I prefer counting pitchers’ actual wins to hypothetical stats like WAR

12 01 2014
Ron Guidry, photo I took in 1977 with his daughter

Ron Guidry, photo I took in 1977 with his daughter

Pitchers’ wins don’t get much respect as a pitching statistic.

I don’t understand that. Isn’t the starting pitcher’s job to win the game? Wins are simple, usually fair, and easy to understand. Maybe that easy-to-understand part doesn’t appeal to the sabermetricians. Some of them appear to prefer stats like WAR that only they can understand.

I’d try to figure out WAR if it measured actual performance or measured something that really matters. Or if it were accurate. But I just analyzed how it measured Ron Guidry‘s 1978 performance, and it wasn’t anywhere close to accurate. It doesn’t give one shred of understanding about his performance that year, while his 25-3 won-lost record tells you immediately that he was historically dominant.

After I blogged my comparisons of Guidry to Greg Maddux and Tom Glavine, Andy Smith told me that wins weren’t an important statistic for pitchers. And he asked me why I’m not a fan of WAR:

Here’s why I think wins are one of the best measures of a starting pitcher’s value: The starting pitcher’s job is to win the game. And in most games the starting pitcher has more impact on whether his team wins than any other player.

Like any stat, wins are not a perfect measure: A powerful offense can help a pitcher win a game when he doesn’t have his best stuff. Lousy defense can lose a game for a pitcher who’s throwing well. A bullpen can blow a win that the pitcher deserved (or even get the win after blowing the save). It’s a team sport and the team context matters. But over time, starting pitchers who win lots of games are always really good pitchers. And starting pitchers who don’t win games don’t last.

Here’s why I don’t place much value on WAR (wins above replacement): I prefer measuring actual performance to hypothetical stats. WAR doesn’t actually measure real wins compared to a real replacement player. Both the W and the R in this stat are hypothetical, projected by computer analysis rather than measured by performance on the field.

I appreciate every attempt to improve analysis of baseball. I used to read the annual Bill James Baseball Abstracts and have his Historical Baseball Abstract and The Politics of Glory. I’m not going to argue that the Triple Crown stats are the best way to measure an offensive player (though they are good).

But I don’t embrace a stat just because it’s new or just because the sabermetricians do. I’m not a fan of WAR for pitchers or position players. It doesn’t measure what the player actually did. It tries to strip players in a team sport away from their teams and speculates about how many more games this player’s own actions “won” above what a mythical “replacement” player from the bench or minor leagues would have won. I won’t try to explain WAR further here (I don’t fully understand it), but you can read for yourself how they figure WAR for pitchers and for hitters.

I’ll just explain, using Guidry’s 1978 season, which Andy and I were discussing, why WAR isn’t nearly as accurate or as valuable as measuring what the player actually did.

In 1978, Guidry’s record was 25-3, 22 games over .500. We don’t have to speculate about what a “replacement” player on the Yankees would have done. Jim Beattie was a replacement pitcher, a rookie who started 22 games, making the team in spring training after some injuries to veteran pitchers. He lost seven games in a row at one time and slipped out of the rotation for a while but ended up with the fourth-most starts on an injury plagued staff. Since Beattie actually played pretty much the whole season, he’s actually a notch above replacement (which WAR confirms by giving him a rating for the year of 0.7 WAR). But he’ll do.

Beattie was 6-9. So one way to measure his wins over replacement would just be Guidry’s wins (25) minus Beattie’s (6), or a WAR of 19. I think that’s a fair way, because a replacement player isn’t generally going to play a full year anyway. But that doesn’t count losses, and Beattie lost three times as many games in his 22 starts as Guidry lost in his 35 starts.

A better way to count in my view would be to compare how far they were above or below .500. The Yankees were 22 games over .500 with Guidry getting the decision, three games under when Beattie got the decision, so that way, Guidry’s WAR would be 25.

Or maybe you want to project Beattie’s performance over 35 starts, which would give him a record of 10-14. That would change the WAR to 15 just counting wins or 26 counting wins and losses.

But WAR doesn’t work that way. The WAR formula gives Guidry 9.6 wins above a mythical replacement pitcher. It tries to slice and dice a player’s contributions to the wins to figure hypothetically what that player’s number of wins is. And it’s flat wrong.

Maybe you don’t want to credit the pitcher for every win the team got. Some of those (or fractions of every win) surely must belong to the offense or defense. But can’t we agree that Guidry was the most important factor in the nine shutouts he pitched in 1978 (including his epic 18-strikeout game in June and three September two-hit shutouts, two of them against the Red Sox on consecutive starts). Beattie didn’t have a shutout, so there’s nine wins above replacement right there. So Guidry was just .6 wins above replacement in his other 26 starts? Hardly.

Let’s set the hypothetical projections aside and look at Guidry’s actual starts, breaking them into different categories, according to the pitcher’s importance in the game and how much the pitcher contributes to the likelihood of a win:

Dominant starts. I’m going to define a dominant start as pitching eight or more innings and giving up no more than one run. If a starting pitcher does this, his team nearly always wins. If not, the blame goes to the bullpen or the hitters.

Should-win starts. These will be starts where the starting pitcher goes at least six innings and gives up no more than two runs. Usually if the starter does this, his team will win, but you need at least average offensive support and maybe some bullpen help. This is a more-demanding standard than a “quality start,” which is pitching six or more innings and allowing no more than three unearned runs. I think the pitcher often bears partial responsibility for unearned runs, so I just look at runs on the scoreboard.

Keep your team in the game. This will be when the starting pitcher goes at least five innings and gives up no more than four runs, without qualifying for the top two categories. The offense and bullpen play a bigger role, but most teams will come somewhere close to breaking even in these games.

Lucky if your team wins. If the starting pitcher gives up five or more runs or pitches less than five innings, the offense and bullpen really won the game, if you win at all.

Here’s how Guidry’s 1978 year broke down:

Dominant 

Guidry pitched eight or more innings allowing zero or one run 15 times in 1978. All 15 times, Guidry won the game. Thirteen of them were complete games where the bullpen was no factor. It’s a team game and you have to score at least a run to win even with a dominant pitching performance. But in every case, Guidry was by far the most important Yankee in winning the game.

Should-win

Guidry pitched seven games of six or more innings in which he gave up two runs. In three games, he gave up only a run but came out sometime after the end of the sixth inning and before the start of the ninth. The team wins the vast majority of these games, and the Yankees won eight of Guidry’s 10.

In the one-game playoff with the Red Sox (more on that later), Guidry pitched on three days’ rest. He gave up two runs on six hits and a walk in 6 1/3 innings, strong but not dominant. Mike Torrez pitched better for six innings and led 2-0 starting the seventh. A famous (infamous if you’re a Red Sox fan) three-run homer by Bucky Dent gave Guidry and the Yankees the lead and Thurman Munson added an insurance run that inning on an RBI double. After a strikeout and a single in the bottom of the seventh, Goose Gossage replaced Guidry, leading 4-2. Reggie Jackson hit another homer and Gossage gave up two runs in the eighth, but the Yankees never lost Guidry’s lead. The game exactly fits this category. When the starting pitcher gets into the seventh allowing just two runs, he’s done his share and if the teammates provide some offense and relief pitching, he’ll get the win. Guidry did.

On the other end of the season, Guidry pitched seven innings Opening Day and gave up only one run. He left a tie game and Gossage lost the game, 2-1, in the eighth. So there’s a game Guidry pitched outstanding in and got a no-decision.

Guidry’s second loss was a 2-1 complete-game duel with Mike Flanagan. He gave up two runs, one of them earned, on a homer following an error by Dent.

In addition, Guidry left a game with the Twins in the seventh inning, leading 2-1 with the bases loaded and one out. Gossage allowed the tying run on a sacrifice fly and ended up getting a win and a blown save as the Yankees won, 3-2. Both runs were unearned but charged to Guidry.

So we have three should-win situations where Guidry didn’t win. In this context, at least, Guidry’s win total was actually diminished, not padded, by the contributions of his teammates in these three games.

On the other hand, Guidry got notable help in two games with the Tigers. Once he trailed 1-0 after eight innings. Four Yankee runs in the top of the ninth gave him the lead and Gossage got the final three outs (and allowed a run) after Guidry walked the lead-off batter in the ninth. The other time, he had a 2-2 tie after eight innings. A run in the bottom of the 8th gave him the win and Gossage saved it with a scoreless ninth.

So in 10 should-win situations, the Yankees won eight of the games and Guidry got seven wins. Adding dominant wins to should-wins, in 25 of Guidry’s 35 games he pitched six or more innings, giving up two or fewer runs. His teammates did their share most of the time. Guidry was 22-1 in those games and the Yankees were 23-2.

Keep your team in the game

Guidry kept his team in seven games, going five or more innings and giving up no more than four runs (but not strong enough to make one of the other categories). The Yankees won six of the seven, so you have to give the offense some credit for scoring runs. One was a 5-3 complete game win over the Blue Jays, so the bullpen wasn’t a factor then. Guidry also pitched into the ninth in a 5-3 win over the Indians, saved by Gossage, who also saved a 5-4 win over the Orioles after seven innings by Guidry.

In three of the games, the Yankees won but Guidry didn’t. Twice the bullpen blew saves. Once he left with the game tied 3-3 (with the tying run unearned). The offense won the games, overcoming the runs given up by Guidry, the bullpen and the defense.

In the only game in this group that the Yankees lost, Guidry pitched five innings (one of only two appearances shorter than six innings) and left leading 3-2 (only one of his runs was earned) and Gossage blew the save in a 5-4 loss.

In the seven keep-your-team-in-it games, Guidry was 3-0 and the Yankees were 6-1. Even when he gave up a few runs, he contributed notably to the victories. The Yankees were 29-3 and Guidry was 25-1 in the 32 games where Guidry dominated, pitched well enough that he should win or kept his team in the game.

Lucky if your team wins

Guidry had only three of these games, and the Yankees were lucky once. He pitched nine innings against the White Sox, giving up six runs. The game went into extra innings, tied 6-6. Gossage got the win in 11 innings. The other two games were Guidry losses and rightly so. He gave up five runs in six innings in a 6-0 shutout by the Brewers. He deserved the loss, but the offense didn’t give him a chance to win either. In his worst game of the year, and the only one shorter than five innings, he gave up five runs, three of them earned, in 1 2/3 innings against the Blue Jays. That was an 8-1 loss, sandwiched among those September two-hitters.

When you give up five runs in a game, as Guidry did in those two losses, you can’t blame the loss on the offense. But Guidry had little offensive support in his three 1978 losses, with the Yankees scoring a combined two runs.

Adding up the four categories, in the 35 games Guidry started, the Yankees were 30-5, 25 games over .500.

More on Guidry’s amazing year

The Yankees had some other good pitchers that year. Ed Figueroa was 20-9, Catfish Hunter 12-6. Beattie and Dick Tidrow had losing records, each with more than 20 starts.  They weren’t exactly “replacement” level. The Yankees, with exactly the same hitters, fielders and relievers as Guidry played with, were 69-59 in the other pitchers’ starts. His starts were 15 wins more above .500 than the rest of the staff. How does that amount to 9.6 wins above a replacement pitcher?

Without Guidry, the Yankees were a .539 team, fourth place in the American League East instead of winning the division and ultimately the World Series. In the games Guidry started, they were an .857 team, way better than any team ever over a full season. Figure those winning percentages over a full season, and the Yankees would win 87 games with the rest of the staff pitching and 139 if Guidry could pitch every day. If you want to make up a hypothetical stat, say that Guidry was 52 WAT (wins above teammates) that year.

I prefer counting actual wins. With a little luck and team support, Guidry could have approached 30 wins. But I’ll settle for 25-3 as a fair measure of his year, maybe the best year any pitcher ever had.

Do you keep your team from falling behind?

OK, perhaps I exaggerate when I say a pitcher’s job is to win the game. In the American League (there was no interleague play in 1978), a pitcher can’t score and you can’t win without scoring. So really all a starting pitcher can do is keep his team from falling behind.

In 22 games in 1978, the Yankees never trailed while Guidry was in the game. They won 21 of those games. He only trailed by three runs three times, including two of his losses. He trailed in less than 20 percent of his 273 2/3 innings.

And Yankee Stadium didn’t particularly help Guidry. His won-loss record was nearly identical at home (12-1) and on the road (13-2). His ERA was a shade better on the road (1.69 to 1.79). He dominated anywhere.

How the pitcher impacts the game

Let’s take a close look at that Bucky Dent game, which the Yankees had to win to earn the championship. Dent was the star of that game, with his surprising three-run homer (it was his fifth homer of the year) that put the Yankees ahead.

But let’s quantify which Yankees had how much impact on that game:

Dent played the whole game at shortstop. He had three other at-bats, all outs, and threw out two runners in his only defensive chances. Of the 78 plate appearances of the game by either team, he was involved in six of them, three of them positively contributing to the victory.

Jackson was the DH and the homer was his only hit in four at-bats. First baseman Chris Chambliss went 1 for 4, advanced to second on a Roy White single and scored on Dent’s homer. He made eight putouts at first base. He was involved in meaningful ways in 14 plays, 11 of them positive. Second baseman Brian Doyle went 0-for-2 and didn’t make a defensive play. Pinch-hitter Jim Spencer was 0-for-1 and second-base replacement Fred Stanley was 0-for-1 and didn’t field the ball. So the four of them combined for four meaningful plays, none of them positive. Third baseman Graig Nettles was 0-for-4 with a putout and three assists, so he was involved in eight meaningful plays, four of them positive.

Center fielder Mickey Rivers got a double and two walks in four plate appearances. He stole second twice, scored on Munson’s double, caught two fly balls and fielded four hits to center. So he figured in meaningful ways in 13 plays, at least eight of them positive (the game log doesn’t tell whether he made a play on a hit that might have saved a base, which would make another positive play; I’m assuming these were routine plays that neither cost the Yankees nor helped them win). His replacement, Paul Blair, got one hit in his only at-bat and didn’t field the ball. He was forced at second, so he was involved meaningfully in two plays, one of them positive. Lou Piniella got a single in four at-bats, was forced at second base and caught four balls in right field and fielded three hits, so he was involved in 12 meaningful plays, at least five of them positive. White had a single and a walk in his four plate appearances. He scored on Dent’s homer and caught four balls in the left field and fielded three hits. So he figured in 11 meaningful plays, at least seven of them positive.

Gossage faced 14 batters, retiring eight and giving up five hits and a walk.

Munson was in the game for all of Guidry’s and Gossage’s pitches, but I’m not counting each pitch as a play, just those that resulted in outs or runners getting on base. Certainly he played a meaningful role in calling the pitches, but he also called the pitches of the other Yankee pitchers, who didn’t pitch nearly as well as Guidry. I’ll count the strikeouts as plays for Munson, since they are recorded as putouts. He had seven putouts, an assist and a passed ball as well as a double in five at-bats. So he figured in 14 plays, nine of them positive. That’s probably an undercount, but I’m not sure how to quantify the catcher’s involvement in pitching.

The game log doesn’t count every involvement, such as taking a cutoff throw, backing up a base or a late throw trying to catch a runner at home. But that’s a pretty close assessment of the players’ involvement.

Guidry faced 26 batters. He gave up six hits and an intentional walk, so 19 of his plays were positive. No one was involved in more plays or more positive plays.

No Yankee figured in anywhere near the 26 plays that Guidry influenced. Only Munson, Gossage, Rivers and Chambliss figured in even half as many plays. And no one made anywhere near Guidry’s 19 positive plays. Only Chambliss was in double figures and seven of those were putouts at first, catching balls from Guidry, Munson and the infielders. Not quite comparable to the role of a pitcher striking out a batter or inducing a ground ball or pop fly.

Quantitatively, Guidry impacted the most plays that contributed to the win. Qualitatively, you can still say (and I wouldn’t argue) that Dent’s homer was so big he still had the biggest role in the win. You can even argue that Munson’s role behind the plate, plus his RBI double, places him even with Guidry. But clearly, Guidry had a major impact on the win.

And, as I’ve noted before, he had 15 dominant starts with significantly better impact than he had in the playoff game. Break those down and his role will be even bigger. Every time. In addition, this wasn’t one of his stronger performances in the should-win category, so he easily had 20 games where he played a bigger positive role. Guidry went deeper into the game in 28 starts than he did in the playoff game.

Wins matter. Wins get you to the post-season and to championships. No one plays a bigger role in most wins than the starting pitcher. I’d much rather judge a pitcher based on his actual wins than on some hypothetical wins above “replacement.”

Stats and other details here come from Baseball Reference.

Other posts about Ron Guidry:

Ron Guidry compares well to three Hall-of-Fame Dodger pitchers

Bert Blyleven belongs in the Hall of Fame, but not before Ron Guidry

Ron Guidry elevated the great teams he played on

Great pitchers (Justin Verlander, Ron Guidry) really are the most valuable players

Advertisements

Actions

Information

13 responses

12 01 2014
Dale Alison

Two words: Steve Carlton, esp. 1972

Like

12 01 2014
Steve Buttry

I presume you’re supporting my point, Dale. Because Steve Carlton 1972 certainly supports it. As I said, the pitcher is the most important factor in winning the game. When he pitched, a lousy team won. Wins were indeed the best way to measure his success that year.

Like

13 01 2014
Brian Cubbison

You sound like a curmudgeon 🙂 I told my SABR friends, “No, no, keep reading. He’s got a point way down in there.”

The point being, here’s a rotation with the same team behind it, with an ace, a couple of quite good pitchers and even a replacement player for comparison. One of them wins way more for the same team than WAR suggests. What’s up with that?

Maybe Guidry’s season was just off the charts. Maybe after a certain point, WAR doesn’t make enough distinction between ultra dominance and ultra ultra dominance. Guidry’s WAR was at the top of the majors, after all. To be persuasive, we should see how it tracks the rest of his career. But real value in WAR is not in measuring just how dominant Guidry was in 1978, but whether Ed Figueroa could win 20 games with Texas.

You do a good job of listing all the ways that a starting pitcher doesn’t account for a win, including the abominable “vulture win” by a reliever, yet you feel that a pitcher’s win is a more real stat that takes place on the field. But it’s filled with arbitrariness and assumptions. I suppose that’s what SABR is fighting against.

Like

14 01 2014
Steve Buttry

Brian, I just threw the word “starting” in there a couple more times to make clear that I feel wins are the best measure of starting pitchers. Actually not very meaningful at all for relievers. I did say “starting” several places in the initial story. The “whether Figueroa could win 20 games with Texas” question is exactly the kind of hypothetical stat that I don’t care about. The developers of WAR might think that they know that. But I’ve just proven that it’s not even close to measuring Guidry’s actual performance against someone who was really a replacement pitcher on his team that very year. Why would I care about its projection in a hypothetical situation if it doesn’t measure actual performance accurately. You want to tell me Guidry wouldn’t have won close to 25 games for Texas that year going eight or more innings with one or fewer runs 15 times and six or more with two or less another 10 times? That guy wins more than 20 for every team (as Steve Carlton did in 1972, Dale accurately noted). And he means well more than 9.6 wins better than a replacement pitcher. For any team.

As for the curmudgeon remark, check out my defense of the DH. That’s the mark of a baseball curmudgeon, someone who still thinks 40 years in that the DH is ruining baseball. 🙂

Like

14 01 2014
Ryan

how can you even begin to critique a stat that you yourself admit you don’t understand? If you don’t understand it, then maybe the problems you have with it lie in the areas that you don’t understand.

Like

14 01 2014
Steve Buttry

Ryan, I actually understood WAR pretty well years ago when they first came out with it. I wasn’t very convinced then, so I never fell in love with it. I didn’t bother trying to explain it here because I know they have refined the formula over the years and because that would be boring. I apologize if my self-deprecation hurt my credibility with you. I do clearly understand these two facts about WAR: 1. It’s a hypothetical stat (and I prefer measuring actual performance) and 2. It attempts to measure individual performance, stripped out from the team context. And baseball’s a team sport. I think it says something (that sabermetricians have been unable to measure) that Guidry did such a good job that year (and much of his career) of hanging on to a lead (or keeping a game tied, if his team hadn’t scored yet). As I said in his post, that’s his job. I notice you avoid addressing the fact that WAR didn’t come close to accurately reflecting Guidry’s 1978 performance.

Like

14 02 2014
Numbers can be misleading in journalism or in sports | The Buttry Diary

[…] you can overanalyze and lose sight of a goal. In my baseball blog, I wrote recently about why I prefer wins as a measurement of a starting pitcher’s performance over WAR, a stat attempting to measure the hypothetical wins a player would produce above those delivered by […]

Like

17 02 2014
Project Unbolt question: How do you measure success in culture change? | The Buttry Diary

[…] Update: Matt Waite suggested this Brian Abelson post, Pageviews above replacement. It’s a detailed, sophisticated look at digital metrics, despite its misguided appreciation of the baseball stat wins above replacement. […]

Like

21 10 2014
Kansas City Royals’ ‘all-lost years’ team | Hated Yankees

[…] and if this team got down, he could always get pinch hit for. Dad note: Tom has more faith in WAR than I do. Kurt Stillwell wasn’t very good, so I’m not going to argue for him here, […]

Like

1 10 2015
Yankee pitchers win more championships than Cy Young Awards | Hated Yankees

[…] I provided a detailed analysis of Guidry’s Cy Young year in an earlier post. […]

Like

18 10 2015
Other notable Yankee starting pitchers: Al Downing, Don Gullett, Jim Beattie … | Hated Yankees

[…] argued in a post about Guidry’s performance that year that Beattie was the R in WAR, a replacement player against whom another pitcher’s […]

Like

19 10 2015
The Yankees’ 50 best starting pitchers | Hated Yankees

[…] had four 20-win seasons, one more than Guidry. But he didn’t have anything approaching Guidry’s spectacular 1978 season, maybe the best season ever by any Yankee. So I ranked Guidry higher than Ruffing. (Guidry also led […]

Like

30 10 2015
Were the 1986 Red Sox better than the 2015 Royals? | Hated Yankees

[…] one on the Royals was as bad a pitcher for the season or career as Nipper. I’m usually dismissive of WAR as a valuable stat, but I’ll use it here: Nipper’s ’86 WAR was -0.9 in ’86. The Red Sox’ […]

Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s




%d bloggers like this: