OBP question

[pmg4]

I'm mostly a lurker, and I don't know squat about stats. So, that's now out in the open.

 

There's a discussion in the Major League Forum that argues BA vs. OBP that triggered this question, if you're wondering, and this seemed like the right place to ask.

 

Most people agree that a BB is less valuable than a hit, statistically. Is there a version of OBP that reduces the value of walks compared to hits? I think that would silence a lot of people who feel that BA is not being taken seriously in the current atmosphere of walk-love in stat discussions (myself included).

[TooLiveBrew]

One problem I see right off the bat is that a single many times isn't worth more than a BB... and the situations in which a single is worth more are entirely out of a batter's control. I agree, though, that it'd be cool to see an adjusted OPS with slightly more weight going to hits. As it is (though not calculated in plain ol' OPS), OBP is seen as 1.8 times as valuable as SLG, iirc... but I have no idea about a random H's value vs. that of a BB.

 

Try RC (Runs Created), a stat Bill James created, and has been tweaked many times. Here's The Hardball Times's write-up on the stat:

 

 

"Runs Created. Invented by Bill James, RC is a very good measure of the number of runs a batter truly contributed to his team's offense. The basic formula for RC is OBP*TB, but it has evolved into over fourteen different versions. We use the most complicated version, which includes the impact of hitting well with runners in scoring position, and is adjusted for ballpark impact. RC/G refers to Runs Created Per Game, which Runs Created divided by the number of outs made by the batter, times 27."

 

 

I like that it rewards hitting well w/RISP & addresses Park Factors, among other adjustments. For a smaller/easier-to-look-at version, RC/G seems interesting (see the same link for RC, directly below RC in the glossary).

 

Another one that I personally don't understand as well (bc I'm not a numerical guru like some here) is EqA. The following definition is from Baseball Prospectus:

"Equivalent Average. A measure of total offensive value per out, with corrections for league offensive level, home park, and team pitching. EQA considers batting as well as baserunning, but not the value of a position player's defense. The EqA adjusted for all-time also has a correction for league difficulty. The scale is deliberately set to approximate that of batting average. League average EqA is always equal to .260. EqA is derived from Raw EqA, which is (H + TB + 1.5*(BB + HBP + SB) + SH + SF) divided by (AB + BB + HBP + SH + SF + CS + SB). REqA is then normalized to account for league difficulty and scale to create EqA."

 

 

I know neither one of these two answers your query perfectly, but I figured it'd be a start until the real stat guys can chime in.

[logan82]

Depending on how hitting with RISP is factored in, it could really skew things.

[sbrylski]

Problem is you are trying to use OBP to measure something its not intended to measure. OBP is a measure of how often someone doesn't make an out - that's about it. It is also logically the most basic definition of a successful trip up to the plate, so that's why its so important.

 

You can do it, but the average value of a hit is going to vary from player to player, which is measure pretty good by SLG. I'll take a Fielder hit over a Kendall hit every time. So it would be better to just look at OBP and SLG together, simplified in forms like OPS, OXS (OBP*SLG), or GPA (OBP*1.8+SLG).

 

To get what you want, I'll start with Pete Palmer's Linear Weights and multiply them by each events' occurrence per hit (data from last season, entire MLB), and then add those values together to get the average run value of a hit:

 

 1B 2B 3B HR
Run Value .47 .78 1.09 1.40
Occurrance/Hit .664 .204 .021 .110

Hit Average Run Value
HARV = [.47*.664]+[.78*.204]+[1.09*.021]+[1.40*.110]
HARV = .65 

Comparing it to the average run value of a BB or HBP (.33) will give us the factor to multiply the hit component of OBP by, and give us the resulting formula. Lastly, if you multiply that result by .33*PA, you'll get the actual runs created.

 

 X = .65/.33
X = 1.97
Adjusted OBP = BB% + [H% * 1.97]
Runs Created = Adjusted OBP * PA * .33 

Remember though, this is all assuming league average hit distribution. The 1.97 will vary from hitter to hitter. For example, Jason Kendall's career constant would be 1.74 while Prince Fielder's would be 2.32.

 

So, in short, I calculate a hit is worth about twice as much as a walk on average. Can anyone confirm (I couldn't find a source to verify) or tell me where I might have went wrong?

[sbrylski]

I just noticed that the weights for extra-base hits go up by .31 each time, which allows the formula for the constant to be simplified so you don't have to break it down to each type of hit that each player gets. So for this 'Adjusted OBP' you just need H, TB, BB+HBP and PA:

 HARV = .47 + [.31*(H-TB)/H] X = 3*HARV Adjusted OBP = [BB+HBP]/PA + [H/PA * 3 * (.47 + .31*{H-TB}/H)] Runs Created = Adjusted OBP * PA/3 

The only problem I have is my Runs Created are coming out weird. I have to look at it again.

 

EDIT: http://www.sju.edu/~sforman/research/talks/baseball/node12.html

 

My RC formula based on the Adjusted OBP matches what I get when I use the linear weights on the site linked above, but neither result equals anywhere close to the 'basic runs created' forumla on the site. Anyone know why? I'm assuming they should be exactly the same...

[pmg4]

Gents, I appreciate the time spent on this. I have to say, by the time we get to those last two answers, some of the formulas are beyond intuitive understanding for me. I get algebra, but I find myself constantly referring back to figure out what each of the variables represents.

 

 

Anyway, I wanted to let you know that I did check back to see what the stat-heads had to say about my theoretical question. I have a couple of responses.

 

TooLiveBrew: One problem I see right off the bat is that a single many times isn't worth more than a BB... and the situations in which a single is worth more are entirely out of a batter's control.

You are entirely correct. Every at-bat is an individual situation, but I have read this as the fall-back position of anyone who wants to stop arguing about a statistic's relevance. As in: you can't credit singles as better than walks because you don't know every hitting situation. But that's the whole point of stats, as I understand them. On average a certain percentage of all ABs will be with men on base, in scoring position, bases loaded, etc. In those cases, a single is much more valuable than a walk. Also, there will be another percentage where the bases will be empty, and in those cases, a BB is worth exactly the same as a single, assuming no errors are committed. I'm just bothered by those who claim (Homer True is a notable example) that BA should be supplanted by OBP as the relevant stat of a player's value at the plate, because for that to be really true, a form of "adjusted OBP" must account for the differential in value between hits and walks.

 

Sbrylski06: Problem is you are trying to use OBP to measure something its not intended to measure.

I completely agree with this also. "On-base percentage" measures exactly what is proclaims to. I was hesitant to frame my original question using OBP as the name for the stat, but that's the starting point for what I think would help identify a player's true "plate value" (there we go: PV = H + .5BB + .5HBP/ PA). That's probably not exactly right. I'm just having fun writing up an equation.

 

I promise you all, I don't know nearly enough to start calculating and throwing out "PV" for every player that the Crew considers in a trade. You'll probably never hear it again. I needed a place to air my own grievance with OBP as everyone's favorite stat. I'm admittedly a traditionalist who likes to know a player's BA, but I recognize that an Estrada who may hit .300 and take 9 walks per season is not more desirable than a Hardy hitting .280 but taking 60 walks.

 

I'd like to see knowledgeable people point out that OBP is not the thinking man's BA. It is a flawed stat because a .340 OBP is not just as good as a .340 hitter with no walks.

[sbrylski]

I'd like to see knowledgeable people point out that OBP is not the thinking man's BA. It is a flawed stat because a .340 OBP is not just as good as a .340 hitter with no walks.

 

 

 

I think all stat guys would acknowledge they are not as similar as they sometimes are presented to be - they measure different things. But OBP alone is better than BA alone, and when paired up with SLG is very accurate at evaluating a players plate contributions - it leaves BA as a stand alone in the dust.

 

 

 

Look at this graph:

 

http://www.hardballtimes.com/images/uploads/dlf_ops1.JPG

 

 

 

Its just depends on your definition of the goal of the batter. It had always been for the goal of the batter to get a hit, so BA was the stat to measure it. But logically, wouldn't not making an out be an even better goal? Statistically it has been proven to be so, and therefore OBP, the stat that measures the ability to not make outs, takes precedence.

 

 

 

Batting average alone can be okay, but its also simple to just look OBP and SLG, or better yet OPS, and get a more accurate evaluation. The graph above also displays how all those fancy statistics some whip out are actually not much better than OPS is anyway, on the team level at least.

[logan82]

So basically you really don't need to go to much farther than BA/OBP/SLG/OPS to get a good feel for a hitter?

[Phenomenal Smith]

So basically you really don't need to go to much farther than BA/OBP/SLG/OPS to get a good feel for a hitter?

 

I would say that's generally correct. If two player have a similar OPS, you should look at which has a higher OBP since that's more important than Slugging (by a factor of about 1.7 ). Looking at all 4 of them together gives you a good idea of what kind of hitter the player is.