# Barry Bonds’ situational wins

When should you walk Albert Pujols? Josh Hamilton? Cesar Izturis? There’s no easy answer because the decision depends not only on the base/out situation, the inning and the score; it depends on the quality of the batters coming to bat next, the nature of the pitcher/batter matchup and maybe even the phase of the moon. Even Izturis was intentionally walked once last year (with two out, runners on second and third and the pitcher up next; the moon was in its first quarter.).

Tom Tango, MGL and Andy Dolphin wrote the definitive chapter on intentional walks in The Book: Playing the Percentages in Baseball (a must-have for all serious baseball fans) and I’m going to dive into the intentional walk question by taking one of Tango’s best (and least understood) statistical concepts—Situational Wins—and applying it to the all-time leader in intentional walks, Barry Bonds.

Do you know what Linear Weights are? They’re the weights that can be applied to events on the field in order to properly credit players for their impact on the game. A single is typically worth about .45 runs, a double .80 runs, a triple 1.0 runs and a home run 1.4 runs. There are several ways to create your linear weights, but my favorite is the “Value Added” approach, which works like this:

{exp:list_maker}Figure out the 24 basic out/base situations (three outs—0, 1 and 2—and eight base situations—bases empty, runner on first, etc.).

For each base/out situation, go through the season and figure out how many runs subsequently scored during that inning. Call that the Run Expectancy Matrix.

Go through the season again and figure out how much each batting event changed the team’s place on the Run Expectancy Matrix (easy example: a single with no one on and one out results in a runner on first and one out).

Use that average change to assign a Linear Weight to each type of event.{/exp:list_maker}The very best on-line example of this methodology is Tom Ruane’s work at Retrosheet. If you want more detail about the method, read Tom’s article.

Linear Weights assigns an average weight to each type of batting event, depending on how often each one occurred in each type of situation. This can lead to some inefficiencies in the system. For instance, pitchers rarely walk batters with the bases loaded. Wild pitchers who do give up walks with the bases loaded, however, have hurt their teams more than their Linear Weights suggest. This is where Situational Wins comes to play.

Situational Wins goes two steps beyond Linear Weights:

{exp:list_maker}It takes the **specific** base/out situation of each event into account. This is only possible thanks to Retrosheet’s fantastic data collection efforts.

It adds the inning and score to the context because it measures impact on wins, not runs.{/exp:list_maker}

Let me give you another example to show the difference. Bases are loaded in the bottom of the ninth, tie score. If the batter makes an out, the game goes into extra innings. If the batter gets a hit, home team wins. It doesn’t matter whether the batter hits a single or home run; the home team will win regardless. So, each type of hit has the exact same value to his team.

Linear Weights treats the single and the home run differently. Situational Wins doesn’t.

And how do we compute Situational Wins? Easy. It’s Win Probability Added (WPA) divided by Leverage Index (LI). It’s so simple to calculate that both Fangraphs and Baseball Reference include it in their player “win stats,” but most people probably don’t understand what it is or what it measures. Now you know: It measures the contribution each batter made to his team’s wins, based on contributing specific batting events in specific situations.

The reverse is also true for pitchers, by the way. I’ve just decided to describe the system from the batter’s perspective because it fits my topic of the day. If you want to read more detail about Situational Wins, try this article from Tom Tango. If you want even more detail, read this one.

I pulled the event files from 2001-2004 and calculated the average Situational Win of each type of batting event. These were my findings:

Event SW Single 0.040 Double 0.068 Triple 0.095 Home Run 0.142 Walk 0.030 Intentional Walk 0.006 HBP 0.030 K -0.027 Ground out -0.025 Fly Out -0.026 DP -0.054

Notice anything familiar about these numbers? They’re roughly equal to 1/10th of each event’s Linear Weight … which makes sense when you consider that it takes about 10 runs to turn a loss into a win. (By the way, here is a mathematical proof, based on the Pythagorean Formula, that a win must equal two times the number of average number of runs scored per game and team.)

Anyway, this is an indication that Situational Wins work, that they make sense. And now I’m going to show you how we can use them.

It seems like ancient history, but Barry Bonds had four amazing years from 2001 to 2004. Here are a few key stats for each year:

Year PA HR NIBB IBB SO BA OBP SLG 2001 664 73 142 35 93 .328 .515 .863 2002 612 46 130 68 47 .370 .582 .799 2003 550 45 87 61 58 .341 .529 .749 2004 617 45 112 120 41 .362 .609 .812

73 home runs in 2001 is ridiculous, of course (and leave your comments about steroids for another article). Perhaps even more ridiculous, however, was the fact that he received more intentional walks than non-intentional walks in 2004. Did Bonds deserve such a high level of respect? Is there any way we can put this many free passes in perspective?

Situational Wins to the rescue. Let’s start with Bonds’ average Situational Wins per at-bat for each year in question. This is how much he contributed, on average, when the opposing pitcher didn’t walk him or hit him with a pitch:

Year Avg SW 2001 0.017 2002 0.016 2003 0.012 2004 0.013 Total 0.015

Every time he was allowed to swing away, Bonds added 0.015 wins for his team. That’s a very good number. Keep in mind that the sum of all Situational Wins is zero—all the positive and negative events, all the wins and losses, cancel each other out—and that walks are a positive event for the batter. That means that the average at-bat, not including walks, is a slightly negative event. Bonds’ at-bats were positive by a good margin.

These numbers give us an important baseline. The answer to “When should we walk Barry Bonds?” is simply “When the Situational Win resulting from a walk is less than his average Situational Wins in at-bats.” Intentional walks are a choice, and you know that if you don’t walk Barry Bonds, he will produce .015 Situational Wins, on average. If you don’t want to take your chances, then you should walk him when the impact of the walk will be less than 0.015 Situational Wins.

*Quick aside: I’m oversimplifying things here, because when you do pitch to Barry Bonds you often wind up walking him anyway and that affects the baseline Situational Wins. I’m choosing to ignore that wrinkle for now.*

Did managers follow this rule? Let’s look at each year.

In **2001**, when Bonds batted third in the order and had Jeff Kent hitting behind him, he was intentionally walked 35 times. In 31 of those IBB’s, the resulting Situational Win was less than 0.017, Bonds’ at-bat baseline that year. The average for all of his intentional walks was 0.011, significantly less than 0.017. Opposing managers were being appropriate, perhaps even conservative.

In **2002**, manager Dusty Baker continued to have Bonds hit third in the lineup with Kent fourth, and then flip-flopped the two for the last two months. Regardless of where he batted, opposing managers issued intentional walks to Bonds twice as often … 68 times in all. 51 of those intentional walks created Situational Wins of 0.16 or less and the overall average climbed to 0.013 WS per intentional walk. Managers weren’t as conservative as they had been in 2001, but they were probably still appropriate on average.

In **2003**, Bonds primarily batted fourth and Kent’s protection was replaced by several batters, such as Edgardo Alfonzo and Benito Santiago. Opposing managers continued to walk him at about the same rate, resulting in 61 intentional walks. 45 of those walks resulted in a Situation Win of 0.014 or less—an overall average of 0.011. Still appropriate, I’d say.

All hell broke loose in **2004**. Bonds got off to an otherworldly start, batting .472/.696/1.132 with 10 home runs in the first month. Managers panicked, walking him a staggering 120 times in the year with batters like Alfonso and Pedro Feliz batting behind him.

61 of those intentional walks resulted in a Situational Win of 0.013 or less (his average at-bat SW that year) and 69 were less than 0.015 (his average at-bat SW all four years). Just a little more than half of them were clearly appropriate; the others, not. The overall average Situational Win for Bonds’ intentional walks in 2004 was 0.015, higher than his average Situational Wins in his at-bats.

Managers must have thought that walking Bonds helped their overall cause. On May 1, for example, Marlins’ manager Jack McKeon intentionally walked him 4 of 5 times* and Bonds went on to bat only .250 in May. Perhaps some people saw a a connection. In fact, all the intentional walks may have had an impact on Bonds. This is the average Situational Win of Bonds’ home runs each year (remembering that the overall major league average was 0.142):

Home Runs Only Year Avg SW 2001 0.164 2002 0.161 2003 0.150 2004 0.132

By walking Bonds in high-leverage situations, managers clearly eroded the impact of his home runs. However, the decrease in home run impact was more than offset by an increase from those intentional walks. Bonds’ overall Situational Wins average rose in 2004.

All Plate Appearances Year Avg SW 2001 0.020 2002 0.019 2003 0.015 2004 0.017

As great as Barry Bonds was, we all lost our perspective in 2004.

By the way, these calculations are based on league-average hitters. Say what you will about the likes of Alfonso and A.J. Pierzynski, they weren’t worse than league average at the time.

To wrap this up, here is a table of all the intentional walks given to Bonds during these four years, broken into three groups (Okay, Maybe and No Way).

WPA/LI Number Pct. Less than 0.016 199 70% 0.016-0.019 21 7% 0.019 and Over 64 23% Total 284 100%

39 of the 64 “No Way’s” occurred in 2004.

**That game included the most egregious example of a bad intentional walk when McKeon had Bonds walked with one out and no one on base in the fifth inning, with a 3-2 lead. That represented a Situational Win of 0.037. The Giants didn’t go on to score in that inning, but they did eventually win the game.*

**References & Resources**

You can calculate the Situational Win of a walk by using our WPA Inquirer.

This article was inspired by Steve Treder’s piece in walks to sluggers.

Many, many thanks to Tom Tango and Retrosheet.

Tom also developed a WPA framework for when/when not to walk Barry Bonds. Here’s one for when the Giants are on the road and here’s one for when they’re at home.

Print This Post

I wonder about the unintentional intentional walks. (those 4-0 or 4-1 counts where the batter doesnt really face any pitches worth swinging at) Does this perhaps factor into some of those walks in 2001 and 2002? I realize that its a really hard thing to look at but Barry Bonds batting eye at this point was really good so maybe it got a little overboard in 2004.

It would have only gotten them one guaranteed leadoff at-bat per game, and presumably no manager would walk him to lead off a game- It would have also caused a tremendous amount of uproar- but if the Giants put Barry first in the lineup it may have been a good move after that hot first month (and reduced some of the ‘no ways’ in 2004).

Of course, statistically the Giants benefited from those ‘no ways’, I suppose.

I have some thoughts about using Situational Wins to judge non-intentional walks, but that may take a while.

My favorite way of putting those walks in perspective:

2004 H+BB+HBP = 376

2004 AB = 373

But this is definitely a more methodical way of looking at it.

you forget everytime he waled in 2004 the next coupkes batter hits an drove in runs so it faild