Loading ...
jarmenia2
Other
17
0
Try Now
Log In
Pricing
", -', \ .-" < " ':-<1 Wi th this issue, The Baseball Analyst, a.fter fOUl;'" ·:y.ea.rs of talk; 1 • .. . J. and promises, finally gets off the oouoh..An outstanding artiol~ by Paul Sohwarzenba.rt detailing the effects ot;tyarious.; ballparks on ;th~ /' 'r.: produotion of errors c.nd other fielding oocurwmes 'has,:a.nd':"Ode-serves,·.. , the distinotion of being our first artiol~,:. P~ul's -a;rtioi~ a~m-o:n;,;; :~:! 1;':1:.' stra.tes tha.t fielding statistio.s, like ba.ttin.g and 'pi tohiig;s~~iiti:-(is~'; l· but apparently even more so, ,ar.e th~ prq.ducts in :9~t"of ott~(hii'rrst~ft5es·- i as well as men.. Following that, Dallas "Adams.":exami-n,es .th~~·gt~$':: :;;;). tion of runs soored by teams aoross· a wid,,~~a~e o.~··'offe!il.t.~";;?li:tI~-:;;;;'·.'l~~··"·; ties, and emerges with a ohart that will·.~t,ell you,:' 'if yo-ci')2:itii. how"',':;:;'::" j; many runs per game your team soores,~' how 'of;~en theyt',fJhou!~"b-~~~hut ;':)~f. ' out, how often they should soore l.~~n, 2 runs "--;"0 ·:ori."g1>'!'t9:·;l.95~:r.;·: .-:. more runs. This article may be::ac li~t:tlJ~~ mQ~Et·:.d±rfioult ,¥o"~1:ij-~d>1;banil1lt .. ' the first, but make a note of i t,L as. this is Q-)le ··a.rtiole:·~;l,1~ "'~'t~~ .. :n r..'.~ in whioh is potentially of enormOU:;3~ lJ'Se'i.:n:'·~s~a·r·<thiiig anY.iW3:fltf:ziCU'~dB:·'i: , • -.' ~ I- " ' , of issues, some of whioh Dallas suggests a. t tA~ oon..olusion of ~fie.,;. , t ' I ... ""'''.',"",''''':'.,,\.- ... '~' . ar loc e.. ", .':;::. ~ ... ~::;.:.:~ ... <..:.~./:.'!. ~~~~<~ .OJ Tom Jones on page 11 puts on reoord a f,~1i' fa,c,ts"iif9rlh knowing" . about 'Nolan Ryan's fifth career no-hitter.. Taking: .. a:·:b'r~aJ.c from th~: releIJ,tless march of informationt ZlIark Pankin propoge~£f·i'~'S;gt~m:of·· :;.1.= ... ; -rlins and Losses for all players, and;.;li'E;?;·p;z:o:esent-a. 'oo;nt:f±b~tion ".. , ..... t i ll:' from the la.te Bob Kingsley seeking to examine th~ 'fnipaot,of:'player of a tti tudes on run production in var.i.Q.uf5 pa.rks.. "'.... :. ' ~ Baseball Analyst begins wi th. '38.:"r~9de,~s .~:nd;~npp:~h in,'ater-: ~, ial on hand f'or three more issues .. ,.One thing,., .. and.',.on.Ef;;t~~rl8"qnlYt.' .~ iiill determine whether or not we can cq:qtinue,·from· tfuit.rtiEit£8&i4: ~:.::. ~'::. ~.t. , beginning to provide a place where peopl~"who··:,~?iv'e,:'.:~es~~~,)·lia.t;:, .. l,tt ~' they want to do can find a place to print it .. ,-:. Reaaers?,.;;·~-re"na;li":!'::-, live :wi th vem or wi thout 'em.. This ,.i.s,.aQ.i~cus·~fo:r;~:.~6~.?f~'enct~~; '" ~. not a oommeroial opera·tion.. 'ivhat ~i·s·impQ.~ta~t ·ifi:'w~etb.ei-·Jor.';i~oi:n .~. .~ you contribute to the discussion~ . linen 'yoli.:ba.ve a'i(li~i;; ~WKenr~ .. ·~:n:iSt. , you do some research, write it up and ·send·, ft 'to- us: .. " j:i y.oii 'do .!o.t ", :,' ~' that, .the Analyst will thrive and-grqlfj. if;~,Y'Oiita:O~:"'t'~":·,:i"t,~~ji~t-b.;,."'·": Iii t her a.nd fold.. It's up to you.. "-. '. '( ,J.l.';- .J C'" a.CO :'~ .. :;.!.~ ":~)~ t·1;"·"1·"il1 .j Bill James 8allpark Effects .. August I, 1981 BALLPARK EFFECTS ON THE PRODUCTION OF INFIELD ERRORS AND DOUBLE PLAYS Since the introductiori of the morlern astroturf ballpark, th~re has been some controversy about the effects of astroturf. ThF.! rparh'?r has dOIJbtless ,been exposp.ri to many of the argumr::nts, fnr instancel' 1. Thp. ball travels faster on the astroturf, thersfore, infielders are forced to play deeper; 2. Accepti~g ·thii jiist premise. people argue ~hat astrolurf either raises, lowers, or has no net effect on batting averages (probably the most commonly accepted statement is that it raises averages 10- 20'points); 3. That because astroturf gives a "tr~e"bounce, infieiders will commit fewer errors playing ph astroturf than on natural turf. This discussion of the differences between astroturf.and natural surfaces may lead the baseball ohserver ,- . . ' to wonder if there isn't also some significant degree of variation among the playing surfaces of these ceneral classifications. After all, astroturf surfaces get harrier with age as the surface gets !ijorn' and the ground unriF.!rneath !]2ts pounded into l,!:?Llr-nt from tha l~ck of natural drainaJe and plant roots. LikR~ise, not all ~a~ural surfaces arR exactly alike. They have diff~rent soil chara~teristics; they get differing degrees of care; some ~ulti-purpose stadiums with natural surfaCES are harder to kEep in s~ooth condition. Several years ago I decided to test some of these statements by 'collectinlJ statistics from 'the box scores in the :iport.ing ~,lel!Js -in order to perform some home/road analysis. In those 'days T .. had more ·availablp. time to IIlork on the statistics col1ection and ... _/- Ballpark Effects !l.!as ahle to compile 10 team/years warth of data. This article" (;<;) stJr.1marizes my findinlJs on National League ballparks for infield errors and double plays. The statistics compiled are from the following yearsl 1. For the years 1973, 1976~ 1979 and 1980 I compiled comparative data for each team, home and away, and fbI' its 'oppbnent~ playin~ in the original team's park- and at home; 2. For the years 1972 and 1978 I compiled comparative stats. for only fbr each tea~'s performance at home and on the roarl. For example, for 1973 T have the Cubs' performaoce:at Wrigley Field compared to on the road, and the performaoce d~ Cub~' opponents at Wiigley Field compared to the totals"nf:C~hs' opponents playing at home in Cubs' road games. For 1t;172, I have only the Cuhs at ~tigley Field compared to the Cubs en the roari. The statistics which are compiled here are infield errorS; outfield errors (th~resu1ts'of which s~em intbnc1u~ive'and not worth much further discussion) and double plays. The numbers used for comparative purposes are infield eirors, outfiEd Errors and double plays per thousand plays. Plays are figured by multiplying Innings Pitched times three and suhtract~ng strikeouts. The next step in ma~ing the compariso0 o£ performance is to divide the "home-park" statistics by the"road-par~~ stats. The result is a ranking of ballparks by their amount o.f yariation from the "average" of all other ballparks other than th~~ome park. The idea is to measure the difference in production, for example, of infield errors in Wrigley Field in games between the Cubs and their opposents, as opposed to the produciio~ of infield erros in Cubs' games on the road. -. 'i ....... "' ... /:: . ., I will again use my favorite ba11team. the Cubs, as an example of how I produced the comparative statistics; 10Ye ar I ( I P X 3 ) Team Totals K DP EOF ((IPX3)-K) LUrig1ey Road :: 530 17717 21916 21740 :: 29.91 4199 4346 806 745 Road EIF 1000 M14 484 :: 17394 124 123 17717 = 27.83 29.91 27.83 The result of the comparison is that the probability of an infield error .in a Cubs v. Opponent game is about 7.5% greater if the game is in Wrigley Field than in the "~~erage opponent's home stadium." The complete final data follows: Team CHI *mON-J MON-O NY PHIL PITT SL ATL CIN HOU LA so "llSF-A SF -t·J EIF 29.91 32.17 27.49 27.91 24.08 28,43 27.61 34.q1 23.90 27.06 28.29 32.3R 35.16 35.32 Home 7.00 7.92 8.64 7.56 6.02 8.89 6.92 8.03 6.16 5.14 5.05 6.07 7.26 10.03 OP 45.49 47.44 38.49 41.39 40.88 40.30 46.35 43.28 38.58 41. 39 42.07 40.60 38.03 39.56 (per 1000) 27.83 31.57 31.06 29.16 27.92 28.33 29,84 29.73 29.30 28.94 29.82 30.09 34.48 29.95 Road EOF 7.07 -6.15 7.09 5.81 7.44 ·8.16 6.68 7.40 6.19 7.47 6.22 7.06 8.14 6.28 DP 42.83 43.42 39.62 42.70 43.83 45.05 42.01 39.33 43.32 40.51 41.65 44.27 39.71 38.72 * Stats from 1972, 1973 and 1976 are for old Parc Jarry; the MON-O stats represent those for Olympic Stadium. years 1978, 1979. 1980. ~ Candlestick Park h~d an astroturf surface in 1972 and 1973, before the city fathers decided to restore the natural surface. 4· :: 1.075 Ballpark Effects The remaining statistics will be listed in order of rank. (:\~ For purposes of the astro/natural argument, an asterisk will ". \.- follow the astroturf park figures. The comparative figures follollJ: Home IAway Rat ios Rank ,Park ErF Park EOF Park OP 1 crN .816* HOU .688* 5L 1.103* '2 PHIL .862* PHIL .809* ATL 1.100 3 ffON-O .884* LA .B12 mON-J 1.093 4 SL .925* SO .860 CHI 1.062 5 HOU .935* SF-A .892* SF -N 1.022 6 l.A .'=349 mON-O .937* HoU 1. 022* 7 NY .957 CHI .990 LA 1.010 8 PITT 1.004* cnJ .995* mON-O .971* 9 MN-J '1. 019 SL 1. 036* NY .969 10 SF-A 1. 020* ATL 1.OB5 SF-A .95B* 11 CHI 1.075 PITT 1.089* PHIL .933* 12 SO 1.076 mN-J 1.288 SO .917 13 ATL 1.174 NY 1. 301 PITT .895* 14 SF -N l.l7g SF -N 1.597 C~J" .891*' l}Avera'1es: Natural 1.061 1.133 1.025 Astroturf .921 .• 921 .968 ~ The averages are simply the means of the figures in the above tables and are not weightpn by either plays. or, in the cases of montreal and San Francisco, by the number of team/years of data. They are put out simply for purposes of discussion, and the extra work of going beyond the simple averages did not seem worth the ffort considering the fact that they are not being used for any purposes other than approximation for discussion purposes. DISCUSSION-CONCLUSIONS: The most obvious conclusion is that the statement about fielding percentages being higher on astroturf is clearly justified by the evidence. However, it is also apparent that there is some substantial variation among the ballparks in the different classes. Atlanta and San Francisco appear to be the banes of National League infielders, and any infi~lder who 5 ...... _'. - .. :1 all ["lark C::ffects expects to win a gold Q10ve while paying half of his games in ei ther place. I}Jhether justi.fied or not, infield gold gloves tend to he Bwardad mainly on fielding percentage (other factors seem to include hitting, range, and o~erall subjective evaluation). I can remember no Giant or Brave infielder winning a gold glove or Sporting News Fielding award in the past ten years, in fact, ~ave Johnson went from perennial American Gold Glove second baseman in ~l al t. imore to a 30 error man in Atlanta, resulting in his endinQ his career as a first baseman (although his deteriorRting range was probably as important a factor in that move). Although the difference between natural and astrolllrf for outfield errors is l?rger than the difference for infield errors, I am less ready to make the obvinus conclusions that the turf itself is responsible for the difference. The absolute numbers of errors are smaller for the outfield errors, so the deviation is actually less significant. IntUitively, the trueness of a bounce seems less significant after the ball hag travelled to the outfield. More likely, the difference in surface comes into play more on throws, since it is usually the player making the throw wh6 is charged with the error when thE'! hn] 1 takes C3. bad hop. For It!hat it's wort.h, it's also true that fiplding percentaQe dOGS not correllate well with the C3.lUard ing of gal d n loves for out f ie 1 ders-Lui tness Dave Parker. Finally, the differences noted in double plays are smaller, leadinQ to the likely conclusion that,whatever the effect that astroturf has on the production of double plays, the effect is le~s sig~ificant then it is for the production of errnrs. The fact that the second baseman and shortstop must play deeper on as troturf Luill prohah 1 y tenr! t C1 reduce the pr obab il i ty of a 6 Flallpark Effects double play; on the other hand, the fact that the ground ball travels faster will tend to increase the probability of a double play. These offsetting factors seem to about cancel each either out. The size of the difference in average and the distribution of results is such that I am not willing to draw any conclusions about the net effects based on this amount of data. ~al1park effects have been shown to be significant on hitters' statistics; it is sensible to expect that the same is true for fielding statistics. Certainly, playing surfacEs are one important difference amon~ the'various ballparks. Others may also he important, fof instances, the swirling winds at Candlestick Park may be as responsible for producino infield errors as the grQund (reputed to be fraught with tiny pphhlES ann clay clumps), although distinguishing among the causal factnrs is probabJy impossible with only numbers avai13hl~. Subjective comments from readers knowl~dgable about your lncal b~l]park would be appreciated, if there are any readers nf this maries tart i cle. Should thE Analyst' ~et off thF. ground, r hope to have .!\mer ican League ball parI- rlata summar i ze ri for thE m~x t r:dition. 7 THE DISTRIBUTION OF RUNS SCORED ( .' by Dallas Adams .. How often will ·a team be shutout? How often will it score exactly one run? Or ten runs, or twenty? What, in short, is the distribution of runs scored in a game? Obviously the distribution will vary between high scoring and low scoring teams; and it will vary, partly due to the "shortness" of the 162 game season, between equivalently scoring teams, although this difference will be less than that between high and low scoring teams .. The approach utilized for this study was to group together equiv- alently scoring teams in order to determine, for each group, an average distribution of runs scored. Since 1968, the Official Baseball Guide has listed, for each team, the -scores of all that team's games the previous season. This furnishes ten seasons, 1967 through 1976, of scoring data; a total of 232 team-seasons. First9 each team was grouped with equivalently scoring teams. the grouping being made on the basis of average number of runs scored per game. In all, there were 11 different groups; each having its own distinct range of average runs per game, as shown in Table 1. ' TABLE 1 Group Range ot Number of Group average number A verage Runs teams in runs per game i I per game for the group the season I 1 2.75 - 2.99 6 2 .. 912 I 2 3.00 - 3.24 14 3.183 ! 3 3.25 - 3.49 23 3.371 4 3.50 - 3.74 26 3 .. 612 i 5 3.75 - 3.99 49 3.864 I 6 4 .. 00 - 4.24 40 4.122 • 7 4.25 - 4.49 34 4.348 8 4 .. 50 - 4.74 24 4 .. 599 9 4 .. 75 - 4.99 13 4.867 10 5.00 - 5.24 2 5.157 11 5.25 - 5.49 1 5.290 For example, the 1967 Cincinnati Reds averaged3.73 runs per game t hence they belong to Group 4. The next step was to tabulate, from the Guides, the number of times each team scored each specific number of runs. Once this was completed for all teams, the totals within each group Were converted to percents. That is, for each group the percentage of times it was shutout was computed, likewise the percentage of times it scored exactly one run, etc. These percentages are shown in Table 2; this table presents the complete grouped information and forms the basic data set tor the remainder of this study .. The inherent behavior of the basic data can be easily seen on Figura 1 which presents for low scoring, average scoring and high scoring teamb the probabilities of such teams scoring an exact number of runs. Note that the solid line on this figure represents the average run scoring 8 EXACT GROUP NUMBER . 1 OF' RUNS 0 14.52 1 18.91 2 17.03 3 16.41 4 10.97 5 9.09 6 5.02 7 2.72 8 1.78 9 1 .. 36 10 1.25 11 0.31 12 0.31 13 0.21 14 0.00 15 0.00 16 0.10 17 0.00 18 0.00 19, 0.00 more T.HE DISTRIBUTION OF RUNS SCORED TABLE 2 THE PERCENTAGE OF GAMES IN WHICH EXACTLY THE GIVEN NUMBER OF RUNS ARE SCORED (ARRANGED BY GROUPS) GROUP GROUP GROUP GROUP GROUP GROUP GROUP GROUP 2 3 4 5 6 7 8 9 . 11.42 9.79 9.98 8.11 7.21 6 .. 49 5.93 4 .. 75 16 .. 90 15 .. 35 13.72 12.90 11.54 10.32 10.00 7 .. 60 17 .. 64 17 .. 79 16.28 14.53 14.29 13.37 12 .. 10 12.11 15.81 16.79 15.13 15.73 14.69 14.58 13.34 14.68 12.22 12.99 12.28 13 .. 78 13.93 13.39 13.11 13.21 9.41 9.22 10.61 11.05 11.09 11.26 11.66 11.78 6.13 5 .. 72 8 .. 00 7.82 8.40 9.15 9 .. 66 9 .. 64 4.95 4 .. 72 5.12 5.76 6 .. 34 7 .. 00 7.69 7 .. 55 2.15 3.53 3.85 3.72 4.02 5.04 5.49 5.65 1.53 1.68 1.94 2.54 3.14 3.34 3.68 3.52 0.83 0.92 1.27 1 .. 58 2.21 2 .. 21 2.36 3.52 0.53 0.81 0.86 0.95· 1.25 1 .. 43 1.81 2.19 0.22 0.33 0.36 0.58 0.68 1.15 1.32 1.24 0.18 0.16 0.24 0.35 0.48 0 .. 53 0.73 0.71 0.00 0.08 0.17 0.25 0.23 0.31 0.54 0.95 . 0.04 0.08 0.05 0.11 0.20 0.18 .0.21 0.43 0.04 0.03 0.02 0.13 0.09 0.05 0.18· 0.05 0.00 0.00 0 .. 02 0.05 0.11 0 .. 04 0.08 0.10 OoOOi 0.00 0.05 0 .. 01 0.05 0.05 0.08 0.19 0.00 0.00 0.05 0.04 0.05 0.09 0.05 0.14 I , FIGURE 1 GROUP GROUP 10 11 4 .. 01 4 .. 94 7.41 8.02 11.42 6.17 13.58 11.11 11.11 17.90 12 .. 04 12.35 10 .. 80 9.26 8 .. 95 6 .. 17 5 .. 86 4 .. 94 3.40 7.41 2.78 2.47 3.70 3.70 1.85 1.85 1.54 2.47 0.31 1.24 0 .. 31 0.00 0 .. 62 0.00 0.00 0.00 0.31 0.00 0.00 0.00 PROBABILITY·OF SCORING EXACTLY A GIVEN NUMBER OF RUNS (COMPOSITE MAJOR LEAGUE DATA, 1967-1976) .20 _. ~ ..... --- '--', ... !" --- !_~=lteam~ ~~Or1~- less i than J /7", ::! 3.74 runs per game .16 - 7·--'; '~ : -- .-r- . I I- all teams I PI........ \'. I tI' .. R .".- i ............... teams scoring more than o .12 7· 'II 4.25 runs per game B tI A I I B iii I .08 -: L / I T Y • 04 ~---!---_t__--t- .. -. . .... -+-----1 .ooL-__ 1-~~ __ ~ __ ~ __ ~~~~~~~~~~~~ o 2 4 6 . 8 10 12 16 18 20 EXACT NUMBER OF RUNS 9 "THE DISTRIBUTION OF RUNS SCORED diltribution for all major league teams in the ten year period 1967 through 1976. On Figure It the abscissa represents the exact number of"" runs in a game, while the ordinate represents the probability of scoring - exactly that number of runs. . The general trends are apparent. As scoring increases (scoring being measured by the average number of runs scored per game over a full season- or other period of time), the peak percentage decreases and shifts to the right. After the peak, the tailing-off portion of the curves show that the lower ~coring the team, the faster the tail-off. The regular behavior of the curves makes it a simple matter to obtain empirical equations expressing, for a team with a known average rate of scoring (R/G), the probability of occurance of each individual run total. If N is the number of runs and PN is the probability of scoring exactly N - runs in a game, then: (l) When N is S runs or fewer: PN = A + B(N) + C(N)2 where A = .38498 - .10839 (R/G) + .00800SS(R/G)2 B = .0010306 + .024139 (R/G) - .002943(R/G)2 C = -.01879 + .002Sl4(R/G) - .0000lS06(R/G)2 (2) When N is 6 runs or more: PN : D(EN) where D = 6.6210 - 2.496S(R/G) + .27Sl8(R/G)2 E : .058479 + .24022(R/G) - .02229l(R/G)2 As mentioned, R/G is the average number of runs scored per game over tht.,-. full season or other period of time. Possible applications of the runs scoring distribution described above- include: (l ) (2 ) The determination of whether a given pitcher on a team received above average or below average batting support. The determination of whether the batters on a team had a tendency to get hot simultaneously (and/or to slump Simultaneously). For if so, the number of high (and/or low) scoring games for ·the team ought to be more than theoretically expected. Whereas if the batters on the team got hott or slumped, independently of each other, then some would be hot while others were slumping, thus tending to cancel out the effects on the team's overall offense; the probable result being that the number of high and low scoring games would·· be close to the theoretically expected values. (3) An examination of. the records of the teams involved in close pennant races might reveal that the ultimate winners were consistently closer to (or perhaps consistently farther away from) their theoretically expected runs distribution than were the non-winning teams. (4) The data from Table 2 can be utilized to obtain an equation relating won/lOSS percentage as a function of runs scored and opponents' runs. (This item will be the subject of a future report). ._
