THE BOOK--Playing The Percentages In Baseball

PMR had Tulowitzki saving 50 outs at shortstop this year.  So 40 runs a year at shortstop doesn’t surprise me that much.

Don’t forget that I’m quoting 0.24 runs saved per game, for the top 50 SS of the last 50 years for their entire careers.

50 outs (40 runs) is pretty much in-line with what UZR, WOWY, TotalZone, Dewan, PMR, et al will give you for a top-end SS in a given year.  It’s quite another for that to be the average for a player’s career; and still quite another for that to be the average for 50 players.

I’ve calculated all seasons using the same methodology from my 1956-1986 article.

Tulowitzki at +33 comes out with the best season of any shortstop in that span, beating out Mark Belanger’s 1975 (+32).  What a player.

The Rockies had a tremendous season defensively—752 runs allowed is their fewest in a full season. Even the ‘95 team gave up more runs in 18 fewer games. Their pitchers’ babip was the lowest in team history, too. Probably just a coincidence that they added Tulowitzki and a real defensive CF for 2007. I wish I had more opportunities to see Tulowitzki play.

It’s possible that there is a bias in that good fielders attracts other good fielders and good pitchers.

I think it’s just as possible that it’s an organizational thing. An organization with good scouts, will find a lot of the top fielding talent. It makes intuitive sense that it’s harder for a scout to spot the next Ron Hansen than the next Cecil Fielder.

Anything having to do with the Rockies has to be taken with a grain of salt, or at least some further research has to be done.

To start quoting the Rockies runs allowed or runs scored (such as this year’s 752) is meaningless without talking about the Coors park factor.  Prior to the humidor being installed, the Rockies obviously had a ridiculous hitters’ park.  Since the humidor, the PF has gone way down.  Since 2006, when supposedly they began keeping balls in the humidor for a longer period of time, the park plays just a little above neutral.  So to compare anything unadjusted the last 2 years with anything prior to that and especially prior to 2002, without doing any kind of park adjustment, makes little sense.  Even unadjusted defensive data for the Rockies’ players has to be taken with a grain of salt.

The Rockies had a tremendous season defensively—752 runs allowed is their fewest in a full season. Even the ‘95 team gave up more runs in 18 fewer games.

For example, in 06, the Rockies only gave up 37 more runs on the road.  In 04 and 02, the gave up less than 30 more runs on the road.  Sure the Rockies had a pretty good pitching and defensive team this year, but given the park conditions, I would say that the 07 team was probably not a whole lot better defensive (and pitching)-wise than the 02 and 04 teams, and maybe worse than the 95 team.  And according to B-R and MLB.com, the Rockies gave up 758 runs in 07, not 752.  No big deal of course. Maybe you didn’t include the play-off game.

Yes, it’s 758, not 752—my error. And of course that includes the extra regular-season game. Still, the home babip was a record low and the road babip was the second-best in team history. I don’t think it’s just the humidor. I guess we’ll see if they can do something similar this year.

Pinto’s charts on Tulo and Atkins are interesting:  http://www.baseballmusings.com/archives/024862.php.  Tulo appears to be exactly average going to his left, but makes a lot of extra plays to his right.  But Atkins is terrible going to his left, so Tulo is likely getting at least some balls that a typical 3B would take, perhaps quite a few.  This raises an interesting possibility, which is that Tulo contributed 50 outs to the Rockies (assuming that Atkins couldn’t have handled any of Tulo’s balls, which may not be true), but would have only contributed, say, 20 outs to a team with an average 3Bman.  And if so, which is the “correct” assessment of his defensive contribution?

* *

Tango’s 40 run spread on the top 50 SSs is amazing, and I have to think at least half of that has nothing to do with the SS.  Looking at the list, a lot of these guys played a lot of years on good teams—not surprising, since many of them were stars.  I would guess that the “extra” runs saved come from these sources, in order of importance:
* better pitchers
* better fielders at other positions
* worse opposing hitters (since these SSs are themselves good hitters, and I’m guessing their teammates were as well).

Ozzie had Pendleton.  It’s possible that there is a bias in that good fielders attracts other good fielders and good pitchers.

At least in regards to the Cardinals, it seems it was organizational.  The Cards traded away a fast 24yo great hitting (for a SS at the time) 2 time all-star, silver slugger winner who had received votes for MVP twice already and had good range for a 25yo Ozzie Smith who was no hit/great field.  They traded for Oquendo and McGee, while bringing up Pendleton and Oberkfell.

A long time ago, I looked at whether infielders can “steal” outs from other fielders or not (as in the case with Atkins and Tulo).  IIRC, I found that there really isn’t much overlap, so I doubt that much of Tulo’s 50 extra outs are because of Atkins poor range.  But I can look into it also.  There are all kinds of reasons why a player can appear to have a great or terrible season, by (partial) chance.  That is why we use multi-year data or regress the data a lot when we only have small samples.  50 extra outs is not really 50 extra outs.  It is probably around 25 extra outs (50% regression).  That is the essence of regression in the first place, no?  What usually happens is that while on “paper,” (the data) it looks like a player has saved 50 outs, if you actually looked at the plays, you might see that he had balls that were a little closer or a little softer or had an easier bounce than the average ball represented by the written data.  In addition, a player can make extra “lucky” catches that are not sustainable.  There are all kinds of “reasons” for the sample record of a player’s performance not to be “sustainable”, which is why we regress. So let’s not assume that Tulo “really” made 50 plays that an average SS might have made.  That is extremely unlikely.  And even if “stealing” balls from the 3B is a factor, it is likely only a factor in maybe a few of those 50.