfan]

OK, Jackaroo Dave ... I've found some time to go over my notes and revisit Baseball-Reference with its Player Page refinements of Bases Taken [under Player Values] and the base runner advance data when you hit the MORE STATS prompt.

I think it's great that there are folks poring over all this data and making details available that were not nearly so only a few years ago. However, even with those specifics, there are some gaps that defy player comps beyond 1950 or so. These may never be fully addressed due to incomplete data for several seasons in several categories.

I don't believe this is a closed door problem for those willing to entertain models of base runner run creation to induce or deduce specifics that are credible from data we already have.

Here's how I went about creating my final model player and how I arrived at AVERAGE, which refuses to adjust itself to ZERO.

1. The Player, Al Soran, has had a 10 year MLB career ... a model of deadly consistency with each year being identical:

AB: 570
BB: 37
Hits: 145
BA .254
1B: 102
2B: 27
3B: 4
HR: 12
St.B 8
C/S: 6

Excuse my slovenly preparation; but I have him never being HBP [which may explain his post career modeling contract as spokesman for Pond's skin care products].

2. All that matters for this exercise is Al's Singles, BBs, St.B and C/S numbers, because I am starting out on the most basic premise that the most meaningful advances for a base runner are those he makes from first base ... either scoring on a double or moving to third on a single. It is, as I see it, far more expected of a runner on 2B to score on a single, provided the single is not an infield hit.

3. The next supposition is based upon WHIP and other data which consistently suggests that a batter will have an on-base opportunity each time he comes to the plate; but that 40% of the time, a teammate will already be on SOME base. This raises the concern of possible cancellation of certain "opportunities." A slower teammate on 2B will stifle St.B opportunities for Al Soran, who is not much inclined to base theft to begin with.

4. Another consideration is the dependency on hitters that follow in the batting order. A 6-4-3 or other DP will erase our base runner and all his chances to excel, with a 2 for the price of 1 efficiency.

5. If Al Soran, over a 10 year career, has compiled 1,450 singles, 370 BBs and swiped 80 bases, having been nailed in 60 attempts, we have a start:

1,450 + 370 = 1,820 gross opportunities in which Al Soran was assuredly on 1B. We subtract 40% for the probability that others [or OTHER, already on base before Al stepped to the plate] will/may impede advance. That leaves Al with 1,092 net base advantage opportunities, with even these co-dependent on the effectiveness of batters in the situations.

6. We can tell from Al's St.B numbers that he is a MINUS in that column. Since we are measuring base running as a skill, we must weigh his successes against his failures [80 stolen bases * .4 = 32]; [60 caught stealing * .6 = -36] = -4 St.B Runs.

I am a bit jammed up with "stuff" right now [like an episode of "The Killing" on AMC]; so I'll continue this later.

[leewileyfan]

OK, picking up where I left off.

Jotting notes while trying to determine a "Speed" factor, I realized it wasn't really going to be a measure of "speed" at all. Base Stealing, only ONE of two components that belong here, is a very specialized skill that [IMHO] has much more to do with "reading and assessing pitcher body language and mechanics; and sensing catcher capabilities in both keeping his pitcher in-the-moment and making fast releases, with sure and speedy throws to target bases. Jackie Robinson ran like a football running back; but he drove pitchers [and catchers] crazy, with head-game feints and dead-stop starts.

Inputs for running INSTINCTS [not speed]:

1. Base Stealing success rate;
2. Three times the square root of the quotient Triples/Hits. I was seeking a "jump" range between .20 and .80. It works.
3. Average the two components.
4. If StB > 65% of attempts, then #3 *1, or #3 * .9.

Al Soran has already logged -4 Runs in the StB column. Next comes the EXTRA Bases Taken model.

Al has compiled 1,820 "bases," the sum of his 10 year Singles and BB. We subtract 40% from that to get NET feasible opportunities. Now there are two factors at work, for or against his base
runner "runs" score:

-subsequent batters putting balls in play [the typical BBBBIP seems to be around .315; but we are interested in OTHER opportunities as well, like tagging up and advancing on fly balls and just maybe catching a fielder napping on something "routine" and making the most of it. I hike the factor from .315 to .35.

-the composite INSTINCT profile devised for the player. Here, for example, are the top players in the profile to date, ranging slightly above .80 to slightly below .80:

Reyes
Crawford
W. Wilson
D. Roberts
Fowler
Granderson
V. Coleman
Victorino
B. Gardner

The final tally for Al Soran and the players named above will follow in Part III.

[leewileyfan]

Just drawing a few conclusions before wrapping up my contribution here:

1. In my quest for a "speed" factor that really wasn't ... it was more an INSTINCT and TIMING and REFLEX matter of studying, assessing and going ... trying to wed "speed" elements too closely with great base running is a mistake. Of the 9 players named in my post above for high INSTINCT scores [really quite a bit based on isolated "speed" elements, only 7 make the top 50 or so listed below. I suggest that this may be due to length of career to date, team batting profiles and on-base situations beyons a runner's control.

However, the are other big picture factors, like the essential ability to get on base. Then, too, one must amass a body of work, which means durability and longevity in a stiflingly competitive sphere in a talent set that favors youth [and prime years].

Just taking ALL factors into a big picture, here are the "best" 50 or so base runners [since 1901], from the standpoint of base runner net runs produced per game:

V. Coleman .116
Henderson .104
Reyes .101
Raines .100
Lofton .092
LeFlore .091
Crawford .091
*[EDIT UPDATE]: INSERT Jacoby Ellsbury .091
E. Collins .090
Cobb .089
J. Morgan .088
W. Wilson .088
Suzuki .087
Bourn .086
D. Roberts .084
Lopes .084
Brock .083
Rollins .083
Br. Gardner .081
Beltran .080
Ho. Wagner .078
Figgins .078
Molitor .078
Pierre .078
Dykstra .077
M. Carey .076
R. Alomar .076
Larkin .074
Ba. Bonds .074
Butler .073
P. Waner .073
O. Smith .073
Podsednick .072
Kinsler .071
Victorino .070
Hornsby .070
Damon .070
Cuyler .070
Campaneris .070
Ashburn .069
Crisp .069
Jeter .069

That's it. Feedback welcome.

*[EDIT] Above; addition.