# Average Is as Average Does

Before you dig too deeply, “average” seems like such a simple concept. But like most things, the way we talk about average production in baseball doesn’t always correspond to simple arithmetic.

Earlier this week, I posted 2006 positional averages on my blog. Those numbers are great to have; if you want to know how good your shortstop is relative to other shortstops in baseball, it’s important to know that the average guy at that position hit .275/.332/.408 last year. However, those numbers don’t exactly correspond to what we think of as average production from a shortstop.

Consider an example. While he didn’t open the year as Milwaukee’s starting shortstop, Bill Hall amassed 463 of the 618 at-bats given to Brewers at that position. In that time, he hit .259/.339/.518, far more productive than the positional average. The other 150 shortstop at-bats didn’t go so well for J.J. Hardy, Chris Barnwell, and others; they dragged down the overall average to .254/.331/.482. That’s still better than average, but it’s much less of an advantage.

At most positions, starters get 75-80% of a team’s playing time. When your team signs a free agent to take over a vacant position, that player probably isn’t going to play 162 games any more than your new mid-rotation starter is guaranteed to make 34 starts. Since the backups who will play the rest of those games will almost almost always be worse, the production you get from the position will be less than what you get from the starter.

###### Starter Averages

In other words, comparing that free agent to league average is misleading unless you know he’s going to play 162 games.

As an alternative, I calculated “starter averages” by finding (obviously) the average production from MLB starters. Since at most positions, a few teams don’t have a set starter, but instead go with a platoon or rotation of minor-league tryouts, I only averaged the 25 players at each position who got the most playing time. Here are the results:

```Position        AVG     OBP     SLG     OPS
Catcher         0.285   0.342   0.435   0.777
First Base      0.290   0.374   0.508   0.882
Second Base     0.284   0.341   0.426   0.767
Third Base      0.284   0.356   0.482   0.838
Shortstop       0.280   0.336   0.422   0.758
Left field      0.284   0.363   0.481   0.845
Center field    0.274   0.341   0.447   0.788
Right field     0.285   0.358   0.469   0.828
Designated Hitte0.261   0.354   0.483   0.837```

To recall, the average shortstop production I mentioned earlier was .275/.332/.408. Here, each number is higher, as we’ve filtered out the thousands of plate appearances accumulated by John McDonald, Tomas Perez, Andres Blanco, and about 100 other guys who got fewer than 200 at-bats at the position.

Another way to look at this is to consider the difference between the starter average and the overall average at each position. Here they are, measured by OPS:

```Position       Overall Starter Diff
Catcher        0.745   0.777   0.031
First Base     0.851   0.882   0.031
Second Base    0.744   0.767   0.023
Third Base     0.805   0.838   0.033
Shortstop      0.740   0.758   0.018
Left field     0.818   0.845   0.027
Center field   0.761   0.788   0.027
Right field    0.807   0.828   0.021
Designated Hitt0.811   0.837   0.027```

I’m not sure there’s much rhyme or reason behind what makes the differences greater at some positions that others. It certainly makes sense that the difference at catcher is large; while starting catchers are not always known for their offensive prowess, backup catchers are frequently defined by their lack of it.

###### The Value of an Ironman

If the difference between the starter average and the overall average is 25-30 points of OPS, there’s clearly a big difference between the typical starter and the typical backup. (This is the kind of cutting-edge analysis you count on here at The Hardball Times, right?) It follows, then, that the more your starter plays, the better off you are.

The difference between the starter average and the overall average was smaller at shortstop than any other position, but the gap between regulars and backups is still dramatic. As we saw, the average starter hit at a .280/.336/.422 clip. By contrast, all other shortstops managed only a .263/.320/.372 line. It’s a substantial difference, but it isn’t one that cripples a team every time their starting shortstop takes a day off.

The average starting shortstop had 535 at-bats last year, while the average team got 631 at-bats from all of their shortstops. In other words, fewer than 100 at-bats came from backups. Using the simple version of runs created, the difference between an average starter’s playing time and a guy making every single start is a little less than three runs.

Yep, only three runs. Granted, that isn’t true in every case. For any specific team, the difference is defined by the production of the starter and the backup, not by league averages. It would hurt a lot more to lose Miguel Tejada for a game or two per week than it would to have Felipe Lopez miss the same number of at-bats. Similarly, it’s much worse to find yourself plugging in Ramon Santiago than Craig Counsell.

We can, however, make a tentative conclusion. Unless there’s a dramatic difference in defensive skill, there isn’t a lot of value in stretching out your starter from 140-150 games up to 162. Especially if squeezing out those extra 10-20 games results in an increased injury risk, the cost far outweighs the benefit.

###### Position by Position

So far, I’ve only discussed shortstops. The difference between the overall averages and starter averages were the lowest at that position, so it’s worth looking a little more closely at the other spots on the diamond.

In the following table, the column labeled “AB-diff” lists the number of at-bats that backups typically accumulate over the course of a season. “RC-starter” shows the runs created by the average starter in that many at-bats; “RC-backup” does the same for backups. The final column is the difference. Theoretically, that’s the difference in production between a team where the starter plays a typical number of games and a team in which the starter plays 162 games.

```Position         AB-diff  RC-starter RC-backup  RC-diff
Catcher          173      26         21         5
First Base       140      27         22         5
Second Base      123      18         15         3
Third Base       109      19         14         5
Shortstop        96       14         11         3
Left field       162      28         24         4
Center field     153      23         19         4
Right field      171      29         25         4```

As you can see, there are slight differences from position to position, but not much. Nothing here challenges the earlier hypothesis that, in the abstract, there isn’t much to be gained from pushing your starter to play more than a typical number of games.

###### One Step Further

Of course, there are other benefits to ironmen, especially ones that consistently turn in 160+ game seasons. They increase roster flexibility by making it less important that you carry a specialized backup just for them. This is particularly valuable for shortstops: for years, the Orioles didn’t have to worry about one of their infielders being able to fill in for Cal Ripken.

That roster flexibility can turn into financial savings, as well. A backup infielder who can hit .265/.320/.380 (think Craig Counsell) costs a lot more when he can play a convincing shortstop. (The same type of thing applies to outfielders who can fill in at all three positions, or corner infielders who can play second base.) A player with Counsell’s level of offense is a \$3 million investment at shortstop. If you can settle for the same offense out of a 2B/3B, you might find it on the waiver wire.

Also, an ironman doesn’t have to be as good as a non-ironman to give you league-average production. For instance, Orlando Cabrera started 152 of his team’s games last year and hit .282/.335/.404. As we saw above, that’s below average in both OBP and SLG for a starting shortstop. On the other hand, because he played much more than the average starter, the Angels’ overall production from shortstop was almost exactly the same as Cabrera’s personal line.

Thus, instead of comparing Cabrera’s stats to the starter average (as we might with, say, Jack Wilson‘s 533 at-bats at the position), we can compare him to the overall average of .275/.332/.408. Next to those numbers, Cabrera looks at least average, perhaps a bit better. In the grand scheme of the Angels’ win-loss record, that distinction doesn’t matter much; as we’ve seen, it’s probably fewer than three runs, especially since the Angels had quality options (Chone Figgins, Macier Izturis, and Erick Aybar) to back him up.

Next time you compare a player to league average, don’t forget his playing time. Since somebody else has to take his place every time he misses a game, a starter’s contribution goes beyond his simple rate stats.

« Previous: THT Links: Lend a Hand
Next: Business of Baseball Report »