The Elusive Clutch Hitter
It’s (almost) spring (training), and a young man’s thoughts turns to baseball metrics. I’ll start with two charts:
This chart shows the batting average of a subset of major league players when runners are in scoring position (RISP) on the horizontal axis and the batting average of those same players with no runners in scoring position (nRISP) on the vertical axis.I wanted to see if batting average increased with RISP, which would indicate a clutch effect. By my logic, any situation with RISP, no matter the score, inning, what part of the season, whatever, is a clutch situation. My underlying hypothesis is that when there are RISP, batters want to bat them in, or at the very least, advance them. If this is the case, we should be able to see an increase in batting average with RISP as opposed to nRISP. I chose an arbitrary improvement of 10%, which equates to about 27 points of batting average (the average average, as it were, of this sample was .277). The black line at the 45-degree angle shows quite clearly that about as many hit worse with RISP (the points above the line) as hit better with RISP (those points below the line). At the end of this article, I’ll fully explain my sample, sample size and methods.
This next chart summarizes the data shown above:
 |
| From Scott Lindholm Stats |
As stated above, I was looking for those players who managed to bat 10% higher with RISP (or 10% worse), and the black lines mark the areas on the left and right which denote the number of those players. To begin, here are the players who managed to bat 10% or more better with RISP:
 |
| From Scott Lindholm Stats |
The AB, H and BA columns show how the player performed with no runners in scoring position, and the spAB, spH and spBA columns show how the player performed with runners in scoring position. The percentage difference in batting average was calculated by (spBA-BA)/BA). As you look at that list, how many of those players jump out at you and scream “Yep, he sure was a clutch hitter”? There are some arguments to be made regarding the player’s non-scoring position batting average, since the lower the average, the greater the opportunity to have a higher percentage increase in batting average, but that didn’t really occur at the macro level. And also, I did black out that person who topped this list on purpose to gently nudge the nice people at FanGraphs to publish this post, at which point I’ll reveal it (and there’s no way you’ll guess who it is–NO WAY).
This chart shows those players who batted 10% worse with RISP:
 |
| From Scott Lindholm Stats |
Again, nothing here that really screams out at me, except that as a Cub fan, I’m not too shocked about Alfonso Soriano. Here we can also see that batting average really wasn’t a factor, since only two .300 hitters (Robinson Cano and Hal Morris) made the list, and the higher the average,the easier it would be for a greater percentage decrease.
I’ve never been a big believer in clutch hitting in the first place, because I don’t think it’s something that can be replicated at will. A PGA golfer can drive the ball 300 yards every time, a good NBA free throw shooter will make it 85-90% of the time and a major league fastball pitcher will be able to hit 95 on the radar gun consistently, but hitting in clutch situations is entirely different. Past success will have no bearing on the next at-bat, and even to the extent it may occur, it will make a very minor difference. For a player who has 600 at-bats in a season, about 1/4 (on average) will be with RISP, or about 150 at-bats. If a player can improve his batting average 10%, or from about .272 to .300, he’ll increase his hits in these situations from about 42 in a season to 45. To put it another way, he will be successful three more times IN A SEASON than if there was no improvement, or about once every 50 games. That’s a pretty small difference.
This last chart will show the wild swings in the performance of some of the best hitters in recent memory:
 |
| From Scott Lindholm Stats |
It’s messy on purpose. Even some of the best hitters swing wildly from season to season in their ability to bat for average with RISP vs. nRISP. I don’t claim this to be the end of the clutch hitting discussion, and there are plenty of other measures that can measure it, but I liked this simple way of defining clutch because it’s clean, easy to explain and a fairly rational definition of clutch. In the end, I don’t think there are clutch hitters–I think there are good hitters, and good hitters will perform as they do, clutch or no clutch situation.
METHODOLOGY
For those interested, I put forth my question in the standard research method. My null hypothesis was there was no difference in batting average with RISP vs. nRISP.
All my information came from split data available on baseball-reference.com. My sample size was 557 players, chosen in three ways:
1. For all players whose career began in 1990 or after, at least 1,000 at-bats with runners in scoring position
2. For current players, at least 200 at-bats with runners in scoring position (I have a few with less, but I was interested in them)
3. For any player who played in 1990 and met the criteria in point 1, I went back and got their data. This would be players like George Brett, Robin Yount, Paul Molitor, etc.
Some would argue that I chose the best hitters, and I probably did, since better hitters will have longer careers, which should make it easier to see if clutch hitting exists. I can run the same tests using all players since 1990 (and might), but I’m not sure the results would vary appreciably. I also noted the effect that a player’s batting average might have, and was quite surprised to see the following data:
 |
| From Scott Lindholm Stats |
At every batting average breakpoint, there was little difference (dBA) between the average with nRISP (nBA) and the average with RISP (spBA). As mentioned earlier, I will do further study of OBP, since I have no clue what causes that difference other than a greater willingness to take a walk, and also to see what effect (if any) sacrifice flies will make. If you want the data set I used, I’ll share the career player data.
Awesome post!
WHO ARE THE BLACKED OUT PLAYERS!?!?! I’m dying to know.
Outstanding study. The best I’ve seen on FanGraphs for some time.
Have you looked at OBP excluding IBB? Given that almost all IBB are given with runners in scoring position, this is bound to increase OBP.
It looks like the years don’t match up correctly with the rest of the columns in your first table (the “clutch” hitters). I don’t think Elvis Andrus had 1068 AB between 1990 and 2004. Neat article.
The AB, H, and BA columns don’t seem to add up. I think what you are showing are the total hits and at bats, but then the BA is just for at bats without runners in scoring position?
If that’s true, then the the blacked out folks are Josh Bard and Akinori Iwamura?
Yes, the blocked out players are indeed Josh Bard and Akinori Iwamura. And yes, that particular table is mistaken, and I can’t quite figure it out right off the top of my head. I’ll have to see where I made my mistake, and hopefully be able to post a correction somewhere.
Bob’s probably right that a fair bit of the OBP bump with RISP may come from IBBs. But I have to hope, for the sanity of MLB managers, that that’s not what’s causing the 10% OBP bump for the sub .250 hitters. If there’s one thing sabermetrics should have taught us, it’s “don’t give bad hitters free passes!” Unless it’s mostly due to NL managers walking the #8 hitter to get to the pitcher? Curious…
I’m going to make another article with just the two charts (only one of which has errors). I made a small methodology change that didn’t affect the overall results, which in itself I find interesting. In case my next post is NOT published, here’s what I intend to do: 1. I don’t know how the years got mixed up on the better hitters, but their data is essentially correct–ONLY the years playeed are incorrect, except for some minor errors I’ve noticed when looking at the data (for example, I shorted Josh Bard about 100 ABs somehow). 2. The charts I… Read more »
I would be morewinterested in the other side….
Who are the chokes???
I like how Barry Bond’s RISP was like 80% higher that year. I guess when you only have to swing at fastballs down the middle, or take a walk, you do pretty good.
Folks reading this should probably know that at last year’s SABR conference I proved that career RISP / non-RISP hitting differences are unquestionably real (the odds against finding the correlations I found by sheer chance are upwards of 500,000,000,000,000,000,000 to 1). You can look for that study sometime this year over at THT. The key point is that the differences have nothing to do with clutch. It’s about different adaptations to the changing dynamics of the pitcher / hitter matchup. Each base / out situation has its own dynamic based on the relative values of a walk and home run… Read more »
Eric:
Could another explanation be the change in fielders’ positions from ideal to less-than-ideal when they have to account for a base-runner? I guess this would only give an advantage to batters with RISP and not a disadvantage but it has to have some effect right?
The definition of clutch is a rather arbitrary one it seems. Batting with the bases loaded in a 10-0 game is hardly a clutch situation. Batting in the 9th with the team down 1 run and nobody on is a clutch situation. Also, anyone with fewer than 2000 AB there is likely going to be have a SSS effect since only 500 AB will be with RISP, or less than 1 full season. There will be a lot of noise in data with samples this small. Furthermore, failing to find evidence of something is not proof it does not exist.… Read more »
1) You could do weighted least squares regression analysis to verify that the regression line is at 45%. There’s no reason to eyeball this.
2) Does previous clutch/anti-clutch performance predict future clutch performance? Again, you could answer this through regression analysis, either using presence in the top/bottom groups as an indicator variable or using a continuous predictor.
the OBP difference is probably the result of intentional walks. Those are far more likely to occur with RISP, than without.
The small up tick in slugging could be explained by a reduction in bunting in RISP situations.
OK. After years and years of either thinking of this question, or having it forced on me, I’ve reached this conclusion: A CLUTCH HITTER IS A HITTER UPON WHOM YOU CAN COUNT TO GIVE YOU A QUALITY AB/PA IN PRESSURE SITUATIONS. I believe that by now, we all accept that luck plays a huge part in the success or failure of a major league hitter. For example, in the 2000 WS, Luis Sojo hit a 22-hopper that rolled right over the second-base bag, inches between the outstretched gloves of both the SS and 2B, and the winning run in the… Read more »
A small reason why you might expect AVGs to go up with RISP: a fly-out with a runner on third may often not count as an at-bat.
Simply wish to say your article is as astounding. The clarity for your publish is just excellent and i can think you are a professional on this subject. Well with your permission let me to seize your feed to stay up to date with imminent post. Thank you one million and please carry on the gratifying work.
1) Pitching from a stretch is a disadvantage, or they would all be pitching out of the stretch with nRISP.
2) A pitcher that is in the stretch is by definition not sailing through an inning.
So why averages not much higher with RISP? Other than the occasional force made at 2B that would not be made at 1B–balls up the middle or in the SS hole.