24 thoughts on “Seth Smith and the (almost) lost art of pinch-hitting”

1. Richard Chester December 16, 2013 at 9:49 pm

   Looks like Dave Hansen holds the lifetime and seasonal records for sacrificial PH appearances. He was PH for as a PH 63 times career-wise. In 2003 he was PH for 13 times as a PH and there were no inning-ending outs-on-base.

   Reply ↓
   1. Doug Post authorDecember 16, 2013 at 10:42 pm

      Good find, Richard.

      I’m guessing Hansen in 2003 also has the single season record for highest percentage of league total with 18.5% of the NL’s 70 PH appearances without a PA.

      Reply ↓
      1. Richard Chester December 17, 2013 at 12:07 am

         Hansen’s percentage was 13/71 = 18.31%. It’s not a record. Lenny Harris had 19.23% in 2005 and Dale Coogan had 20.00% in 1950. Prior to 1943 there were several seasons in which there were fewer than 6 player-games with 0 PH PA so obviously there had to be at least 1 player with at least 20.00% in each of those years. The results are for the NL only.

         Reply ↓
         1. Doug December 17, 2013 at 3:00 am

            Richard,

            Nice work again on Lenny Harris. The 1950 data are described in B-R as “mostly complete”, with data before then only spotty (apparently substitutes are not always shown with positions in the older boxscores; or they may be shown with a position signifying only that they appeared in the game in the field, but without precluding the possibility of having entered the game as a pinch-hitter).

            Actually Richard, it was only 70 PH for PH appearances in the NL in 2003. The 71st that you noted was by Henry Blanco on Aug 20 who pinch-hit for Javy Lopez in the middle of Lopez’s 8th inning AB, and completed a strikeout that was charged to Lopez. Blanco remained in the game to catch the 9th inning, so appears as PH-C in the box score without a PA.

            Reply ↓
            1. Richard Chester December 17, 2013 at 11:29 am

               Thanks for the correction for the NL in 2003 but how did you pick that up about Blanco? Did you check each 2003 PH appearance individually?
            2. Doug Post authorDecember 17, 2013 at 11:53 am

               I had clicked “Played exactly as marked or not” for the PH position and with PA=0 and found 70 occurrences for the NL in 2003. I used the “exactly” option as any pinch-hitter who was pinch-hit for could not have remained in the game and so would only appear in the boxscore as having been a pinch-hitter. This method will be 99% accurate, but can still be foiled by an aborted PA with an inning-ending caught stealing or pickoff (or an oddball appearance like Blanco’s, had Blanco not remained in the game).

               When you said 71 games, I figured you had clicked “Played any of those marked” and picked up an extra game where the player stayed in the game after pinch-hitting (and, so, could not have been pinch-hit for). Looking at the position column in the list of games, it was easy to spot the one where the player stayed in the game.
            3. Richard Chester December 17, 2013 at 12:27 pm

               Thanks Doug. That was very clever on your part.
2. Brendan Bingham December 16, 2013 at 9:58 pm

   Doug,
   Some very nice analysis here.
   A few comments:
   1) To your second graph, the drop off in the NL 4+ PA games line does not seem out of proportion to the increase in the NL pinch hit PA per team game line. Bear in mind that these two metrics are graphed on different scales, and even in the early ‘70s team games with four pitcher PAs were somewhat rare (20%). The decrease to 5% in recent years represents only a marginal increase in the opportunity for pinch hitter PAs.
   2) To your third graph, pinch hitter production has never been great. And to my eyes, it seems that any recent trends are subtle, at best. I suspect that if the MLB OPS+ numbers were graphed as 5-year rolling averages, the trend line would mostly reside between 75 and 85 OPS+.
   3) Interesting that since 1990, there has been a steady decline in pinch hitter platoon advantage (your fourth graph), but this trend is not obviously coupled to a decrease in pinch hitter OPS+ (your third graph).

   Reply ↓
   1. Doug December 16, 2013 at 10:23 pm

      Thanks Brendan,

      To your first point, perhaps 4+ PAs by a starting pitcher was not the right metric to illustrate the point, but I do think the decline in length of pitcher starts has tended to increase NL pinch-hitting, as seen from the early ’70s to the mid ’80s, and which has only been tempered since by the necessity of resorting to the double-switch instead of pinch-hitting.

      Re: point 2 and 3, it’s difficult to see on the OPS+ chart but, if you look closely at the MLB (green) line, you’ll see it was between 80 and 90 in all but two years from 1973 to 1986, and below 80 in all but two years since 1987. I think that’s a meaningful drop that does correlate well with the platoon advantage chart which rose rapidly when OPS+ was in its higher range, and started its drop from peak levels about 1990, just after OPS+ dropped into its current lower range.

      Reply ↓
3. bstar December 16, 2013 at 10:37 pm

   Seth Smith’s career BABIP as a pinch hitter: .392

   Reply ↓
   1. Doug Post authorDecember 16, 2013 at 10:44 pm

      Most of that is Coors, it would seem. Since coming to Oakland, Smith’s seasonal BABIP as a pinch-hitter has been .333 and .286.

      Reply ↓
      1. bstar December 16, 2013 at 10:47 pm

         Me too. I think his 132 tOPS+ in Coors Field is coming into play also.

         Reply ↓
4. Arsen December 16, 2013 at 11:33 pm

   It looks like Greg Gross was more a pinch walker in his career. Amazing how many walks he drew considering he had no power.

   Reply ↓
   1. Doug Post authorDecember 17, 2013 at 3:30 am

      Good observation, Arsen.

      Since Greg Gross conveniently had 733 pinch-hit PAs, a very full season’s worth, I looked for season lines with the most walks without a home run.

      Among homerless seasons, only Eddie Stanky (137 BB in 1946) and Topsy Hartsel (121 in 1905) have more walks than Gross’s walk total of 116 as a pinch-hitter. And only Rabbit Maranville’s 746 PAs in 1922 (but with only 61 BB) exceed Gross’s pinch-hit PA total.

      Rk	Player	BB ▾	HR	Year	Age	Tm	Lg	G	PA
      1	Eddie Stanky	137	0	1946	30	BRO	NL	144	642
      2	Topsy Hartsel	121	0	1905	31	PHA	AL	150	674
      3	Richie Ashburn	116	0	1960	33	CHC	NL	151	672
      4	Max Bishop	116	0	1926	26	PHA	AL	122	529
      5	Roy Thomas	115	0	1899	25	PHI	NL	150	702
      6	Roy Thomas	115	0	1900	26	PHI	NL	140	675
      7	Donie Bush	112	0	1914	26	DET	AL	157	721
      8	John McGraw	112	0	1898	25	BLN	NL	143	652
      9	Johnny Evers	108	0	1910	28	CHC	NL	125	556
      10	Roy Thomas	107	0	1902	28	PHI	NL	138	626
      11	Roy Thomas	107	0	1906	32	PHI	NL	142	620
      12	Luke Appling	105	0	1939	32	CHW	AL	148	633
      13	Max Bishop	105	0	1927	27	PHA	AL	118	491
      14	Jack Crooks	103	0	1891	25	COL	AA	138	631
      15	Eddie Collins	101	0	1912	25	PHA	AL	153	673
      16	Yank Robinson	101	0	1890	30	PBB	PL	98	417

      Provided by Baseball-Reference.com: View Play Index Tool Used
      Generated 12/17/2013.

      Reply ↓
   2. no statistician but December 17, 2013 at 9:34 am

      Two things:

      I remembered Gross only as a name so I looked him up, and 1) he didn’t actually walk that much when he was not pinch hitting; 2) as a highly used pinch pinch hitter he was the worst at scoring runs and driving them in.

      Reply ↓
5. Darien December 17, 2013 at 9:18 pm

   I’m a bit wary of the claim that Seth Smith would still top the pinch-hitting effectiveness list with room to spare if OPS+ were available; my inclination is to think that his years in Colorado trump his 70-point OPS advantage to a significant degree.

   Smith is, though, still a pretty good hitter overall, and I remember never being able to figure out why the Rockies seemed to value him so lightly. He’s not a superstar by any means, but he’s decent with the bat, and they used to leave him on the bench in favor of the Dexter Fowlers and Eric Young Jr.s of the world. This never made sense to me. And then they traded him for nobody and nothing. I just don’t get it.

   Reply ↓
   1. Doug December 17, 2013 at 9:43 pm

      I suppose one could estimate the OPS+ as a pinch-hitter by multiplying the player’s career OPS+ by TOPS+ as a pinch-hitter.

      Thus:
      – Smith: 109 x 1.42 = 154.8
      – Northey: 124 x 1.24 = 153.8
      – Johnson: 125 x 1.15 = 143.8
      – Colbrunn: 106 x 1.14 = 120.8
      – McCovey: 147 x .91 = 133.8

      Reply ↓
6. bstar December 17, 2013 at 10:59 pm

   We’d have to know what OBP and SLG an average offensive player would produce in every pinch-hit AB by these players, in that specific park and era, to know for sure. Then we could generate a lgSLG and lgOBP number for each player in their PH career to judge their individual pinch-hit SLG and OBP against.

   But, for example, just using Seth Smith’s career lgSLG and lgOBP numbers and plugging in his PH OBP and SLG of .413 and .554, I get a 152.6 OPS+ for him, which would probably be a decent estimate as well.

   Reply ↓
7. John Autin December 22, 2013 at 1:24 pm

   Doug, on a tangent re: searching PH splits — When I try to replicate your first split table with the career minimum at 2,000 PAs, I’m not getting Alex Arias, which is a clearly faulty result; Arias had 2,010 PAs, with no sign of any unknown splits (as expected for a career spanning 1992-2002).

   When I lower the threshold for career PAs, Arias does appear, shown with 1,897 PAs. The difference is 113 PAs, which happens to be his rookie total, when he batted only as a SS.

   Checking my results further, I found slight discrepancies in some other players’ career PA totals on the same search: e.g., Derrick May shows 2,324 PAs instead of 2,387, and Rick Miller shows 4,398 instead of 4,440. In both cases, the difference equals their rookie total.

   Have you noticed any such faulty results from the Split Finder?

   P.S. You said the split tables are since 1950, but I think if that were so, Ron Northey would not qualify, having only 161 known PH PAs since 1950.

   Reply ↓
   1. Richard Chester December 22, 2013 at 3:49 pm

      JA: I ran the PI and got the same results as Doug, except that Torgeson was listed above Anderson. Northey’s totals go back to 1946. The years 1946 to 1950 have incomplete results. His 1942 to 1944 PH stats can be retrieved from his batting game logs for those years.

      Reply ↓
      1. John Autin December 22, 2013 at 5:53 pm

         Richard, I do get Doug’s results when I choose “Find Career Totals” for the PH split, with 2,000 career PAs and 200 as a PH. But if I choose “Find Totals Spanning Seasons” for 1950-2013, my results omit Arias, and have those other PA discrepancies that I mentioned.

         Reply ↓
   2. Doug Post authorDecember 23, 2013 at 7:18 pm

      John, I said since 1950 because that is the earliest that PI claims that the event data are “mostly complete”.

      Northey gets picked up because, when running a career Split, P-I does not provide the option of limiting the data by season – you get everything P-I has.

      The table is generated based on selecting the PH within Defensive Positions split with “Set minimums by season totals …” checked off and PA >= 2000 specified as a career filter, and then specifying PA >= 200 under “Select additional criteria …”.

      Reply ↓
8. Richard Chester December 22, 2013 at 7:53 pm

   John: I am on a mobile device now so navigation is difficult. Briefly perhaps when running spanning seasons the PI ignores seasons without a PH appearance.

   Reply ↓
9. Richard Chester December 23, 2013 at 3:34 pm

   JA: This is to confirm my statement of post 22. I ran the PI for careers and for spanning seasons. I checked Carlos Baerga’s stats. The career PI run shows 308 PH appearances and 5895 total PA. The spanning seasons run shows 308 PH appearances but with 4710 total PA. The difference in PA is 1185. I then checked his career splits on BR.There were two seasons, 1992 and 1994, in which he had no PH appearances. His total PA in those two seasons was 1185. So doing a spanning seasons PI run counts only the years in which had had at least one PH appearance.

   Reply ↓