Stat for Leadoff Hitters

[mlstros]

I've wanted to come up or find a stat that would find how effective a leadoff hitter can be. I'm not too much of a sabermetrician, and I'm not too familiar with stats. But anyways... Obviously, OBP would a major component because the leadoff hitter (and for every other hitter, for that matter)'s objective is to get on base. But I don't think Albert Pujols with a .400 OBP would make a great leadoff hitter. The hitter needs speed also with the big bats behind him. So here is what I came up with:

SBE: Stolen Bases / Stolen Base attempts
Basically stolen base efficiency to determine how successful they are at stealing bases.

SPEED: SBE * Stolen Bases
Uses how successful the hitter is at stealing bases and combines it with the amount of stolen bases they have. So the more stolen bases they have, the higher this will be. But this will only be high if their SBE is high as well.

Speed * OBP
This combines their success at stealing bases, the amount of stolen bases they have achieved, and how often they get on base.

For example, let's say we have Leadoff Hitter A, Leadoff Hitter B, and Leadoff Hitter C.
Leadoff Hitter A:
.300/.400/.500
30/50 SB
Stat: 7.2

Leadoff Hitter B:
.250/.350/.450
40/50 SB
Stat: 11.2

Leadoff Hitter C:
.350/.450/.550
10/15 SB
Stat: 0.3

What do you think? Leadoff Hitter A has a good average and OBP and a high amount of stolen bases but isn't successful when attempting to steal.
Leadoff Hitter B has a lower average and OBP, but has more stolen bases and is more successful at stealing. Leadoff Hitter C, however, gets on base at a much higher rate than B, but isn't as successful when stealing and doesn't steal as often.

[Tango Tiger]

No.

However, you are following in the footsteps of everyone else (including me as a kid) that needs to come up with some crazy combination hoping that it will tell you something.

[skyking162]

I'd love to have Pujols as my leadoff hitter.

[Second Base Coach]

Your numbers tend towards showing us the players who are most apt to show off their base stealing ability during a random game.

Is that what you want in a leadoff man? The one most likely to steal second base next saturday, the day you have tickets?

I think you have found a nice way to predict THAT, but maybe not the best way to choose a leadoff man for any given team.

[weskelton]

mlstros,

I think the bottom line here is that you are WAY overvaluing the contributions of the stolen base vs the on-base (or not out) contributions. To demonstrate this, I took the slash stats you provided and converted them into 700 PA stat lines. They look like this...


PA AB H BB XB OUTS SB CS
250/350/450 700 608 152 92 122 456 40 10
300/400/500 700 600 180 100 120 420 30 20
350/450/550 700 592 207 108 119 385 10 5
Note: XB = TB - H, OUTS = AB - H

Then I used the following Linear Weights values to convert these into absolute runs...

H=.46, BB=.31, XB=.31, SB=.22, CS=-.38, OUT=-.10


The results in Runs...


Bat Run Total
250/350/450 90.7 5.0 95.7
300/400/500 107.0 -1.0 108.0
350/450/550 127.1 0.3 127.4
Basically, the preferred lead-off man according to your stat (250/350/450) nets 0.1 runs per SB attempt at the 80% success rate. To account for the difference of 36.4 Batting runs between him and your least favorite (350/450/550), he would need to attempt 364 steals (he's only on base 244 times).

[brett]

A leadoff hitter has one guaranteed plate appearance where he is leading off an inning, and since the production of a player leading-off an inning has more potential to influence run scoring, I think that I would want a leadoff hitter who demonstrates good rates in his first plate appearance.

Slugging is a little less important and OBP a little more when leading off, so I want a high OBP first, and second a guy who is almost automatic going from first to third or second to home on singles, and also who can break up double plays.

Steals matter in that they can avoid DPs here, and for this player, it puts him a sure 1 hit from a run.

Something that looks at OB% in the first plate appearance, and first to third percentage, second to home percentage (since DPs should correlate well).

Maybe OB%(1st PA) x total relative rate of going first to third and second to home on a hit.

So if a player has a .400 OB% leading off the first, and goes from 1-3 and 2-4 at 150% of the league rate, (1.5) you could get a rating of .400 x 1.5 or 0.600

[THE OX]

Seems to me that I remember a quote by Bill James to the effect of "DO NOT multiply stat A times stat B unless there's a damn good reason for doing so."

Tango Tiger sort of reinforces that (IMO) in his remark......

[brett]

Seems to me that I remember a quote by Bill James to the effect of "DO NOT multiply stat A times stat B unless there's a damn good reason for doing so."

Tango Tiger sort of reinforces that (IMO) in his remark......

There is a damn good reason in my example. Player opportunities to take extra bases are multiplied in real fact by the rate at which they get on base!

Actually OB% times a measure of baserunning ability is also appropriate. The original poster's idea of rating speed and multipying it by OB% is a perfect example of when to multiply 2 factors. His evaluation of "speed" has problems though.

[brett]

For a leadoff hitter, percentage of team's runs scored divided by percentage of team's outs would also be a fairly good measure of pure lead-off ability. Not perfect but pretty good.

[kgrifeyjr30]

For example, let's say we have Leadoff Hitter A, Leadoff Hitter B, and Leadoff Hitter C.
Leadoff Hitter A:
.300/.400/.500
30/50 SB
Stat: 7.2

Leadoff Hitter B:
.250/.350/.450
40/50 SB
Stat: 11.2

Leadoff Hitter C:
.350/.450/.550
10/15 SB
Stat: 0.3

What do you think? Leadoff Hitter A has a good average and OBP and a high amount of stolen bases but isn't successful when attempting to steal.
Leadoff Hitter B has a lower average and OBP, but has more stolen bases and is more successful at stealing. Leadoff Hitter C, however, gets on base at a much higher rate than B, but isn't as successful when stealing and doesn't steal as often.

way too much emphasis on SBs

[brett]

way too much emphasis on SBs

Not only is there too much emphasis, but the 30/50 guy is worth less than the 10/15 guy and has a much higher "score"

Again, its not perfect, but if you take (steals minus 2 x CS) x 2 and add it to a players slugging percentage, you get a close approximation of the value.

So a guy going 50 for 70 is going to be worth about an equivalent of 20 more points of slugging (probably a little less). A guy going 90 for 105, like Raines peak is worth about 60 x 2 or 120 more points of slugging.

That's if you don't want to adjust OB%.

[Death to Crawling Things]

Not only is there too much emphasis, but the 30/50 guy is worth less than the 10/15 guy and has a much higher "score"

Again, its not perfect, but if you take (steals minus 2 x CS) x 2 and add it to a players slugging percentage, you get a close approximation of the value.

So a guy going 50 for 70 is going to be worth about an equivalent of 20 more points of slugging (probably a little less). A guy going 90 for 105, like Raines peak is worth about 60 x 2 or 120 more points of slugging.

That's if you don't want to adjust OB%.

Hey that is interesting. I had a "net" stolen base stat for my own figuring out of players years ago(must have been 20 years ago when I was a kid, never had heard sabremetrics at that point, never had taken algebra at that point for that matter, either). It was basically (SB - (CS*2) ) Which would mean Ty Cobb's 96 SB - (38 CS*2=76) 96-76=20 "Net" SB. Nice to know I wasnt completely on the wrong track. Anyhow, I always figured a CS was twice as destructive as SB was constructive due to you lose the out and you lose 1st base. Anyhow, my stat did accomplish what I was looking for which was in conjunction with looking at steal attempts you could see quality vs. quantity. Anyhow, out of curiousity where does the multiply by 2 come from anyhow?




But, yeah. With those 3 players in the original post. I believe I'd rather have Player C. A .450 OBP, oh yeah, baby. :bowdown:

[weskelton]

Anyhow, out of curiousity where does the multiply by 2 come from anyhow?

The factors come from traditional values for linear weights which are a result of analyzing the change in run expectancy from various events. Here's a snippet from the glossary on www.baseball-statistics.com (http://www.baseball-statistics.com/Leaders/glossary.htm) ...


STOLEN BASE RUNS For teams, the Linear Weights measure of runs contributed beyond what a league-average basestealing team might have gained, defined as zero; for individuals, Stolen Base Runs are calculated on the basis of the 66.7 percent success rate that sabermetric studies have shown to be the break-even point for producing runs beyond the average. Availability dependent upon caught stealing data as with Stolen Base Average. The formula is simple:

.30(Stolen Bases) - .60(Caught Stealing).

A man who steals two bases in three attempts is merely spinning his wheels in terms of value to his team, and even a man who succeeds at an 80 percent clip will have to steal a lot of bases - about 65 - to create just one win beyond average. Abbreviated by Total Baseball as SBR.

[brett]

Hey that is interesting. I had a "net" stolen base stat for my own figuring out of players years ago(must have been 20 years ago when I was a kid, never had heard sabremetrics at that point, never had taken algebra at that point for that matter, either). It was basically (SB - (CS*2) ) Which would mean Ty Cobb's 96 SB - (38 CS*2=76) 96-76=20 "Net" SB. Nice to know I wasnt completely on the wrong track. Anyhow, I always figured a CS was twice as destructive as SB was constructive due to you lose the out and you lose 1st base. Anyhow, my stat did accomplish what I was looking for which was in conjunction with looking at steal attempts you could see quality vs. quantity. Anyhow, out of curiousity where does the multiply by 2 come from anyhow?


We know that the "break ever point" for steals ranges from maybe 58% to 73% over different run environments. For most of baseball, it has been right around 66% meaning that every steal above 66% will produce positive offense.

Keep in mind that with a guy like Cobb, he probably had half of his times caught stealing when a batter failed to put down a sacrifice, or hit and run, and he got thrown out. Then again, because he was running so much on sacs, and hit and runs, he probably picked up along the lines of 30+ extra bases that were not recorded as steals.

I wouldn't give him credit for both of these differences, but either way the value is about the same.

[Death to Crawling Things]

The factors come from traditional values for linear weights which are a result of analyzing the change in run expectancy from various events. Here's a snippet from the glossary on www.baseball-statistics.com (http://www.baseball-statistics.com/Leaders/glossary.htm) ...


No I got that, I had already figured 2/3 was break even. What I meant was if someone goes 90 out of 105. He should the way I figured it have a score of 60, but how does he get 120? He was essentially doubling the score at the end. I was just wondering what the reason for that was.

[brett]

No I got that, I had already figured 2/3 was break even. What I meant was if someone goes 90 out of 105. He should the way I figured it have a score of 60, but how does he get 120? He was essentially doubling the score at the end. I was just wondering what the reason for that was.

If you take a guy with 500 at bats and a .400 slugging percentage, he would have 200 total bases. If you add another total base his slugging goes up to .404.

A steal is worth about half of a typical total base. About .4 singles, about .6 of the average base from a double, about .55 for a triple and about .53 of the average base from a home run.

That's using 1B=0.5, 2B=.75, 3B=1.1, HR=1.4, SB=0.2

So to get a quick gauge you can double the "plus SB's" and add the total to slugging percentage. 1 total base=4 points of slugging. 1 steal extra=2 points of slugging.