# Strikeouts, Grounders and Fly balls

 ``` Are some outs better than others? Do the number of times a runner is advanced on a grounder or a ball is booted make up for the times the defense turns two or throws out a lead (unforced) runner? To try to answer this question, I took Retrosheet's play-by-play data for the complete 1982, 1983 and 1987 seasons and calculated the number of runs a team could be expected to score during the rest of the inning for each of the 24 game situations (0, 1 or 2 outs and all 8 combinations of runners on base). Here's what I found: MenOn Number of Outs FST 0 1 2 --- .495 .261 .100 x-- .879 .521 .226 -x- 1.121 .676 .325 xx- 1.472 .912 .431 --x 1.299 .952 .381 x-x 1.730 1.205 .501 -xx 1.955 1.404 .581 xxx 2.209 1.514 .767 The number of plays was in the neighborhood of 500,000. I then figured the value of each offensive play by the formula: ( run-scored-on-play + exp-runs(end) ) - exp-runs(start) For example, a home run with no out and none on would have a value of (not too surprisingly) 1 run ( 1 + .495 - .495), while a foul out in the same situation would have a value of -.234 ( 0 + .261 - .495). This is the same thing that Gary R. Skoog did in the 1987 Bill James Abstract in his article "Measuring Runs Created: The Value Added Approach." I separated these plays into three groups, based upon the speed of the batter. I used Bill James' Speed Scores (also introduced in his 1987 Abstract) to do this, and picked scores of 3.0 and 5.5 as the dividing lines. Finally, I examined all plays where the batter hit into an out (including sacrifice flies and errors) and determined the average loss associated with each strikeout, grounder and fly ball. With that as a somewhat lengthy introduction, here's what I found: Average Weighted Slow batters Plays Loss Average strikeout: 17475 -.278 -.282 grounder: 28802 -.262 -.261 fly ball: 28562 -.261 -.261 Average batters strikeout: 33785 -.276 -.283 grounder: 63691 -.244 -.249 fly ball: 64532 -.257 -.260 Fast batters strikeout: 16776 -.268 -.282 grounder: 36428 -.230 -.240 fly ball: 32803 -.254 -.263 One strange thing about the average loss column is that the speed of the batter seemed to lessen the loss associated with a strikeout (-.278 for the slow runner and only -.268 for the fast one). This was caused because fast runners are more likely to hit in situations where the loss due to a strikeout is low (leading off an inning, for example). The "Weighted Average" column weights each situation according to its percentage in the entire sample. These results agree with common sense: speed at the plate is little help when the batter either strikes out or hits the ball in the air. They also show that a grounder, especially for a fast runner, is on the whole a better way to fail than either a strikeout or fly ball. Of course, this would vary from situation to situation. The following table show the penalties for each class of out in all 24 game situations for the slowest class of batters: Slow batters (speed score < 3): MenOn 0 out 1 out 2 out FST K GO PO K GO PO K GO PO --- -.232 -.210 -.231 -.159 -.146 -.160 -.099 -.093 -.099 x-- -.372 -.463 -.364 -.303 -.356 -.300 -.225 -.210 -.224 -x- -.444 -.261 -.401 -.346 -.294 -.350 -.323 -.300 -.325 xx- -.557 -.548 -.519 -.494 -.567 -.490 -.429 -.392 -.420 --x -.346 -.208 -.219 -.566 -.257 -.225 -.377 -.337 -.372 x-x -.534 -.395 -.409 -.720 -.580 -.341 -.491 -.441 -.489 -xx -.540 -.265 -.373 -.827 -.295 -.366 -.576 -.508 -.581 xxx -.681 -.579 -.454 -.747 -.819 -.404 -.759 -.664 -.732 Most of this make sense. The absolute worst time to strike out is with men on second and third and one out. The difference between a strikeout and a grounder is these situations is a whopping .532 runs. On the other hand, a strikeout is the best kind of out with a man on first and none or one out. The fly out is almost always in the middle, except when the bases or loaded or there's men on first and third with less than two outs. By the way, since the expected runs with none on and two out is .100, you might expect the cost of a strikeout to be -.100 instead of -.099. The difference is the relatively rare cases where a batter reaches first after a strikeout due to a passed ball or wild pitch. The table for medium and fast batters: Medium speed batters (speed score >= 3 AND < 5.5): MenOn 0 out 1 out 2 out FST K GO PO K GO PO K GO PO --- -.232 -.208 -.231 -.160 -.144 -.159 -.099 -.092 -.099 x-- -.373 -.413 -.369 -.305 -.327 -.303 -.225 -.209 -.222 -x- -.442 -.236 -.386 -.352 -.277 -.344 -.324 -.290 -.320 xx- -.589 -.548 -.521 -.494 -.541 -.489 -.430 -.390 -.425 --x -.346 -.167 -.209 -.568 -.208 -.184 -.375 -.326 -.375 x-x -.521 -.391 -.384 -.699 -.509 -.311 -.501 -.444 -.486 -xx -.551 -.277 -.337 -.825 -.302 -.438 -.581 -.503 -.571 xxx -.695 -.425 -.401 -.747 -.595 -.376 -.767 -.651 -.751 Fast runners (speed score >= 5.5): MenOn 0 out 1 out 2 out FST K GO PO K GO PO K GO PO --- -.231 -.207 -.231 -.159 -.143 -.160 -.099 -.090 -.099 x-- -.375 -.392 -.374 -.304 -.300 -.307 -.224 -.209 -.223 -x- -.441 -.250 -.386 -.346 -.289 -.354 -.320 -.295 -.323 xx- -.565 -.412 -.528 -.491 -.439 -.490 -.431 -.384 -.421 --x -.346 -.215 -.203 -.571 -.267 -.197 -.375 -.326 -.381 x-x -.539 -.280 -.405 -.731 -.379 -.369 -.501 -.448 -.497 -xx -.551 -.302 -.387 -.828 -.348 -.461 -.572 -.497 -.571 xxx -.695 -.352 -.450 -.753 -.524 -.413 -.767 -.672 -.731 So what does this mean in real life? How much of an advantage does a slow batter who grounds out a lot have over one who strikes out a ton? To get some insight into this, I made up two hitters who each accounted for 400 outs in a season. Here's the breakdown of their outs: K GO FO K GO FO Hitter A: 60 240 100 Hitter B: 150 100 150 Hitter A would be consider a low strikeout, high ground-ball hitter. His ratio of 2.4 grounders to fly balls would be among the highest in the league. Hitter B is at the other end of the spectrum, with a lot of strikeouts and relatively few grounders. So using the normalized penalties (described in the previous note) of -.282 (K), -.258 (GO) and -.264 (FO): we get the following aggregate loss: Hitter A: -105.66 Hitter B: -107.55 Or less than two runs over the course of a season. Sure, this doesn't seem like a big difference--hitter A's outs cost about 2 runs a season less than hitter B's, but this is true despite the fact that hitter A will ground into around 23 double-plays compared to less than 10 for hitter B. So where hitter A should be getting credit for making slightly less costly outs, he will in fact be charged for all those extra double-plays. This difference gets larger as the batter gets faster. K GO FO GDP PEN K GO FO GDP PEN Slow: 60 240 100 23 -105.66 B: 150 100 150 10 -107.55 Medium: 60 240 100 19 -102.74 B: 150 100 150 10 -106.35 Fast: 60 240 100 12 -100.82 B: 150 100 150 10 -105.75 In other words, the faster you are, the greater benefit you get from hitting a lot of grounders. Which means those grumpy old men are correct when they complain about high strikeout totals. Still, no one goes up to the plate intending to strike out; rather, they have developed a batting style that make them more or less vulnerable to whiffing. Often, these styles arise out of physical limitations and sometimes out of a belief that the increased power every time they DO make contact more than offsets the high strikeout totals. But are they correct? As a start to answering this question, I looked at all the players with 300 or more plate appearances in a season and divided them into 4 equal size groups based upon the percentage of strikeouts per plate appearance. From 1990-96 this gave me 4 groups of 415 players each, with the dividing lines between the groups 11.1%, 14.4% and 17.9%. At the two extremes were Bo Jackson in 1993 (106 Ks in 308 PAs) and Tony Gwynn in 1995 (15 Ks in 577 PAs). Here are the median averages for the four groups: K PCT Bavg Slug OBP OPS < 11.1% .285 .399 .348 .745 11.1% - 14.4% .276 .411 .346 .752 14.4% - 17.9% .269 .419 .336 .756 > 17.9% .259 .438 .334 .770 This is about what I expected. Contact hitters have a much higher batting average than the free-swingers (26 points), but their advantage is much less in on-base percentage (14) and does not come close to making up for their deficiency in slugging average (-39). In short, the more you strikeout, the higher your on-base plus slugging average. Now this does not mean that the free-swingers are necessarily better hitters--as much as I love OPS, I'm not convinced that on-base percentage shouldn't be weighted more than slugging percentage. Also, since the strikeout kings have a lower OBP, that means they are using up a few more outs, and I've already shown that these outs are slightly more costly than those made by hitters with low strikeout totals. Still, a 39-point advantage in slugging is a lot to give up to gain a 14-point boost in on-base percentage. But before the coaches among us start advising their players to swing for the fences on every pitch, perhaps they might want to consider the data for the 1980s. During this decade there were 2296 qualifying players, giving me 4 groups of 574 each. This time the dividing lines were 9.7%, 12.9% and 16.2% of strikeouts per plate appearance, and the ends of the spectrum were occupied by Bo Jackson in 1987 (158 Ks in 434 PAs) and Bill Buckner in 1980 (18 Ks in 616 PAs). Here's the table: K PCT Bavg Slug OBP OPS < 9.7% .278 .378 .341 .717 9.7% - 12.9% .271 .406 .338 .743 12.9% - 16.2% .264 .403 .327 .732 > 16.2% .256 .421 .323 .741 While the relationship between the two extreme groups is about the same, the best group in this decade seems to be the second, those striking out somewhat less than normal. The 1960s and 1970s reverted to form, however. To save space, here are the dividing lines and only the OPS for the groups, ordered from lowest to highest strikeout percentage: 1970s 8.9% 11.7% 15.0% .699 .723 .729 .758 1960s 9.9% 12.8% 16.2% .699 .719 .732 .739 Prior to that, things weren't quite so clear cut. For one thing, the hitters' strikeout totals were only a pale imitation of what they would one day become. A strikeout percentage of 9.9% would qualify a hitter for the top group prior to the fifties, but would land him among the contact hitters from the sixties onward. It's kind of sad what passed for a free-swinger in those days. Here's the data: 1950s 7.0% 9.7% 12.7% .762 .746 .748 .764 1940s 5.3% 7.5% 9.9% .726 .720 .726 .735 1930s 4.9% 6.8% 9.1% .768 .762 .763 .780 1920s 4.0% 5.6% 7.8% .781 .763 .752 .764 Tom Ruane ```