Home Runs + 1

--------------------------

            At Bats +42

       And for "batting average"  I used this:

              Hits + 16

-----------------------------

             At Bats + 60

       It’s what I call ballast.   I do that so that I don’t get crazy results for players who hit 2 home runs in 5 at bats.   Going 9-for-25 with 3 homers doesn’t make you a .360 hitter with home run rate of 72 per 600 at bats; it makes you a .294 hitter with a home run rate of 36 per 600 at bats, which is still pretty good, but normal. 

The Predictive Formula

       A player’s expected walks are:

1)   His home run rate,

2)   Times .7,

3)   Plus .069,

4)   Times his plate appearances,

5)   Times 1.1 if the player is a switch hitter, 1.08 if he is a left-handed hitter, and 0.92 if he is a right-handed hitter, and

6)  Increased by 12% if he hit over .340. 

       Let’s do a couple of players for whom the formula works, to show how it works.   Duke Snider in 1954 hit 40 home runs in 584 at bats, which we change to 41 in 626, figuring that he had a home run rate of .0655.  

       That number we multiply by .7, making .0458.  

       To that we add .069, making .1148. 

       We multiply his plate appearances (679) times .1148, making 77.98.  

       This we increase by 8% because Snider was a left-handed hitter, making an expectation of 84.2 walks.

       Snider in 1954 hit .341, HOWEVER, because of the "ballast" adjustment, we don’t treat him as a .341 hitter, but as a .334 hitter.  Therefore, we don’t make an additional adjustment for his high batting average, and his expected walks stay at 84.2.

       He did in fact walk 84 times.   The formula accurately predicts his walks, in that particular case. 

       George Springer in 2017 hit 34 homers in 548 at bats, which we interpret as a home run percentage of .0593 (35/590).  Multiply that by .7, you have .0415.   Add .069, you have .1105.

Multiply that by his plate appearances, 629, and you have 69.52 walks.    He is, however, a right-handed hitter, so we reduce that by 8%, and his expected walks are 63.96.  He actually drew 64 walks. 

       Chris Davis in 2017 hit 26 homers in 456 at bats, which we interpret to be a home run rate of .0542 (27/498).  Multiply that by .7, you have .0380.   Add .069, you have .107.  Multiply that by his plate appearances, 524, and you have 56.04.   He’s a left-handed hitter, so we’ll increase that by 8%; that makes 60.53.   He actually drew 61 walks.

       Hanley Ramirez exactly hit his expected walks drawn both in 2016 and 2017.   

       Let’s do somebody who has no power.   Luis Polonia in 1991 hit 2 homers in 604 at bats, which we interpret as a home run rate of .00464, or 3/646.   Multiply that by .7, you have .00325.  Add .069, you have .07225.   Multiply that by his plate appearances, 662, and you have 47.83 walks.   He’s a left-handed hitter, so we increase that by 8%, he winds up with 51.66 expected walks.  He actually drew 52 walks, so the formula works. 

       Later on, we’ll do the cases where it doesn’t work.  There are lots of cases where it works perfectly, literally hundreds of them, thousands of them if you included the low-at-bat guys, and many cases where it doesn’t work.  But when it doesn’t work, it is equally likely to be 50 walks too high, or 50 walks too low. 

The Results

       First of all, you have to just throw Barry Bonds in 2004 out the window; Barry Bonds in 2004 is just stupid data.   Note what I am saying; I am not saying that you have to throw Barry Bonds’ data out the window; I am just saying Barry Bonds in 2004.  We all know why this is; we don’t need to talk about it or explain it.   We’re just going to throw it away and move on.

       Throwing away Barry Bonds in 2004, the 25 most exceptional walk seasons of all time—that is, the 25 seasons in which the player most exceeded his expected walks—are these 25 seasons:

       Eddie Stanky and Eddie Yost are, by this chart, the greatest non-threat Walkers of all time.   Wovenstrap said that "Barry Bonds" is not the answer he is looking for, and I agree that it isn’t, but Bonds in 2004 would have been +123 walks, if we were counting that. 

       Running the data for Stanky in 1945, The Brat hit 1 home run in 555 at bats, which we interpret as a home run rate of .00335.  Multiply that by .7, you’ve got .00234.  Add .069, you’ve got .07134.  Multiply that by his plate appearances (726), you’ve got an expectation of 51.8 walks, but he’s a right-handed hitter, so we reduce that by 8%, and he’s down to 47.7, which we will call 48.  He actually drew 148 walks, which I think was a National League record at the time, so he beat expectations by 100.   He is the only player in history, other than Bonds in 2004, to beat his expected walk total by 100. 

       Ed Yost in 1956. . .Ed Yost was known as the Walking Man.  Yost hit 11 homers in 515 at bats, which we interpret as a Home Run Percentage of .02154.   Multiply that by .7, that’s .0151; add .069 and it is .0841.   Yost had 684 plate appearances, so that’s an expectation of 57.5 walks, but he was a right-handed hitter, so we multiply that by .92, and we’re down to 52.9, or 53 walks.  He actually drew 151 walks, so that’s +98. 

       These are the greatest NON-Walking seasons in the data, by this method:

       So the greatest NON-Walker of all time is, let us say, Garrett Anderson in 2000.  I’ll run the data for Garrett Anderson.  Anderson hit 35 homers in 647 at bats, which we interpret as a Home Run Percentage of .0522.  Multiply that by .7, that’s .0366.   Add .069, that’s .1056.  Multiply that by his plate appearances, 681, and he’s expected to walk 71.896 times.  He’s a left-handed hitter, so we increase that by 8%, and we’re up to 78 expected walks.  He actually walked only 24 times, so he is 54 walks short of expectation—the largest shortfall of all time.

       People think that Barry Bonds and Babe Ruth and Ted Williams walked a tremendous amount because they were left-handed hitters and great hitters with high averages and a lot of power, so naturally they’re going to walk a lot.  It’s not that simple.   Garrett Anderson in 2000 was a left-handed hitter; he hit .286 with 35 homers, which is pretty good, similar to what Barry Bonds did in 1995 (.294 with 33 homers).  But when Bonds did that, he drew 120 walks; when Anderson did it, he drew 24.  Bill Terry in 1932 was a left-handed hitter who hit for a good average (.350) with 28 homers—but he didn’t draw walks.  Hal Trosky in 1936 was a left-handed hitter who hit .343 with 42 homers—but he didn’t draw walks.  Cecil Cooper in 1932 was a left-handed hitter who hit .313 with 32 homers—but he didn’t draw walks.   Tony Oliva in 1964 was a left-handed hitter who hit .323 with 32 homers—but he didn’t draw walks.  Chuck Klein in 1930 was a left-handed hitter who hit .386 with 40 homers, but he didn’t draw a lot of walks.  Bonds, Ruth and Ted Williams didn’t draw huge numbers of walks because they were left-handed hitters who hit .350 with power; they took a large number of walks because that was part of their approach.  It was an additional skill that they had. 

Career Numbers

       Well, before I get there, there is a point I should have made earlier.  

The Point I Should have Made Earlier

       Hitters walk, over time, in about 8.8% of plate appearances.  The first-effort approach to predicting walks for each hitter, then, is just to predict that every hitter will walk in about 8.8% of his plate appearances.  

       You can improve that estimate by (1) adjusting for his power, and (2) adjusting for whether he is a left-handed hitter or a normal person, but these improvements don’t actually do a hell of a lot.   You have a certain amount of error in the first effort, and then, by making these adjustments, you can remove about 20% of that error, get the estimates 20% closer to what actually happens.

       And then that’s about all you can do, just remove about 20% of the error.   I’m sure you could do better than I could, if you spent a week with the data rather than a few hours, but I’m also pretty confident that you couldn’t remove 40 or 50% of the error; maybe you could get to 25% or something, but not much better.  

       The reason that is true is that drawing walks is a strong individual skill or trait.   How many walks you draw—and here again, we have to say "with the exception of Barry Bonds in 2004."  But with the exception of Barry Bonds in 2004, how many walks you draw is NOT primarily a function of how much power you have or what your batting average is or whether you are fast or slow or fat or ugly or whether you hit right-handed or left-handed.   It is a primarily a function of the batter’s ability to draw a walk.   Because it is an independent skill of its own, it cannot be indirectly measured or independently predicted with great accuracy based on the player’s other characteristics.  

Career Numbers

       These are the greatest walk-drawers of all time, career totals, in terms of drawing more walks than you would expect them to draw:

       26^th is Wade Boggs.   And these are the greatest walk-drawers in terms of the ratio of expected to actual walks drawn, minimum total of 1000 between expected and actual:

       These are the top 25 NON-Walkers of all time:

Here is a full listing of all players whose walks + expected walks total 1,000 or more.   Data for active players is a year out of date, because I haven’t updated something for 2018 yet: