Where did all the runs go?

The following is an excerpt from the introduction to the Hardball Times Annual 2011:

The Hardball Times Annual proudly displays statistics you won’t find anywhere else, Batted Ball Stats. Even web sites like FanGraphs and Baseball Reference don’t provide what you can find in the Annual. I like to think of these numbers as a combination of statistics and scouting—numbers that quantify not only how well each player performed, but how he did it.

To give you a sense of how these numbers work, I’m going to review them in the context of 2010, the “year of the pitcher.” Here are some facts: In 2007, major league teams scored 23,322 runs. In 2010, they scored 21,308 runs. Two thousand runs have been lost in the past three years. Let’s see where they went. First, let’s look at strikeouts and walks as a percentage of plate appearances for the last four years:

    Year     K%      BB%
    2007    17.1      9.5
    2008    17.5      9.6
    2009    18.0      9.7
    2010    18.5      9.3

The 1.4 percent increase in strikeouts in just three years is huge. The walk rate had also been climbing, helping to offset the run-dampening effect of the strikeouts, but it dropped quite a bit in 2010. Forget home runs or ballparks—this little table explains over half of the decline in offense since 2007.

Our system applies a “linear weight” to each batting event, a weight that reflects the average impact each event has on run scoring. When you multiply the quantity of each event by its weight, you get a number that is roughly equal to the total number of runs scored each year. The system works pretty well.

So when you apply the appropriate weights to the total number of strikeouts and walks the past three years, you find that the net impact has been a decrease of about 400 runs, 20 percent of our total.

But there is another consequence of an increase in strikeouts: The total number of pitches put into play has decreased. Strikeouts and walks have grown from 26.6 percent of appearances to 27.8 percent. As a result, there were about 5,000 fewer balls put into play in 2010 compared to 2007. According to our linear weights, the net impact of this drop was about 700 fewer runs. If you add the 400 and 700, you get 1,100 runs. Over half of the decrease in runs from 2007 to 2010 has been the result of more strikeouts and fewer walks.

Okay, so what happened when the ball was hit? Here’s a list of the percentage of batted ball types for all balls put into play (ground balls, line drives and fly balls):

 Year    GB%   LD%   FB%
 2007   43.5  18.6  37.9
 2008   43.9  20.2  36.0
 2009   43.3  18.9  37.8
 2010   44.3  18.2  37.5

I should tell you that our statistics partner, Baseball Info Solutions, has video reviewers watch every batted ball to classify it. They even have people review the video reviewers. Still, it’s hard to tell the difference between a line drive and a fly ball, so there are certainly some judgment calls and biases in the data.

But when the overall data tell you that ground balls were up a full percentage point in 2010, you can believe the trend. And when the data tell you that line drives were down two full percentage points in just the last two years, you can believe that something was going on. As you can imagine, the average ground ball generates fewer runs than the average line drive. I estimate that the movement toward grounders and away from liners has resulted in about 250 fewer runs.

The rest of the difference, about 650 runs, is the result of the outcome of each type of batted ball. Below you will find a few factoids about batted balls: Outs per ground ball and the average number of runs generated by each grounder (GBR); the average number of runs generated by each line drive (LDR); and three different facts about outfield flies: home runs per outfield fly, outs per outfield fly and runs generated by the average outfield fly.

    Year   Out/GB    GBR     LDR    HR/OF  Out/OF    OFR
    2007    73.9    .046    .398    .103    82.9    .181
    2008    74.1    .045    .386    .107    84.2    .175
    2009    74.7    .041    .388    .109    83.0    .189
    2010    73.9    .043    .383    .102    82.8    .179

This mishmash of data accounts for 650 fewer runs during the three years under review.

{exp:list_maker}The out rate on ground balls had been trending up, and the subsequent run rate of grounders down, but they both came back to norms in 2010.
The run impact of the average line drive has trended down. According to BIS, this is primarily due to fewer doubles and home runs off line drives. This trend could easily be a data classification issue, so we won’t dwell on it here.
The run impact of an outfield fly jumped up in 2009 but fell back again this past year. There were fewer outfield fly home runs, but the out rate (that’s of outfield flies that aren’t home runs) also stayed down, which helped negate the run impact of fewer homers. Again, this may partially be explained by the relative classification of flies and liners. Only partially, however. {/exp:list_maker}Lots of info in just a few paragraphs, huh? Let’s recap. Run scoring has trended down the past three years primarily because there have been more strikeouts and fewer walks. In fact, the strikeout/walk differences account for more than half of the change in runs scored.

The rest of the change has been characterized by two things: a movement toward more weakly hit balls (fewer liners, more grounders) and weaker results from both line drives and fly balls.

Of course, these sorts of statistics can also be used for individual players. In our stats, you’ll find 2010 batted ball stats for all batters with at least 100 plate appearances and all pitchers who faced at least 100 batters. Plus, batted ball stats for all major league hitters and pitchers are available as an Excel download.

To get access to Batted Ball stats, please support the site and purchase the Hardball Times Annual 2011.


Dave Studeman was called a "national treasure" by Rob Neyer. Seriously. Follow his sporadic tweets @dastudes.
23 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Dan Greer
13 years ago

I come up with 4 possible causes, but no test for any of them.

1) Hitting talent trending downward. Cause: increased emphasis on defense as evaluation improves/gains acceptance.

2) Pitching talent increasing. Cause: random variation + 7-8 man bullpens (depth/specialization).

3) PEDs.

4) Weather/random variation.

Thoughts?

Dave Studeman
13 years ago

This is discussed in the Annual in a fine article by Steve Treder, partly borrowing thoughts from Craig Wright.  One thesis is that the answer lies in game theory between batters and pitchers, and that the “game within the game” has swung to pitchers, for reasons outlined in the Annual.

trantor
13 years ago

One significant change that is not factored in is pitch FX data. Whgat % of the increased strikeouts/decreased walks is from the league “training” the umps into a new (slightly larger) strike zone?

Seems like a major potential factor to me!

Steve Treder
13 years ago

“What % of the increased strikeouts/decreased walks is from the league “training” the umps into a new (slightly larger) strike zone?

Seems like a major potential factor to me!”

Without spoiling what’s in the Annual article Studes mentioned, I will say that I agree:  my hunch is that part of the explanation lies in a more pitcher-friendly strike zone than that which prevailed in various periods of the past.

Peter Jensen
13 years ago

Dave – By my calculations there were 2118 more strikeouts in 2010 than 2007.  Also 190 fewer NIBB and 106 fewer IBB.  The value of reduced opportunities are already reflected in the linear weight values for these events.  I believe the correct way to calculate their impact is to compare each event’s linear weight to the linear weight of an average hit ball event, which is about .024 for this time period. By my calculations the total decrease in runs for these events is a little over 750.  Still a lot, but not close to 1100.

DrBGiantsfan
13 years ago

Hasn’t there also been a significant trend of increased use of the cut fastball which is generally a groundball and strikeout pitch?

Dave Studeman
13 years ago

Thanks, Peter.  I don’t know why you’d compare the non-hit events to a batted ball.

I didn’t use “relative” linear weights, but “absolute” linear weights, in which the weight of the event is reduced by the value of an out (so that total events times total linear weights equals total runs scored, not zero).  Doing it this way allowed me to calculate the numbers the way I did.

Also, if you want to pursue your approach, I think you have to go all the way and account for all the missing runs to ensure that your method works.

Dave Studeman
13 years ago

Bickle, look at the data.  How does that support your theory that PEDs are the number one factor?

Peter Jensen
13 years ago

Dave – We’ve had threads on the proper way to calculate absolute linear weights at the Book Blog.  Tango has retracted his previous methodology of subtracting the value of an out.  The current accepted method is to add the average value of a PA to every PA. 

I compared the linear weights of the event to a batted ball since the alternative to a non hit event is a batted ball.  Presumably all the batters that were not striking out or walking (or being hit by a pitch) in 2007 were putting the ball in play.

Dave Studeman
13 years ago

Peter, I’ll look at the alternative way of calculating absolute linear weights.  Did you use absolute linear weights?

Also, would your calculation account for less plate appearances, as a result of more outs through strikeouts?

I do think it’s best to calculate the entire decline in runs scored and then compare notes.  While doing this analysis, I found that an incremental approach often overlooked other factors.

obsessivegiantscompulsive
13 years ago

I think baseball is moving away from the juiced ball, read High Boskage House’s SillyBall article, that opened my eyes, in conjunction with his research into PED as well:  http://highboskage.com/juiced-ball.shtml

That’s the easiest explanation, and best I think.

Dave Studeman
13 years ago

ogc, the data doesn’t fit your theory.  How does the juiced ball affect strikeout and walk rates, which account for over half the difference?  The entire “sub-point” of my article is that things like steroids or juiced ball can’t explain what’s going on.  A change in the strike zone is much more likely.

Dave Studeman
13 years ago

Peter (and others),

I have created and uploaded an Excel spreadsheet that lays out my logic.  Peter, I did change my methodology to adding runs/PA, which changes the result somewhat but not substantially.

Here is the link:

https://docs.google.com/leaf?id=0B-gtYTN4IAHMZWUwNDdhMzQtZDIwOC00NTczLWJiN2YtNzU4OTgxZGIwNTYx&hl=en

I would appreciate any and all feedback.  If there is an error in my methodology, I will upload a corrected document for those who purchase the THT Annual.

Thanks.

John DiFool
13 years ago

It may be, in this so-called post-steroid era, that pwoer hitters are still trying to use batting strategies which were optimal in the PED era, but are now sub-optimal.  Said strategies would include swinging hard on every pitch, in the hope that the HRs will outweight the K’s.

Travis Bickle
13 years ago

PED`s should be listed #1 not #3. Once HGH testing comes along look for a further reduction.

Dave Studeman
13 years ago

Unless steroids make batters hit the ball more often, I don’t see the connection.  Besides, pitchers took steroids as often as batters.

Red Sox Talk
13 years ago

Dave, thanks for crunching the numbers for us. I too have noticed the K and BB trends. One explanation I thought of was the introduction of a lot of young, quality pitchers in recent years; seems like the retread vets are being pushed out of rotations more and more.

Or could it be an impact of DIPS/FIP that teams are emphasizing fewer walks and more strikes to their pitchers? There certainly seems to be more awareness of it these days, even among pitchers if Brian Bannister and his friends are considered representative of some minority. Somehow the run value of a walk wasn’t really appreciated until Moneyball brought the spotlight onto the value of OBP and getting on base.

Dave Studeman
13 years ago

Red Sox Talk, you’re getting close to the answer that Craig Wright had, and Steve Treder discusses in the THT Annual.  I think that some combination of young arms, batter/pitcher “game within the game” and expanding umpire strike zones is probably the best answer.

Peter Jensen
13 years ago

Dave – Thanks for making your spreadsheet available.  The methodology that you used was very similar to what I was doing.  The year to year differences in the lg avg runs per PA don’t seem that great (.124 to .120) but since you have calculated them anyway I would used the actual number for each year.  I also calculated the linear weights separately for 2010.  What worries me is that I got some very strange numbers for 2010.  I am using PBP files that I downloaded from Gameday and I know that I still have some bugs in my parsing program but my run totals are close enough that I don’t think I should be getting the linear weights that I am calculating.  So I am trying to resolve that issue.

Mark F
13 years ago

Dave

I still think that the non-use of PEDs has a place in this theory.  However, perhaps it is the resulting positive effect on pitchers we should be considering!!

With a decreased fear of PED-aided HRs, pitcers are once again willing to throw pitches close to the strike zone. If fewer of a pitcher’s “mistakes” end up in the bleachers, they will be less fearful of making mistakes.

Mark

Mitch
13 years ago

Great stuff. Red Sox Talk’s comment made me think of something that maybe you’ve all thought of: are we seeing a “rubber-band effect” of the sabermetric movement? In its beginnings, sabermetrics was (were?) largely focused on hitting for understandable reasons. This knowledge gradually led to more productive hitters overall. However is the knowledge now catching up to fielding and pitching quality levels? Perhaps teams are starting to value the more important but less obvious qualities in a pitcher; e.g. a good K/BB ratio gets more attention now than the old W-L, ERA metrics. Similarly for fielders’ UZR vs. fielding percentage.

stu
13 years ago

Dave,

Great stuff.

To follow up on what Mark says… if the players are using less PEDs then perhaps that explains why the fly balls are not flying out, why the line drives are being caught and why the groundballs are not getting through- what you call “weaker results from ” batted balls.

Dave Studeman
13 years ago

Well, many thanks to Peter Jensen, who scrutinized my methodology and came up with his own approach. Every time I think I kind of understand linear weights, I am humbled.

I’ve spent the past couple of days immersed in this stuff and trying to sort it out, and have finally given up.  But the bottom line is this: the increase in strikeouts and reduction in walks doesn’t account for over half the decrease in runs…more like 35% to 40%.

The primary flaw in my approach is that I gave too much emphasis to strikeouts and walks decreasing total balls in play.  Other factors contributed to that decrease as well.  Plus, I apparently don’t know diddlysquat about linear weights.

This gives credence to those of you who want to emphasize PEDs as the cause of decreased scoring, though I still wouldn’t rank it first among all possible causes.