Question/Help from the Gurus on Leadoff stat

[rwa12]

Question for the braniacs of the SA forum...

 

I'm trying to work out a stat for leadoff hitters, as I am not a fan of OPS for non-power spots in the lineup. There may already be such a stat, and if there is, I would love to know how it is calculated. I have a rudimentary idea for this that I called OBSB for lack of a name. (OBP+SB%) I know that this calculation has obvious flaws (# of steal attempts being a prime one), but I thought the results turned out interesting for a number of guys I tried, any help guidance would be appreciated. I am pretty new to the stat game, as OPS is still a newfangled idea to me, but am looking to expand my appreciation of the game. The reason that I chose those two stats, is because those are two skills that I think are crucial to a leadoff man. Please add/subtract/modify at will!

Thanks

Rob

 

Results

Player OBSB

Tim Raines 1.229

Rickey Henderson 1.201

Johnny Damon 1.133

Vince Coleman 1.124

Scott Podsednik 1.102

Juan Pierre 1.080

Brady Clark .958

[sbrylski]

Completely ignoring the slugging percentage of a batter leaves out a good chunk of valuable known data that can easily help evaluate a batter. Though SLG for a leadoff is probably a little less important, it still matters. We know on average OBP is about 1.6 to 1.8 times as important as SLG. It's going to be more valuable than that for the top two guys in the order getting on base infront of the boppers and probably a bit less important for the 7th and 8th guys in the lineup because its unlikely that the 8-9-1 hitters will drive the player in if he only reaches 1st.

 

However, for that same reason a SB is going to be more valuable infront of the 7-8-9-1 hitters than infront of the 2-3-4-5 hitters. If the middle of the order guys are going to be knocking doubles around and hitter homers, you might as well sit on first. The 7-8-9-1 hitters would have to string two or three hits together to score a run, so any advancement by the runner by way of a steal is going to be very helpful. Also, getting caught won't hurt as bad because its a lot less likely the runner would have scored had he stayed put.

 

I guess what I'm getting at is that a more accurate stat for leadoff hitters would have to take into account the value of OBP compared to SLG and the SB value out of that spot in the lineup. That's probably a lot more in depth that your original goal.

 

What you have so far looks okay for a rough stat. It be interesting to see how well it correlates to runs scored or runs created or something.

 

Also, for whatever reason when I saw your sample I thought that if you divide them by 4 you drop the numbers into a scale fairly close to that of batting average; i.e. Raines - .307, Pierre - .270. Just a thought.

[rluzinski]

How come hitting yourself to 2B while also moving existing runners arround the bases is less imporant than getting to first and then stealing a base without moving anyone else over? It's a logical nightmare when you think about it.

 

If your goal is to measure the players who best the stereotype of a good leadoff hitter, than you might be on the right track. You need to weight OBP and SB% better, however. .060 is the difference between a horrible and great OBP. for SB%, .25 is maybe the diff. between poor and good. Also, OBP is just WAY more important to begin with. I'd zero SB% to .66. Anything less than that and you are probably hurting your team. Maybe something like this:

 

LO = OBP + [(SB% - .66)/2]

 

.350 OBP

.75 SB%

LO = .440

 

.325 OBP

.60 SB%

LO = .295

 

You need to set some kind of min SB attempts too. Maybe 1 SB attempt for every 15 AB?

[sbrylski]

Did some digging, found this: LINK

 

So based on that, these are the relative values for the leadoff hitter, rounded to the 10ths place:

 OBP 2.5 SLG 1.0 SB/G .2 CS/G - .6

The first hitter in the lineup will recieve about 750 plate appearances a year, or ~4.6 per game - LINK. Therefore, SB/G = SB/PA*4.6

 

An equation like this should give you your best analysis:

 

LO = [OBP*2.5]+[sLG]+[(SB/PA*4.6)*.2]-[(CS/PA*4.6)*.6]

 

Simplified to:

 

LO = [OBP*2.5]+[sLG]+[sB/PA*.9]-[CS/PA*2.8]

 

 

How would I get a more relative comparison of their actual value? I took a replacement player's line (is OBP/SLG of .266/.335 accurate for that?) and plugged it into the LO formula. Then subtracted that from LO to find LORP:

 

LORP = LO - .996

 

Does this make sense/work right?

 

 

Here are the results:

 

http://i35.photobucket.com/albums/d175/sbrylski06/LORP.jpg

 

 

Any critiques or suggestions? I'm sure I've made a mistake or two or there are ways to improve on this.

[sbrylski]

I like this better, Leadoff Above Average:

 

LOAA = LO - 1.257

 

I took the average production of NL teams out of the leadoff spot in 2006 and used that to generate league average LO. Subtracting it from the players LO gives LO above average. This way I'm comparing leadoff hitters to other leadoff hitters, rather than just an average replacement player. (I'm not even sure if my replacement stats were correct.)

 

Here's the results of LOAA:

 

http://i35.photobucket.com/albums/d175/sbrylski06/LOAA.jpg

[sbrylski]

In case anyone is interested, here are his values charted:

http://i35.photobucket.com/albums/d175/sbrylski06/LP.jpg

 

He expressed some concern over there accuracy, but they are decently close. They at least show how much OBP and SLG values can vary between lineup spots.

[mothershipconnection]

I would either multiply something by number of SB attempts, or better yet use percentage of on-base occurrences that were paired with a SB attempt. So you'd have OBP, SB%, and %(on base+attempt). This would show their propensity to "cause a disruption" by stealing or attempting.

 

So you could have OBP+[sB%(%on base with SB attempt)]

Then, attach a weight to the OBP and SB components and you'd have something with at least maybe a little utility.

 

Otherwise, SB% is useless as Damian Miller could be 1/1 and have a better % than Henderson.

[vacin8r]

Ive been tinkering with a leadoff stat myself recently. And except when a team had a dominant slugger(which added more runs scored than they should) the correlation seems to be about right. I must admit however that my sample size in this the planning stage is very very small. The stat im going for is something along the lines of bases reached per plate appearance(BRPA for short). Take (total bases + walks + hbp + sb - cs) / plate appearances. Just using last years stats as an example:

sizemore:.6072 BRPA

utley: .5886 BRPA

reyes: .5946 BRPA

hanley ramirez: .5714 BRPA

furcal: .5271 BRPA

eckstein: .3967 BRPA(and these guys won the WS)

weeks: .5036 BRPA

gross: .5476 BRPA

brady clark: an astonishingly low yet better than ecksteins .4046 BRPA

 

for many of the major leaguers if i divided bases reached by their run total the numbers come out around 3.4. rickie is an anomoly in that he scores at a much higher clip than his peripherals demonstrate with 2.84 bases earned per run. On the other end of the spectrum brady clark needs 3.82 bases earned per run i would guess demonstrating his lack of aptitude for running the bases. also as i said above teams with players like howard or pujols naturally have lower bases/run although not that pronounced as utleys is 3.32 barely below the norm, and ecksteins is 3.22.

 

like i say, still kinda in the experimental stage, but i like the results so far.

-Don

[rluzinski]

Take (total bases + walks + hbp + sb - cs) / plate appearances.

 

That's just so arbitrary, though. To generate a meaningful metric, you need to define what the metric is trying to measure and there needs to be a reason WHY you weight various components the way you do.

 

For instance, according to your above definition, a guy with 50 SB and 50 CS in a year has no effect on your leadoff metric. A guy like that probably really hurt his team, however.

[vacin8r]

It is true, while statistical data suggests the actual break even point for SBs is at or around 72% and for the sake of my statistic i am setting the break even point at 50% for no other reason than simplicity. it would probably be more correct of a stat to have it be (tb + bb + hbp + (sb * .72) - cs and po) / pa but i went the simpler route:

1) because it would be much easier to calculate/explain

2) one would hope that if a batter is stealing less than 60% anyways his manager would nail his freakin foot to the bag.

My main interest was a stat i could easily fire out with a calculator and a stat sheet that would combine obp, slg, and steals to gauge not only a leadoff hitter, but any hitter. My guess would be that Pujols prolly had the highest BRPA (.7256 in 2006) in the league but his ridiculous slg puts him into the roll of a 3/4 hitter 1st. Its not exactly perfect, but based on the fact it does seem to match well with the number of runs scored that it is a usable stat.

-Don

ps-yeah the 50 sb 50 cs player you describe according to statistics would cost his team about 20 runs on the season vs never running at all. just thought id throw out a little statistical analysis towards your gripehttp://forum.brewerfan.net/images/smilies/smile.gif

[rluzinski]

There are dozens of "runs created" equations already out there, from simple to extravogent. The most simple is basic runs created:

 

RC = OBP x TB

 

It ignores SB and CS because, for most players, it has little or no effect anyway. If you wanted to know the average value of each baseball event, just use something like linear weights. Each event has a value associated with it... then you just add them all up.

[TooLiveBrew]

vacin8r, I don't mean to nitpick for the sake of doing so, but Chase Utley is not the leadoff man for the Phils. That's Jimmy Rollins. Utley, IIRC, batted 3d for most of the season(at least that's where he was down the stretch, I think).