Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Primate Studies > Discussion
Primate Studies
— Where BTF's Members Investigate the Grand Old Game

Wednesday, November 27, 2002

And the Beat Goes On: Derek Jeter and the State of Fielding Analysis in Sabermetrics - Part 7

Mike celebrates Turkey Day with a big helping of play-by-play data.

Part 7: Details, Details: A Look at the Play-by-Play Data

 

So what does the play-by-play data have to say about
Derek Jeter?

Because of the earlier problem with the 1998 data noted by Chris Dial, I
limited this section to analysis of the PBP data for 1999 and 2000 only. For
this analysis, I defined the hole area as zones z56 and z56D, the direct area
as zones z6 and z6D, and the middle area as zones z6M and z6MD. In each zone,
I counted the number of ground balls that were either fielded by the shortstop,
or that went through to the outfield (excluding the handful of plays where the
outfielder actually made a play). I did not include plays made by another
infielder in the same area - thus excluding all plays made by 3B in the SS
hole as well as the handful of plays made by other infielders in the direct
and middle areas.

The results of this analysis are summarized in Table 16. I included the
summary for Jeter and for all AL shortstops, broken down by zones.

Table 16. Plays Made by Jeter in SS Zones, 1999-2000, compared to
AL Average

Inn GBIP Hole PM Hole BIP/9 %BIP %PM GBIP Direct PM Direct BIP/9 %BIP %PM GBIP Middle PM Middle BIP/9 %BIP %PM
Jeter 2674.3 275 105 0.93 29.1% 38.2% 316 286 1.06 33.5% 90.5% 353 223 1.19 37.4% 63.2%
AL 40217.7 4780 2075 1.07 28.1% 43.4% 5389 4894 1.21 31.7% 90.8% 6857 4365 1.53 40.3% 63.7%

As noted earlier, the distribution of GBIP around Jeter’s area
favors the SS hole; there were significantly fewer balls hit up the middle
than there were in the hole, as a percentage of the total. Note also that
Jeter was, as indicated, slightly below average in the percentage of plays
that he actually made in the direct and middle zones, and significantly below
average toward the holes. Finally, note also in that every assigned SS area,
Jeter had fewer opportunities per nine innings than the league norm -
significantly fewer in the case of balls hit up the middle. Had there been a
league-average number of opportunities in each area of the field, with all
else being equal (of course, it’s unlikely that all else would have been equal),
Jeter would have picked up about .4 more plays per nine innings.

Breaking the data down further, by LHB and RHB, we get Table 17.

Table 17. Plays Made by Jeter in SS Zones, 1999-2000, compared to
AL Average, by Batter Hand

vs LHB Inn GBIP Hole PM Hole BIP/9 %BIP %PM GBIP Direct PM Direct BIP/9 %BIP %PM GBIP Middle PM Middle BIP/9 %BIP %PM
Jeter 2674.3 47 16 0.16 19.5% 34.0% 75 65 0.25 31.1% 86.7% 119 82 0.40 49.4% 68.9%
AL 40217.7 854 296 0.19 19.2% 34.7% 1257 1092 0.28 28.2% 86.9% 2344 1652 0.52 52.6% 70.5%
                                 
vs RHB Inn GBIP Hole PM Hole BIP/9 %BIP %PM GBIP Direct PM Direct BIP/9 %BIP %PM GBIP Middle PM Middle BIP/9 %BIP %PM
Jeter 2674.3 228 89 0.77 32.4% 39.0% 241 221 0.81 34.3% 91.7% 234 141 0.79 33.3% 60.3%
AL 40217.7 3926 1779 0.88 31.2% 45.3% 4132 3802 0.92 32.9% 92.0% 4513 2713 1.01 35.9% 60.1%

Both LHB and RHB tend to hit fewer balls up the middle than
might be expected. Here we see that most of Jeter’s below-average conversion
percentage in the hole came from balls hit by RHB; he was much closer to
average against LHB, although there weren’t a whole lot of balls hit there
by LHB (about one every 6 games for Jeter), so standard sample size caveats
apply.

I also chose to take a look at the breakdown of plays made with a runner
on 1B, vs those without a runner on 1B. One might expect that the shortstop
would find it easier to make a play with a runner on 1B than when there is no
runner on 1B, on the theory that it’s sometimes easier to get a 6-4 force play
than it is to throw out a runner (especially when the ball is hit into the
hole). Since the Yankees had a very good pitching staff, one might also expect
that Jeter had fewer opportunities to make force plays than the average
shortstop might, eliminating some easier opportunities for plays. After looking
at the preliminary results from this analysis, I made a further division of the
1B situations into double play situations (outs<=1) and non-DP situations
(outs=2). Table 18 summarizes the results.

Table 18. Plays Made by Jeter in SS Zones, 1999-2000, compared to
AL Average, Various Baserunner Situations




















































































































































































































DP Sit Inn GBIP Hole PM Hole BIP/9 %BIP %PM GBIP Direct PM Direct BIP/9 %BIP %PM GBIP Middle PM Middle BIP/9 %BIP %PM
Jeter 2674.3 56 17 0.19 28.1% 30.4% 56 45 0.19 28.1% 80.4% 87 63 0.29 43.7% 72.4%
AL 40217.7 1023 353 0.23 30.0% 34.5% 936 839 0.21 27.5% 89.6% 1447 1054 0.32 42.5% 72.8%
                                 
RO1/no DP Inn GBIP Hole PM Hole BIP/9 %BIP %PM GBIP Direct PM Direct BIP/9 %BIP %PM GBIP Middle PM Middle BIP/9 %BIP %PM
Jeter 2674.3 30 11 0.10 51.7% 36.7% 28 27 0.09 25.9% 96.4% 50 29 0.17 46.3% 58.0%
AL 40217.7 580 270 0.13 28.0% 46.6% 624 566 0.14 30.1% 90.7% 869 575 0.19 41.9% 66.2%
                                 
no RO1 Inn GBIP Hole PM Hole BIP/9 %BIP %PM GBIP Direct PM Direct BIP/9 %BIP %PM GBIP Middle PM Middle BIP/9 %BIP %PM
Jeter 2674.3 189 77 0.64 44.9% 40.7% 232 214 0.78 36.4% 92.2% 216 131 0.73 33.9% 60.6%
AL 40217.7 3177 1452 0.71 27.5% 45.7% 3829 3489 0.86 33.2% 91.1% 4541 2736 1.02 39.3% 60.3%

The data contradicts the notion that SS should do better with a
runner on 1B than they do overall; in fact, they do better only when going up
the middle. Upon further reflection, these tendencies can be readily explained.
The shortstop is likely playing in and closer to the bag in a DP situation in
order to be able to cover. The move up has the effect of reducing the area that
the SS can cover, mitigating the effect of the possible shorter throw to make a
play. When going up the middle, however, the SS can get plays on flips to the
2B covering the bag where it would be difficult or impossible for him to turn
and throw to 1B, which gives the SS an extra range of plays to make; when you
throw in the fact that he is closer and has a shorter throw, the reduction in
range going up the middle is more than balanced.

Jeter actually saw more of his GBIP in DP situations (21.1%) than did the
average AL SS (20.0%). He did far worse in runner on 1B situations than he did
overall; he fielded balls in the direct and middle zones at a higher rate than the
league average without a runner on 1B, although when considering that he was
likely shading up the middle (as I discussed in part 6) that performance isn’t
quite as good. But the runner on 1B performance drags his overall fielding numbers
down.

I decided after looking at the runner-on-1B data to review Jeter’s fielding
in DP situations, to see whether there are secondary effects that might be
affecting those numbers. Table 19 provides a breakdown of Jeter’s numbers in
DP situations. I present the totals, the combination of hole+direct plays, and
the plays made up the middle (since Jeter was closer to the league average
there in DP situations). I broke the plays down into double plays (DP - if there
were any triple plays on GB they were included in this category), errors that
allowed a runner to reach base (Err), force outs at 2B (FO), plays on which
only the batter was thrown out (GO), hits (H), fielders’ choices that allowed
a runner to reach (FC), and plays where other runners besides the runner on 1B
were thrown out advancing (OA). The %Out column is the percentage of plays on
which an out was recorded (DP+FO+GO+OA/total).

Table 19. Plays Made by Jeter in SS Zones, 1999-2000, compared to
AL Average, in DP Situations




















































































































































































































all DP Sit Inn DP FO GO Err H FC OA %DP %FO %GO %Err %H %FC %OA %Out
Jeter 2674.3 58 49 18 7 65 2 0 29.1% 24.6% 9.0% 3.5% 32.7% 1.0% 0.0% 62.8%
AL 40217.7 1438 605 187 80 1072 7 17 42.2% 17.8% 5.5% 2.3% 31.5% 0.2% 0.5% 66.0%
                                 
Hole & Direct Inn DP FO GO Err H FC OA %DP %FO %GO %Err %H %FC %OA %Out
Jeter 2674.3 28 25 9 4 45 1 0 25.0% 22.3% 8.0% 3.6% 40.2% 0.9% 0.0% 55.4%
AL 40217.7 669 396 111 49 713 4 17 34.2% 20.2% 5.7% 2.5% 36.4% 0.2% 0.9% 60.9%
                                 
Middle Inn DP FO GO Err H FC OA %DP %FO %GO %Err %H %FC %OA %Out
Jeter 2674.3 30 24 9 3 20 1 0 34.5% 27.6% 10.3% 3.4% 23.0% 1.1% 0.0% 72.4%
AL 40217.7 769 209 76 31 359 3 0 53.1% 14.4% 5.3% 2.1% 24.8% 0.2% 0.0% 72.8%

In almost all of these situations, Jeter was very likely positioned
at double-play depth in the infield. 

Going up the middle, Jeter was actually getting to more balls than his shortstop
peers in the AL, but making fewer plays because of a higher error rate. That would
be consistent (again) with the idea that he was shading in that direction. What’s
striking about this data, however, is how few DPs Jeter was able to get from those
plays, and how many FOs and (especially) GOs resulted. The up-the-middle data is
especially interesting - Jeter had about twice and many FOs and GOs as did the
average shortstop, and only about 2/3 as many DPs. The GO rate (and to a lesser
extent the error rate) suggests that Jeter was passing up opportunities to take
the easier throw to 2B in favor of the longer throw to 1B. It’s not particularly
clear why this might have been happening - the PBP data doesn’t suggest anything
one way or another - but I did note that Jeter had a relatively low rate of
unassisted putouts at 2B, which suggests that part of the problem might
be that Jeter was waiting for his 2B counterpart to cover rather than taking plays
that he could make himself.

With Jeter at DP depth, he had his usual problems going into the hole, but he
was also strikingly bad at handling balls hit directly at SS. This was mostly, as
far as I can tell, a 2000 problem; his 1999 out rate was a lot closer to average
(he had an 86.2% PM rate in 1999, a 74.1% rate in 2000). He still wasn’t turning
DPs in either case, though.

Finally, let’s take a look at the same breakdown in DP situations on the 2B
side of the diamond, comparing how Jeter’s DP partners did to the performance
of AL 2Bs. That data is presented in Table 20.

Table 20. Plays Made by 2Bs with Jeter at SS in DP Situations, 1999-2000, compared to
AL Average









































































































































































































Result DP FO GO H Err FC OA %DP %FO %GO %H %Err %FC %OA %Out
Jeter 58 39 32 60 11 0 0 29.0% 19.5% 16.0% 30.0% 5.5% 0.0% 0.0% 64.5%
AL 1169 556 422 1005 110 4 2 35.8% 17.0% 12.9% 30.8% 3.4% 0.1% 0.1% 65.8%
                               
Hole & Direct DP FO GO H Err FC OA %DP %FO %GO %H %Err %FC %OA %Out
Jeter 38 32 31 42 10 0 0 24.8% 20.9% 20.3% 27.5% 6.5% 0.0% 0.0% 66.0%
AL 703 411 385 732 73 2 1 30.5% 17.8% 16.7% 31.7% 3.2% 0.1% 0.0% 65.0%
                               
Middle DP FO GO H Err FC OA %DP %FO %GO %H %Err %FC %OA %Out
Jeter 20 7 1 18 1 0 0 42.6% 14.9% 2.1% 38.3% 2.1% 0.0% 0.0% 59.6%
AL 466 145 37 273 37 2 1 48.5% 15.1% 3.9% 28.4% 3.9% 0.2% 0.1% 67.5%


I think it’s pretty clear here that the 2Bs on the Yankees
were shading toward the hole on the right side; the number of hits that go
through the middle, compared to the number that go through the hole, support
that assertion. The Yankee 2Bs were getting to more balls than their
counterparts when the ball was not hit up the middle on the 2B side. Here
again we see a similar pattern - fewer DPs, more force outs and ground outs,
more errors on the balls hit in the hole and directly at the 2Bs - and
certainly these plays were likely to involve Jeter in the pivot. On the
up-the-middle balls, on the other hand, the 2Bs actually turned a few more DPs
and force outs and had a few less ground outs and errors on balls that they
actually fielded (realizing that, here again, not that many balls went through
the z4M and z4MD zones in DP situations).

I think that the PBP data allows us to draw the following conclusions:

  • there is no possible construction of the data that would support Jeter as
    being significantly better than average defensively;
  • the PBP data suggests that Jeter is weak convering the hole, but this
    deficiency may be attributable in large part to positioning;
  • if Jeter is weak going up the middle, as analysts have suggested, that
    weakness is also mitigated by positioning;
  • Jeter is more of a liability defensively when there is a runner on 1B;
  • Jeter’s one clear defensive weakness is in turning the DP, whether feeding
    or turning the pivot.

    Finally, just because Tango asked for it, even though it has nothing
    directly to do with Derek Jeter, here’s one last table:

    Table 21. Distribution of Errors by Ball Location, AL SS, 1999-2000



























    Location BIP Errors SIZE=2>%Err
    Hole 8376 126 SIZE=2>1.50%
    Direct 7336 208 SIZE=2>2.84%
    Middle 8820 157 SIZE=2>1.78%


    Mike Emeigh Posted: November 27, 2002 at 06:00 AM | 15 comment(s) Login to Bookmark
      Related News:

  • Reader Comments and Retorts

    Go to end of page

    Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

       1. tangotiger Posted: November 27, 2002 at 02:07 AM (#607424)
    Ahhhhh! That was a satisfying climax to the series. This is exactly the kind of breakdown that defensive analysis should aspire to.
       2. Joey Numbaz (Scruff) Posted: November 27, 2002 at 02:07 AM (#607425)
    Great job Mike.

    Now the question is, when will we get data like this for every player at every position, in a nice easy to read book . . . :-)

    I'm not sure if this is in part VIII or not (roman numerals seem appropriate for this series, no?), but how can we use this PBP for those of us that want to go back historically, based on the traditional stats (adjusted all over the place of course) and evaluate fielders accurately? What I mean is, can we construct a system based solely on the traditional stats, and adjust it so players rate similar to how they do w/PBP data?
       3. jeff angus Posted: November 27, 2002 at 02:07 AM (#607429)
    Lovely work. Thank you for this series.
       4. MGL Posted: November 27, 2002 at 02:07 AM (#607431)
    To Joe and Art,

    Basically, for any year in which you don't have PBP data, you can use other PBP data to estimate how many balls are hit into each player's "zone" (however you want to define a player's zone), based on the total estimated number of ground balls and fly balls, and the estimated percentage of RHB and LHB's faced when that player is on the field. Then you can use an infielder's assists and some of his putouts (by estimating unassisted GB outs) to come up with a ZR, and OF'ers putouts for the same.

    Or....

    You can do essentially the same thing as above by using the PBP data to see how many outs a SS is "supposed" to make given the total # of GB's hit when a RHB is at the plate and when a LHB is at the plate, and then use these numbers to see how many outs a particular player "should" have made compared to how many he did make.

    This is not my idea BTW. It was already formulated by "I forgot who" in a previous article on "context-adjusted fielding stats" or something like that. Maybe that was even by Mike - I don't remember.

    I think that is about the best you can do applying "other year" PBP results to years and players in which no PBP data is available.

    So basically, all you need as far as the PBP results are concerned, is the following:

    For each defensive position:

    What % of fly balls are fielded (turned into outs) by the LF/CF/RF, on the average, versus a RHB and the same versus a LHB. Alternatively, what percentage of fly balls are hit in a LF's/CF's/RF's "zone" versus RH and LH batters.

    Same for GB's and the infielders.

    Once you have that "one-time" information, you can create rough ZR's for players when you don't have their PBP info available. These rough ZR's will be better than any other non-PBP defensive metric, because right now, all of these metrics (RF, Palmer's defensive runs, FA, etc.) necessarily use only the number of balls fielded by a fielder without regard to the number of balls available to be fielded...
       5. MGL Posted: November 27, 2002 at 02:07 AM (#607434)
    Art P, you present some interesting and valid points. Sure, you could break down the PBP data even more in order to apply it to non-PBP data, as you suggest...

    Your point about perhaps regressing FR before adding it to BR is an excellent one.

    When people present sample results, and then pass them off as estimates of ability or even if they don't and other people use those results to maker inferences about ability, they often forget about at the very least mentioning the need for regressing.

    I can live with not regessing sample results, as long as we are consistent. We don't want to say in one breath, that Vlad Geurrerro is a great hitter because his career OPS is .920 and then in another breath say that he is a great hitter because his career OPS is .908 - oh and BTW, that "career" OPS of .908 is really his actual career OPS of .920 regressed!

    When we combine metrics that have the same currency, like batting and fielding runs, however, we do run into more serious problems of regression, as you mention. This is not a good thing! Let's say, for example, that we were confident that batting lwt runs (BR) had a pretty low measurement error for some decent sample size, so that we didn't feel an urgent need to present "regressed" BR's. Let's also say that our fielding runs (FR's) had a very large measurement error such that they had almost no correlation with actual fielding skill. That could mean that each player's sample FR's came out exactly the same, that they were different but random, or some combination. Let's say that it was the second alternative, that each player's sample FR's were differnt but random (had no relation to actual fielding skill). What happens if we combine each player's BR and FR? Disaster! Utter and complete disaster!


    Well, what if Palmer's FR's (or some other defensive metric) are not correlated very well with fielding skill, which they probably arent? What happens when we combine BR and FR, as Palmer and others doe all the time? Semi-disaster! Yes, it is true that you don't want to combine metrics even if they have the same currency, IF THEY HAVE SUBSTANTIALLY DIFFERENT MEASUREMENT ERROR (REGRESSION COEFFICIENTS). The correct way to combine the metrics os to "normalize" the regression first (regress each one and then combine, or regress the one with the most measurement error, don'e regress the other one, and then combine. This is the basic idea, BTW, with regressing a player's component stats individually and then combining them, in order to get good projections, rather than computing a combined stat, like lwts, OPS, or BaseRuns, and THEN regressing the final number in order to get a good projection...

    Great insight Art!

    Have a good T-Giving guys! I'll be off-line for a few days.
       6. Mike Emeigh Posted: November 28, 2002 at 02:07 AM (#607440)
    how does the PBP data lead to the inference that it is Jeter and not the 2B that are the cause of any weakness on DPs

    I touched on this in a comment to an earlier article, and probably should have gone into it in more detail here. Yankee 3Bs were average in the rate of DPs turned on GBs fielded over 1999-2000 (44% vs a league average of 43.1%). Since almost all of these are 5-4-3 DPs, one would expect that, if the 2Bs were a large part of the problem, the 3B rate would also be low.

    -- MWE
       7. tangotiger Posted: November 28, 2002 at 02:07 AM (#607441)
    ...one would expect that, if the 2Bs were a large part of the problem, the 3B rate would also be low.

    We would expect the 3B rate to be lower thanit would be with an average 2B. Brosius/Ventura are excellent fielders, and so, I think that the 2B did drag them down to average.

    Jeter might have been dragged down from average (or below) to much below average.
       8. MGL Posted: November 28, 2002 at 02:07 AM (#607442)
    Yes, one really has to control for various variables (read "pther players") in order to isolate cause/effect relationships in these types of analyses. I'm in the process of attempting to do that with my UZR which is one reason why I haven't redone my methodology and results yet, as I said I would. I'm stuck again on lots of things, including how to handle ROE's (should I treat them as missed balls in UZR or should they command a higher weight even in UZR, etc.)...
       9. Charles Saeger Posted: November 30, 2002 at 02:07 AM (#607448)
    Mike looked at the second basemen issue a bit. Knoblauch was not the problem with DPs, looking at 5-4-3 dps.
       10. Mike Emeigh Posted: December 02, 2002 at 02:07 AM (#607452)
    Knoblauch was not the problem with DPs, looking at 5-4-3 dps.

    In fact, in the 2000 season (the only season that I evaluated in which other players besides Knoblauch got significant time at 2B), Jeter's DP conversion rate was actually higher with Knoblauch than with the other 2Bs that the Yankees used (Bellinger, Delgado, Sojo, and Vizcaino). The Yankees got double plays on 30.2% of the GBIP in the 2B and SS vicinity in DP situations with the Jeter/Knoblauch combo, 28.7% when Jeter was paired with another 2B.

    -- MWE
       11. Mike Green Posted: December 02, 2002 at 02:07 AM (#607453)
    Great series, Mike E. One fairly easy way to check the defensive capabilities of infielders on ground balls is to compare performance (i.e conversion of ground balls into outs) of 3 sets of pairs (1b,2b), (2b,ss)and (ss,3b) with league averages. So, if Jeter is shifting towards the middle, the (1b,2b) or (2b,ss) rates for the Yankees should be better than expected (depending on whether the second baseman also shifts or remains at the usual positioning). A third reference point is the overall team conversion rate on ground balls.

    Ground balls are much easier to deal with than popups because one is only dealing with 2 players who have the realistic opportunity to make a play. It is hell to deal with the short fly-ball/popup in the Bermuda Triangle where 4 or 5 players may have a legitimate shot at it depending on their positioning.

    A similar approach is useful in dealing with outfielder's range, but with 2 pairs (RF,CF) and (CF,LF). The purpose of using the pairs is to find out if an individual's conversion rate is affected by unusual positioning.

    It seems to me that in evaluating defence, it will not be practical to tell the whole story with a single stat. Rather 5 or 6 stats will paint a better statistical picture, whereas in evaluating offence 2 stats will do (OBP and slugging-OPS being a mixture of apples and oranges). You have shown the lead in identifying what 3 or 4 of those defensive stats might be for a shortstop. Well done.
       12. Mike Emeigh Posted: December 03, 2002 at 02:07 AM (#607457)
    One fairly easy way to check the defensive capabilities of infielders on ground balls is to compare performance (i.e conversion of ground balls into outs) of 3 sets of pairs (1b,2b), (2b,ss)and (ss,3b) with league averages.

    DFT and CAD do this with middle infield putouts now, and while I don't know if Clay still does the pairing on assists, he has done it in at least one earlier version of the DFTs for SS/3B and 1B/2B. I have had some thoughts that we might be able to use combined assists to get a better handle on the location of *hits* in play (which are in my initial draft of the final article in the series). Since the overwhelming majority of GBIP that go for hits are singles, if we can get the number of singles allowed we can get a boundary for infield opportunities, and if we can figure out where those singles go we can get a boundary for positional opportunities. Combined assists might be one way to get us there.

    It seems to me that in evaluating defence, it will not be practical to tell the whole story with a single stat. Rather 5 or 6 stats will paint a better statistical picture, whereas in evaluating offence 2 stats will do (OBP and slugging-OPS being a mixture of apples and oranges).

    MGL splits out arm (for OF) or DP (for IF) runs from UZR runs. That's a necessary step, IMO - you need at least those two stats separated. Infield putouts should probably be separated from infield assists as well, because they generally represent very different kinds of plays. In CAD, Charlie also treats error rates separately. It probably makes sense to look at least at *range* (assists/opp for IF, putouts/opp for OF), *advancement prevention* (DP rate for infielders, assist rate for OFs, with an estimator for extra bases taken by runners tacked on if we can come up with one in the absence of PBP data), and error rates. Independent putouts are very important at 1B, because most of them are GB3U. James noted that putouts weren't very important at 3B, and I'm inclined to think that they aren't very important at 2B or SS either.

    -- MWE
       13. Mike Green Posted: December 03, 2002 at 02:07 AM (#607459)
    To decide how important putouts are, we need more information by position. For instance, putouts by a third baseman reflect 5 types of plays in my opinion:

    1. the line drive
    2. the popup in the infield (where the IF fly rule is called, or would be called had there been a runner on first and less than 2 outs)
    3. the looper over 3rd
    3. the popup in foul territory and
    4. the play at third on a runner

    The number of plays of type 1 and 3 clearly reflect on defensive ability, the number of plays of type 2 clearly does not reflect on defensive ability, the number of plays of type 3 and type 4 may very well reflect on defensive ability (in parks with lots of room in foul territory, nos. of foul pop-ups caught reflect on a third baseman's range). I have no idea what the proportion of putouts by a third basemen fall into each type.

    All I do know is that the average third baseman has between 1/3 and 1/2 the number of putouts as assists. As a guess 1/2 of those putouts may be reflective of defensive ability. So, my best guess is that number of putouts may have about 1/6-1/4 the impact of number of assists in context. This is important enough to take into account in my view.

    For shortstops, putouts fall into the same categories as for thirdbaseman, but the numbers in the final category (the play at second) are of course much higher. In my view, most of these plays do reflect on the shortstop's defensive ability (whether the play goes 4-6, 4-6-3, 1-6, 1-6-3, 3-6, or 3-6-3). As well, on average shortstops have over 1/2 the number of putouts as assists. I would estimate that putouts have about 1/3 to 1/4 the significance of assists for a shortstop in the context of opportunities.

    Obviously, statistical data on the breakdown of the proportions of putout by type for each position is required to better assess the importance of putouts for that position.

    By the way, I loved the Briles story.
       14. Charles Saeger Posted: December 03, 2002 at 02:07 AM (#607460)
    Putouts by third basemen:

    About half of these are popups, and something like a third are line drives. (I have an e-mail from Mike Emeigh somewhere giving a breakdown.) They do correlate positively with outfield assists (as do independent putouts by catchers), they correlate negatively with doubles (as do independent putouts by first basemen). Bill James wrote they correlate positively with flyouts (PO-A-SO) and balls in play against a right-handed pitcher, and negatively with wins (probably due to the outfield assists). I am sure they correlate positively with foul territory, and I know they correlate negatively with putouts by a left fielder, who is grabbing more and more plays from the third baseman as time moves forward.

    Bill wrote much about putouts by third basemen in Win Shares. His essay on this is top notch, FWIW. Bill looked at this many ways , and could make little sense of it. There may well be something there, and I suspect there is, but right now, there are better things on which to work.

    Putouts by middle infielders:

    Clay Davenport does a good job with these. Anyone who chats with me at all knows I have been working much on these, and may send Dan Szymborski an article I have written about these (and first base putouts) once Mike and I can work some on the PbP data.

    Many putouts by second basemen are throws from third base and shortstop, so they correlate positively with assists by these fielders and runners on first base (needed for a play at second base). They correlate positively with balls in play against a left-handed pitcher, mostly because there are more groundballs to the third baseman and the shortstop, and some because batters tend to hit flyballs to opposite field (this tendency stays when I take out the fielder's choice plays). They compete some with putouts by pitchers. Of the rest, some are line drives, some are popups. I'll check to see if they make more independent putouts with a runner on first base, since the first baseman covering the bag makes the second baseman play in the hole.

    Putouts by shortstops are the most pure, you are right. Some are fielder's choice plays by the second baseman, and this is why they happen a little more often with a runner on first and with a left-handed pitcher on the mound. They correlate positively with those Bill James net first baseman assists (A.1b-PO.p), so much so (it's on my home computer, but I recall r=0.40) that I now call those plays 3-6 assists. They are the most "pure" (meaning "independent," or "fewest came on a throw") of the infield putouts other than third baseman putouts.
       15. Mike Posted: January 03, 2003 at 02:16 AM (#607995)
    Mike says:

    "The up-the-middle data is especially interesting - Jeter had about twice and many FOs and GOs as did the average shortstop, and only about 2/3 as many DPs. The GO rate (and to a lesser extent the error rate) suggests that Jeter was passing up opportunities to take the easier throw to 2B in favor of the longer throw to 1B. It's not particularly clear why this might have been happening..."

    Wondering if this might have to do with the situations. They Yankees have been very successful during the Jeter years, and I wonder if there were more cases of Jeter attempting to secure the "sure out" at first rather than attempting a double play, because the Yankees were leading by a certain number of runs or simply because of the confidence that not risking a big inning gave the Yankees a much better chance of winning.

    Also wonder, with a bad defensive catcher like Posada, if more runners were on the move from first when the balls were hit against the Yankees, then the league average.

    You must be Registered and Logged In to post comments.

     

     

    << Back to main

    BBTF Partner

    Support BBTF

    donate

    Thanks to
    Phil Birnbaum
    for his generous support.

    Bookmarks

    You must be logged in to view your Bookmarks.

    Syndicate

    Page rendered in 0.9130 seconds
    47 querie(s) executed