FanPost

DLD 7/27: Offense primer

Is the A's offense too heavily focused on on-base percentage versus batting average?  Would they do better if they got more hit-for-average types?

The quick answer is yes.

The most important offensive value calculators employ some kind of linear weights system.  Linear weights is a method by which we can estimate the number of runs certain offensive elements translate to.  One common implementation is:

Runs = .47*(1B)+.78*(2B)+1.09*(3B)+1.4*(HR)+.33*(BB+HBP)+.3*(SB)-.6*(CS)-.25*(AB-H)-.5*(baserunning outs)

In this equation, the number next to each offensive element (the coefficient) is called the "weight."  The weight is a way of assigning importance.  The typical single, for example, is worth .47 runs.  You'll notice that the weight for the single (.47) is greater than the weight for the BB or HBP (.33).  This is because a single allows runners to advance farther and more frequently than a walk.  A metric like linear weights is a refinement of OPS, which is a great stat but not entirely perfect.

In that light, let's compare Jay Payton 2006 (.296/.325/.418 (.743 OPS)) with Marco Scutaro 2006 (.266/.350/.397 (.747 OPS)), ignoring for a moment defense and positional scarcity.  The Hardball Times computes runs created per 27 outs, and in 06, Payton ranks ahead of Scutaro 5.1 to 4.6.  A large reason is because Payton has a higher batting average, so when he got on base he tended to advance baserunners.

Baseball offenses tend to be linear, so you should not be concerned that the A's are not getting the expected marginal benefit of additional OBP.  In particular, the A's offense has a middling OBP, and we would expect to seem differences to marginal benefit at very high or very low team OBP.

In the same vein, if you had the choice between two guys with the same runs created per 27 outs and one had the higher batting average and the other had the higher OBP, both would probably help the offense the same amount.  A stat like runs created corrects for distortions caused by high/low average, patience, or power.

The secret problem with the A's offense is that it's just not very good.  It has little to do with whether we value OBP or AVG incorrectly.

It is worth pointing that building on offense on a bunch of players who hit for high average is very risky.  Batting average is a skill, but it tends to have more random variance and is more strongly regressed to the mean than, say, patience or power.  Classic example: the 2002 Angels.  Their collective team line was .282/.341/.433.  Their offense was excellent.  In 2003, returning largely the same cast of characters, the team collectively hit .268/.330/.413.  That's a similar isolated slugging (SLG-AVG), a similar isolated patience (OBP-AVG), but a 20 point drop in batting average.  Their lineup:

Player     AVG 2002   AVG 2003
Molina     .245       .281
Spiezio    .285       .265
Kennedy    .312       .269
Glaus      .250       .248
Eckstein   .293       .252
Anderson   .306       .315
Erstad     .283       .252
Salmon     .286       .275
Fullmer    .289       .306

Update [2007-7-27 13:9:54 by salb918]: By popular request: a link.