Hitting with RISP

[Appling]

A player's batting average with RISP (Runners in Scoring Position) may be a simple example of "hitting in the clutch". I sampled 14 good hitters using data from www.retrosheet.org to see if their career Batting Averages with RISP tend to be better or worse than BA when there are no runners in scoring position.

Retrosheet has no splits data prior to 1959 AL or 1960 NL, so that leaves out many of the old-time greats. The 14 players I sampled are: Hank Aaron, Barry Bonds, Wade Boggs, Roberto Clemente, Norm Cash, Tommy Davis, Rickey Henderson, Harmon Killebrew, Mickey Mantle, Willie Mays, Brooks Robinson, Frank Robinson, George Brett and Sammy Sosa. Of this group, five players have career Batting Averages with RISP that are ten or more points higher than their BA without RISP: Aaron, Boggs, Clemente, Killebrew and Mays. Some of these have a much better BA with RISP: Aaron .319 w/RISP and .288 without RISP; Killebrew .273 and .253; Mays .312 & .287.

Why would these players have a significantly higher BA with RISP? I can think of several reasons: (1) With RISP, the pitcher cannot use a wind-up; (2) the pitcher is unlikely to pitch a breaking ball (curve, splitter or change-up); also (3) the hitter is more likely to swing just to get a hit, rather than swing for the fences.

Another five players had lower career BA with RISP than with no runners in scoring postion: Cash, Davis, Henderson, Mantle, and Sosa. Mantle had his best seasons before 1959 (we have no splits data prior to 1959) so his early career may have had a higher BA with RISP. Henderson was such a great base-runner that pitchers might often "groove" the ball to him with bases empty rather than risk putting him on with a walk -- but with RISP they would be willing to pitch to the corners of the plate.

This table (attached) shows that most of these hitters averaged .450 RBI per AB with RISP, versus about .100 RBI when there were no runners in scoring position. Players not known for HR power have even few RBI when hitting without RISP: Wade Boggs .024; Roberto Clemente .049; Tommy Davis .039; R >Henderson .039; Brooks Robinson .038; and George Brett .050. These numbers confirm the thought that players in the middle of a batting order, who more often hit with runners on base, have many more opportunities to drive in a run than a player who usually hits with no Runners in Scoring Position.

Also please note that I included situation splits data for each hitter's best RBI season. In their best RBI season, some of these hitters had incredibly high BA with RISP; look at Hank Aaron (.409 BA with RISP but only .285 otherwise in his 1963 season); Barry Bonds (.382 RISP v .315 all other in 2001); Roberto Clemente (.407 v .295 in 1966); Harmon Killebrew (.314 v .261 in 1969); Mantle (.364 v .305 in 1961); Willie Mays (.361 v .285 in 1962); Frank Robinson (.390 v .324 in 1962); -- and George Brett (,469 with RISP v. "only" .357 other in 1980).

Norm Cash in 1961 and Tommy Davis in 1962 had BA and RBI totals much better than they had in any other season. Any suggestions why this might be?

[Ubiquitous]

SF

Sacrifice Flies gobble up most of the differences in batting average and SLG. The Flyball out that is fortunate enough to drive in a run saves the batter from using up an at bat. For instance someone mentioned that Thurman Munson said that when he had runners on he used to concentrate more, and on the surface that appears to be the case. His line with men on and with runners in scoring position is higher then with bases empty. But when you figure in SF his batting average and SlG fall to below bases empty level. Though Thurman did increase his walk rate in these situations.


AS to why you might get some outlier, well a lot of times its luck. We are talking about an extremely small amount of AB spread out over an entire season and over all of the country.

[SABR Matt]

The major league average with RISP is higher than the major league average with none on base.

There are a couple of obvious reasons this is the case.

1) If a pitcher has gotten to the point where there are runners on base...he's not at his absolute best...there is a selection bias involved in choosing at bats where the the pitcher is already in trouble.

2) With RISP the defense is often forced to change the way they're covering the field (infield in...a shift..the wheel play...outfield in...covering the lines...etc), and the new arrangement is not advantageous to the prevention of hits compared to the way the field is typically covered (it's covered that way for a reason...through the years we've determined the best way to position fielders to stop most hits).

"Clutchness" requires more than a demonstration that a player has a higher average with RISP...it requires the domstration that the player has hit significantly better than their average with bases empty...significantly being defined as a larger difference than is normally expected for RISP situations.

[Ubiquitous]

Actually it is a few reasons with the number one reason being the one I mentioned, sacrifice flies.

[SABR Matt]

I didn't see your post til after I made mine, but I agree...SF do help a lot too..but I do believe the selection bias and the fielding situation play a role.

[Appling]

Actually it is a few reasons with the number one reason being the one I mentioned, sacrifice flies.

I have now come to understand and agree with you. I had always looked at the SF rule as it affects overall BA for the season, and found that it usually adds 2 or 3 points to the hitter's batting average (compared with what that BA would have been under the 1940s rules which did not credit SFs).

The record for Sac Flies in a season is 19; but only 52 players since 1954 had more than 12 Sac Flies in a season. The more common number is between 3 and 8.
Case in point: Hank Aaron is credited with 121 sac flies in his career. His official career BA is .305 (3771 AB in 12364 at-bats) but if he had been charged with another 121 AB (the number of SF) his overall career average would be only .302. (A 3-point drop)

Splits data for Aaron's 1963 season shows 201 hits in 631 AB (.319 BA) with 5 SF. Adding those 5 SF to his season at-bats total gives 201/636 = .316. (Again a 3-point for the overall season)

I never considered this 3-point "inflation" due to the SF rule to be very important, but I now understand how important the SF is to a player's BA with RISP. The SF can happen only when there are Runners In Scoring Position, and that occurs for most players in only 20% to 27% of their total career AB. So a 2 or 3 point inflation on the overall season BA would cause and 8 or ten point inflation of the player's BA with RISP.

To see the impact of the SF rule on my data for career BA of selected hitters:

Hank Aaron career BA of .319 with RISP but just .306 if we discount 88 SF. *
This is still better than Hank's career BA w/o RISP: .288.
* Note: SF totals here are only those reported in splits data available to Retrosheet.

Barry Bonds -- career RISP: .305 v .293 if we add 53 SF to offical AB (12 point impact)
Bonds career BA w/o RISP: .297

Tommy Davis -- RISP: .286 v .277 if 68 SF are added to AB
Davis career BA w/o RISP: .297

Rickey Henderson -- RISP: .266 v .258 if 48 SF are added to AB
Henderson career BA w/o RISP: .322

Wade Boggs -- RISP: .348 v .332 if 61 SF are added to AB
Boggs career BA w/o RISP: .335

Roberto Clemente -- RISP: .338 v .328 if 45 SF are added to AB
Clemente career BA w/o RISP: .326

Harmon Killebrew -- RISP: .273 v .263 if 75 SF are added to AB
Killer's career BA w/o RISP: .253

George Brett -- RISP: .312 v .299 if 110 SF are added to AB
Brett career BA w/o RISP: .305

Willie Mays -- RISP: .312 v .302 if 56 SF are added to AB
Mays career BA w/o RISP: .287

Also look at the selected "high RBI seasons" for these same players:

Aaron 1963: RISP: .409 v .394 if 5 SF are added to AB
Aaraon 1963 BA w/o RISP: .285

Bonds 2001: RISP: .382 v .374 if 2 SF are added to AB
Bonds 2001 BA w/o RISP: .315

Davis 1962: RISP: .376 v .362 if 8 SF are added to AB
Davis 2001 BA w/o RISP: .332

Henderson 1985: RISP: .288 v .276 if 5 SF are added to AB
Henderson 1985 BA w/o RISP: .322

Boggs 1987:RISP: .339 v .319 if 8 SF are added to AB
Boggs 1987 BA w/o RISP: .370

Clemente 1966: RISP: .407 v .402 if 2 SF are added to AB
Clemente 2001 BA w/o RISP: .326

Killebrew 1969: RISP: .314 v .306 if 4 SF are added to AB
Killlebrew 1969 BA w/o RISP: .261

Brett 1980: RISP: .469 v .445 if 7 SF are added to AB
Brett 1980 BA w/o RISP: .357

Mays 1962: RISP: .312 v .302 if 3 SF are added to AB
Mays 1962 BA w/o RISP: .285

This small sampling confirms the argument that most hitters have a higher BA with RISP, but that much of this "improvement" is artificial -- because the SF rule discounts outs that score the runner from third. Those same outs would be included in chargeable At-Bats with no RISP.

Nonetheless, batting average with RISP is an important stat in evaluating a hitter's performance: how does he perform when he has a chance to drive in runs? High RBI hitters seem to hit well when there are RISP.

Interesting to me also that players who usually do not hit in the power part of the batting order (such as Henderson and Boggs) hit better when no runners are in scoring position -- especially after making adjustment for SF with RISP.

[Ubiquitous]

2004 Major Leagues with SF counted
None on: .262 AVG
Men on: .266 w/o .271
RISP: .256 w/o .264
BaseLo: .255 w/o .276

[Appling]

2004 Major Leagues with SF counted
None on: .262 AVG
Men on: .266 w/o .271
RISP: .256 w/o .264
BaseLo: .255 w/o .276

Until now I had not noticed that Retrofit.com also provided situation splits information for an entire league.

I sampled league or overall MLB numbers for random seasons: 1962-1967; 1979-1981 (shortened season); 2001 and 2004. For overall league totals, the reported BA with RISP was generally within a few points of the league BA in all other situations. In some cases, the reported BA with RISP was five points or more higher than for "other": 1962, 1963, 1967 AL and 1979 AL.

However, as noted by several previously, this BA with RISP is artificial in the sense that a fly out which would count as an official AB in all conditions gets credit for a Sacrifice Fly if it scores a runner from third. I had been sampling well-known players to create an estimate of the magnitude of the Sac Fly Effect on Batting Averages with Runner-In-Scoring-Position.

Using league totals, this "estimate" is now a much more accurate number. The true BA with RISP (when SF is added to the official AB total) is 8 to 10 points lower than the official reported BA with RISP. In my sample, only once was the adjusted (true) BA with RISP higher than that league BA in all other situations: BA -- SF -- Adj. BA
1967 - AL w/ RISP .242 348 -- .235
1967 - AL - other - .234
In that season, the official AL average with RISP was .242, eight points better than BA w/other: .234. When 348 SF were added to the official AB with RISP, the adjusted BA is .235 (one point higher than without RISP).

Here is the full table I created from this sample data:

League Batting Average with RISP
Yr-League Situan BA SF Adj. BA

1962-mlb RISP 0.261 833 0.253
Other 0.256

1963-AL RISP 0.250 388 0.242
Other 0.246

1963-NL RISP 0.250 376 0.242
Other 0.243

1964-mlb RISP 0.251 735 0.244
Other 0.250

1965-mlb RISP 0.247 758 0.239
Other 0.245

1966-AL RISP 0.242 379 0.234
Other 0.240

1966-NL RISP 0.254 356 0.247
Other 0.257

1967-AL RISP 0.242 348 0.235
Other 0.234

1967-NL RISP 0.248 383 0.241
Other 0.249

1979-AL RISP 0.275 765 0.265
Other 0.268

1979-NL RISP 0.260 549 0.252
Other 0.261

1980-AL RISP 0.269 719 0.259
Other 0.269

1980-NL RISP 0.257 577 0.248
Other 0.260

1981-AL RISP 0.259 450 0.250
Other 0.255

1981-NL RISP 0.256 379 0.247
Other 0.255

2001-MLB RISP 0.264 1363 0.259
Other 0.267

2004-MLB RISP 0.266 1424 0.257
Other 0.263

Conclusion: for all players taken together, there is little difference in BA with RISP and BA without RISP. When SF are counted as AB, the BA with RISP is almost always lower.

Yet when the BA with RISP is computed for star players, their BA with RISP is very often higher -- sometimes by five or more points. This suggests that good hitters perform even better in potential RBI situations, while most other players do not.

[misterdirt]

Yet when the BA with RISP is computed for star players, their BA with RISP is very often higher -- sometimes by five or more points. This suggests that good hitters perform even better in potential RBI situations, while most other players do not.

You have not done enough analysis to be able to draw this conclusion. You would have to calculate BA and BA with RISP - SF for all batters, stars and non-stars, and look at the distribution to see if there are significantly more stars than non-stars whose BA w/RISP-SF exceed their normal BA by a given number of points. And your definition of stars should not include the ability to get RBI's because that would be assuming what you are trying to prove.

[Ubiquitous]

Over on THT John Walsh last month looked into sac flies and found:

as a group batters do seem to be able to change their approach at the plate to increase the probability of getting a fly ball to score a run in a sacrifice fly situation. However, the increase in fly balls comes simply from putting more balls in play (by striking out and walking less often) and not by batters putting more of their batted balls into the air.

How many extra sacrifice flies result from the extra fly balls? Here is a table of how often each type of batted ball results in a sac fly:

+------------+------+------+-------+
| Ball type | SF | Opps | Rate |
+------------+------+------+-------+
| F | 2666 | 4422 | 0.603 |
| L | 16 | 2667 | 0.006 |
| P | 12 | 1225 | 0.010 |
+------------+------+------+-------+

I was surprised by the low number of sacrifice flies off line drives, but this is what the play-by-play data tells us. Part of the reason is that about 75% of line drives go for base hits and hence do not produce a sacrifice fly. In any case, about 60% of fly balls will score a runner from 3rd with less than 2 outs, so the 229 extra fly balls result in about 137 extra sacrifice flies. One may ask, though, if there is some price to pay for the modified approach. We've already seen that hard hit balls (line drives) are reduced in sacrifice fly opportunities. Well, this will be the subject of a later analysis, but I'll leave you with one more comparison, the "biggest" defense-independent outcome of them all:

Fly HR
All Opps: 4193 603
SF Opps: 4422 507
Diff: 229 -96
Diff Pct: 5.5 -15.9

In other words, despite the increase in fly balls, the number of home runs is down significantly. We'll look at this more closely in the next installment.

Then he looked a little more into the subject in his next article.

What can we take away from this little study of batting performance in sac fly situations? That, as a group, batters modify their hitting approach in such situations. They adopt a more "contact-oriented" approach, cutting down on strikeouts and walking less, in an effort to put more balls in play, with the goal of driving in the runner from third. I have not tried to determine if this is the correct strategy in sac fly situations; such an investigation could be performed with the play-by-play data and would be an interesting analysis in its own right. What I've tried to do here is answer the question "Is the 'contact-oriented' approach generally more productive than the standard approach?" The answer appears to be "no," as can be seen after translating the aggregate performance in sac fly situations into a defense-independent context.

Can we say anything about what kinds of approaches individual batters should adopt? Based on this study, probably not. I have looked at groups of batters, but each batter is different and there is no a priori reason to believe that the group performance reflects the situation of any particular batter. As we saw in the sac fly article, the samples for individual batters are too small to draw any conclusions. However, the method of using particular situations to isolate a particular batting approach may prove fruitful in this regard. It may be possible to identify more inclusive situations, say "RISP" instead of just "sac fly," where a particular batting approach can be studied. With some more work, more years can be included in the study (I've used 2003-2004 data for this analysis). It may ultimately be possible to answer the burning question:

If Jim Edmonds put the ball in play more often, would he be a more productive hitter?

somebody else did a study I think or it might be one of these studies but I missed it in which the author broke batters into three categories based on their hitting, and found that flyball hitters do rather well (obviously) in sac fly situations while groundball hitters do rather poorly in sac fly situations. I bleieve more pop flies and so for these guys.

[Appling]

"Yet when the BA with RISP is computed for star players, their BA with RISP is very often higher -- sometimes by five or more points. This suggests that good hitters perform even better in potential RBI situations, while most other players do not."

You have not done enough analysis to be able to draw this conclusion. You would have to calculate BA and BA with RISP - SF for all batters, stars and non-stars, and look at the distribution to see if there are significantly more stars than non-stars whose BA w/RISP-SF exceed their normal BA by a given number of points.

The data listed in post #8 above is enough to convince me that, on average, the overall BA of an entire league with RISP is very near that without RISP. I admit that much more data is needed to quantify the VARIANCE for individual hitters -- or even the shape of the distribution for individual hitters -- but ...

When SF totals are added to the official AB number, the league "adjusted" BA with RISP is a more realistic measure of the average hitter's performance -- and it is usually 7 to 10 points lower than this average was BEFORE this adjustment for sac flies. Overall BA may not be as meaningful a stat as OBP or Slg. Pct., but BA with RISP seems to be a very important and meaningful stat.

I now understand and agree that, by not counting SFs as at-bats, we have in the past inflated "BA with RISP" such that a direct comparison with "BA without RISP" is not meaningful. Although the "raw" BA with RISP (for an entire league) is often a few points higher than BA without RISP, the more realistic "Adjusted" league BA with RISP is lower than the league "BA without RISP" in 15 of the 16 seasons I sampled. The only exception was the American League in 1967, when the "raw" BA with RISP was .242 and the "adjusted" league BA with RISP dropped to .235 -- but still one point better than the league BA without RISP.

Yes, my sample of 14 "good hitters" may be anecdotal rather than statistical. It may not reveal much about the overall distribution of all hitters (when batting with or without runners in scoring position) but I think it gives me interesting comparative data for these 14 hitters.

Retrosheet provides splits data for nearly 8400 of Hank Aaron's career at-bats. In 6,408 at-bats with no runners in scoring position, Hank averaged .288. In 1,985 chargeable AB with RISP he averaged .319; and when 88 sac flies are added to these 1985 official at-bats his adjusted BA with RISP is still a very respectable .306 For whatever the reason, Hank clearly hit better when there were runners in scoring position. Over his long career, this difference is too large to be ascribed only to chance.

Roberto Clemente, Harmon Killebrew and Willie Mays also have a career BA with RISP (even after adjusting for SFs) that is higher than than their careet BA without Runners In Scoring Position:
Clemente: .329 RISP (adjusted for SF) v .326 without RISP
Killebrew: .263 (adj.) v .253 without RISP
Mays: .302 (adj.) v .287 without RISP.

(Career BA with RISP for Aaron, Killebrew and Mays is ten points or more better than their own BA without RISP. Since -- for an entire league -- the adjusted BA with RISP is usually 3 to 10 points lower than the league average without RISP, the fact these these hitters consistently hit better when runners are in scoring position is remarkable. They may be "good" hitters all the time, but they are GREAT when there is a chance to drive in runs.)

Compare the adjusted career BA with RISP with the reported BA without RISP for the other ten hitters:
B Bonds: .293 adjusted (v .297 career BA without RISP)
W Boggs: .332 adj (v .335) Boggs was a high-average hitter all the time.
N Cash: .256 adj (v .274)
T Davis: .277 (v .297)
R Henderson: .258 (v .286)
M Mantle: .265 (v .285)
B Robinson: .259 (v .269)
F Robinson: .283 (v .292)
G Brett: .299 (v .305)
S Sosa: .258 v .278

These are all great hitters, but over the course of a full career they seem to hit for a higher average when there are no runners in scoring position. For six of these ten hitters, the difference is ten points or more. Over a long career (each hitter has more than 4000 AB with situational splits data), this large a difference must have an assignable cause -- even if we don't know it yet.
* * *
I also compared "Batting Average with RISP" (adjusted for SF) with the "BA without RISP" for the best RBI season reported for these 14 hitters. As expected, the hitter's BA with RISP (even after adjustment for SF) is almost always better than his BA without RISP in the hitter's best RBI year. Looking at just the hitter's adjusted BA with RISP shows some great numbers for these selected high RBI seasons:
Aaron 1963: .394; Bonds 2001: .374; Clemente 1966: .402; Cash 1961: .363;
Davis 1962: .362; Killebrew 1969: .314; Mantle 1961: .348; Mays 1962: .354;
B Robinson 1964: .326; F Robinson 1962: .379; Brett 1980: .445.

Surprising to me, for three of these hitters their high RBI season was not due to a high BA with RISP:
W Boggs 1987: .319 with RISP (adj) v .370 without RISP
R Henderson 1985: .276 with RISP (adj) v .322 without RISP
Sammy Sosa 2001: .298 with RISP (adj) v .329 without RISP

If I searched long enough I am sure I could find seasons for each of these hitters in which BA with RISP was higher than BA without RISP.

I will attach a text file showing my data for these hitters.

[misterdirt]

This suggests that good hitters perform even better in potential RBI situations, while most other players do not."

I don't understand the point you are trying to make with your last post. You state that you examined the hitting of 14 stars with men in scoring position and 4 of them hit better than they did without scoring position and 10 of them hit worse. How does this data suggest that good hitters perform better in potential RBI situations and most other players do not? What would the data be for 14 non-stars chosen at random? Would all of them hit worse with men in scoring position or would 3 or 4 or 5 hit better? That you found 3 out of 14 stars that had the ability to raise their performance levels a significant amount with men in scoring position over the course of their careers is interesting. But if you study non-stars as well you may find that a similar percentage had this ability as well.

Stars get more RBI's than average players because a: they hit better than average players and b: they hit in more situations with men on base. Whether they are better than the average player at raising their performances in RBI situations remains to be demonstrated.

[Appling]

That you found 3 out of 14 stars that had the ability to raise their performance levels a significant amount with men in scoring position over the course of their careers is interesting. But if you study non-stars as well you may find that a similar percentage had this ability as well.

Stars get more RBI's than average players because a: they hit better than average players and b: they hit in more situations with men on base. Whether they are better than the average player at raising their performances in RBI situations remains to be demonstrated.

I guess you are referring to my post #8 more than to #11 above. Using splits data for complete leagues, I stated:
"Conclusion: for all players taken together, there is little difference in BA with RISP and BA without RISP. (but) When SF are counted as AB, the BA with RISP is almost always lower.

Yet when the BA with RISP is computed for star players, their BA with RISP is very often higher -- sometimes by five or more points. This suggests that good hitters perform even better in potential RBI situations, while most other players do not."

On further review, I must agree with your last paragraph: that stars get more RBI because (1) they hit better than the average player (with RISP or not); and (2) they get to hit more often with RISP.

Nonetheless, my admiration for Hank Aaron went up a few notches when I saw that his career BA with RISP was so much better than his career BA in other situations -- and this kind of "clutch hitting" seems to be rare among all hitters. But you are right: my small sample does not prove anything about "all" star players.

This exercise did teach me a few things I didn't understand before:
(1) Overall BA may not be as important as OBP or OPS, but Batting Average with "Runners In Scoring Position" is very important.
(2) An initial look at situation hitting splits can be deceiving: by not charging a hitter for an At-Bat when he hits a Sacrifice Fly, we greatly distort any comparison with his "BA when no runners are in scoring position". (A Sac Fly cannot be earned unless there is a runner on third base.)
(3) After adjusting for SFs, most hitters have a better BA when there are no runners in scoring position. (Based on my sampling of seasonal league totals data.)

This last "finding" truly surprised me: I expected that most MLB players would hit for a higher average when hitting with RISP -- that is, when the pitcher can't use a wind-up and is not likely to throw a breaking ball.

I still don't understand why the adjusted league BA with RISP is generally lower, but it makes me better appreciate the "clutch hitting" of those (few?) hitters who -- over a long career -- hit far better with RISP than they do otherwise. Aaron, Mays and Killebrew hit best in RISP situations, while Mickey Mantle, Sammy Sosa and Frank Robinson certainly did not. Even Reggie Jackson, that great "Mr. October", hit only .256 (adjusted for SF) with RISP, compared with his career average of .262 in situations without runners in scoring positon.