30-game winner Paul Skenes?! A new formula to bring pitcher wins back to life

There have been 2,664 pitchers who have made at least 30 career starts since 1901.

Three of those pitchers -- or one out of every 888 -- own a career ERA below 2.00. Two of them are Hall of Fame deadball era greats: Ed Walsh (1.82) and Addie Joss (1.89). The third is Pittsburgh Pirates superstar Paul Skenes.

The chances of Skenes, who has made just 39 career starts, remaining in that class are slim. That's nothing against him. It's the reality of math and the era in which he plays. The careers of Joss and Walsh overlapped in the American League from 1904 to 1910, when the aggregate ERA was 2.61. The collective ERA in the majors since Skenes debuted is 4.04.

This season, Skenes' 1.85 ERA leads the majors, and he's first among all pitchers in bWAR (4.4). The latter figure is actually tops among all National League players, period. The current numbers generated by my AXE system and the futures at ESPN BET both mark Skenes as a solid favorite to win his first NL Cy Young Award.

Incidentally, Skenes' won-loss record for the woeful Pirates is a meager 4-6. Should we care?

Yes, we should care about pitcher wins

Won-loss records for pitchers are no longer part of the evaluative conversation, so if your response to the previous question was "no" then congratulations for paying attention. If your response was anything else, then it's almost certainly because you're in a fantasy league that still uses pitcher wins, not because you think Skenes' record actually tells us anything about his true value.

But what if I could tell you this and prove it: Skenes' real won-loss record is 11-5, the win total tied for the third-most in the majors. I'm going to explain how I got there, but first, let me explain why I think it matters.

Just to illustrate how starting pitchers were written about for most of baseball history, I pulled up the 1980 MLB preview from the Sporting News and went to the page where the Pirates (defending champs at the time) were analyzed. Here's a bit on their pitching:

"The Pirates last year won without a 15-game winner. The staff won in bunches. Five pitchers won 10 or more games."

There were no other pitching statistics in the staff outlook. No ERAs, no strikeout rates, nothing about walks. This was it. This is just how pitchers were discussed back then.

It's good that we understand how to assess pitchers now at a deeper level and, even back in 1980, people like Bill James were already doing it. But pitching wins still meant something as one of the baseball statistics James might allude to as having achieved "the power of language."

That is: To describe a pitcher as a 20-game winner had real meaning. It was an avatar for quality, and if someone was a five-time 20-game winner, that was an avatar for greatness.

Pitcher wins have always been an imperfect measure, but its flaws have ballooned over time as the game and the responsibilities of the starting pitcher have evolved. Last season, 41.3% of decisions went to relievers. One hundred years ago, that number was 18%.

A good win statistic clears away a lot of contextual noise. In every game, you have two starting pitchers, on opposing teams, pitching on the same day, at the same ballpark and in the same weather conditions. While starters will never admit they are competing against each other ("my job is to get the opposing lineup out" is the standard refrain), they actually are. Their job is to pitch better than the other pitcher, because doing so means giving up fewer runs than him and, if you do that, you win. Well, at least before the bullpens get involved, but a good win stat would filter out that factor, too.

Take anyone who has ever pitched for the Colorado Rockies. The Rockies have been around for more than 30 years and it's still exceedingly difficult to make heads or tails of their pitchers because so much of their data has to be greatly adjusted for ballpark context. And, while park effects are necessary and sophisticated, they are also estimates.

The Rockies have never had a 20-game winner. The closest was Ubaldo Jimenez, who won 19 in 2010, when he also became one of two Rockies starters to top 7 bWAR. (The other was Kyle Freeland in 2018.) Jimenez is Colorado's career ERA leader as well, with a mark of 3.66. Every other qualifying Colorado starter in franchise history is at 4.05 or above.

Thus, when we talk about the best pitchers of the current era, Rockies pitchers are almost always going to be left out of the conversation. Their numbers just don't seem telling or comparable.

This is where a better win statistic would be so useful. Because whatever the precise effects Coors Field might have on a game's statistics on any given day, a good win stat would be comparing two starters on that field in almost exactly equal conditions. If we do it that way, maybe the Rockies do get some 20-game winners on their ledger.

Is such a win stat possible?

A better way to win

For me, the pitcher win should strictly be the domain of a starting pitcher. This dictum is clouded by the use of openers to start games and bulk pitchers who are used like starters but just not at the outset of games. For now, let's try not to think about that.

The question about each game I want to answer is this: Which starting pitcher was better in that game? The starter who becomes the answer to that question gets the win; the other gets the loss. And that's all. It's as simple as that. Every starter in every game gets a win or a loss and no-decisions don't exist.

Well, the no-decisions would still exist, because I'm not proposing that we erase traditional won-loss records from the books. There's too much history attached. Early Winn is remembered in part for clinging to his career in pursuit of 300 wins, and he finished with that number exactly. Cy Young is remembered for his unbreakable career record of 511 wins. Likewise, Jack Chesbro's claim to immortality is that he owns the modern single-season record of 41 wins. We don't want to erase those things -- we want to add to our understanding of starting pitchers.

Something I've proposed on a number of occasions is to use James' game score method to assign wins and losses. In fact, I've tracked game score records for several years and for this piece, I expanded my database back to 1901 to see how the historical record might look.

There are other game score methods, but I like James' version for its simplicity, though the modified version created by Tom Tango for MLB.com has the same virtue. With either, you can look at a pitching line and easily calculate the game score in your head, once you've got the formula down. (If you can't do that calculation, study more math.)

I also would try to account for short, opener-style outings. I use James' version but dole out a heavy penalty for going fewer than four innings. To avoid ties -- when the starters end up with the same game score -- you can give the W to the starter on the winning team.

Awarding pitcher wins like this isn't perfect. The conditions for the starters aren't truly equal because the quality of the lineups they face won't be the same. When Skenes beat Yoshinobu Yamamoto earlier this season, for example, his task against the Los Angeles Dodgers' lineup was a bit more difficult than Yamamoto's figured to be against Skenes' teammates. Likewise, the quality of the defenses behind opposing starters won't be the same in any given contest.

Despite those disparities, the mandate for both starters is identical: Out-pitch the other guy. And you know what? The game score method of assigning wins and losses to assess the success of that assignment works pretty well.

How game score wins would change history

Let's call a game score win a GSW and a game score loss a GSL. Do you know who owns the single-season record in GSW?

It's Chesbro, still. In fact, his 1904 feat looks just as impressive by this method. Here are the top five seasons by GSW:

Jack Chesbro, 40-11 (1904)
Christy Mathewson, 35-9 (1908)
Iron Joe McGinnity, 34-10 (1904)
Mathewson, 34-12 (1904)
Ed Walsh, 34-15 (1908)

Still all deadball guys, sure, but that's just the top of the leaderboard. There have been 21 30-win seasons by the traditional wins method since 1901 but only three during the last 100 years: Lefty Grove (31 in 1930), Dizzy Dean (30 in 1934) and Denny McLain (31 in 1968).

By the game score method, the list of 30-game winners grows to 36 and it's not so dusty -- 12 of them land in the expansion era (since 1960) and we even get two 30-win seasons during the wild-card era (since 1994). Here are the most recent instances:

33 GSWs: Sandy Koufax (twice, 1965 and 1966) and Mickey Lolich (1971)

32: Steve Carlton (1972, for a last-place team), Denny McLain (1968)

31: Koufax (1963)

30: Whitey Ford (1961), Juan Marichal (1968), Jim Palmer (1975), Ron Guidry (1978), Randy Johnson (twice, 2001 and 2002)

The Big Unit! Johnson won the last two of four consecutive NL Cy Young Awards in 2001 and 2002, during which his combined traditional record was 45-11. His combined game score record is 60-9.

When you go down the list to 29 wins, the roster is just as interesting -- and more recent. Here are the last five instances:

• Dwight Gooden (1985)

• Mike Scott and Roger Clemens (1986)

• Curt Schilling (2001)

• Gerrit Cole (2019)

I mean, are we having fun now, or what? Imagine those seasons and the coverage that would go with their pursuit of 30 wins. Schilling would be trying to match Johnson to give the Arizona Diamondbacks a pair of 30-game winners. And Cole, only a few years ago, would have been racing for 30 wins in his last season for the powerhouse Houston Astros in advance of free agency. Wouldn't you have liked to have had this headline at ESPN to react to that winter?

Yankees sign 29-game winner Cole to $324 million deal

None of this is a product of a fantastical what-if scenario. This is all based on what these pitchers actually did, just framed and measured a little differently. And I think it adds to their accomplishment (or lack thereof in the case of Homer Bailey's 0-20 season in 2018) and improves the conversation about pitching, which now is too bogged down by statistical complexities that many or even most fans roll their eyes at.

Advanced measures would still matter a great deal of course, but barroom conversations about pitching would be much improved. I imagine somehow sitting down for one more baseball chat with my late grandfather, who was one of the people who taught me about the sport. If I told him something like, "Gerrit Cole had 7.8 WAR last year and a 28% strikeout rate," it wouldn't mean anything to him. But if I told him, "Gerrit Cole won 29 games last year," he'd understand that and would not be misled about what it meant.

Thinking about pitcher wins in this way brings the past back into conversation with the present. For all of the differences between what was expected of Christy Matthewson in 1904 and Tarik Skubal in 2025, the core mission outlined by this framework is identical: To outpitch your opponent when you take the mound.

This becomes evident when you look at the list of those who have reached 300 career game score wins since 1901, a roster of greats that covers every period of the modern era ... and is about to grow by one:

Next up, at 299: Clayton Kershaw, who will join Verlander and Scherzer as active 300-game winners, at least by this method. By the traditional method, none of them are likely to reach 300.

What about Skenes?

There's a reason we chose Skenes as our jumping off point. As mentioned, Skenes' 4-6 mark over his first 16 starts tells you nothing about a pitcher with a 1.85 ERA. His game score record (11-5) is a lot more on the mark. Here's Skenes' game score log entering his start Wednesday against Milwaukee Brewers rookie sensation Jacob Misiorowski:

For his career, Skenes is now 30-9 by the game score method. He's 15-9 by the traditional formulation. Same number of losses, but double the wins. Which version is more indicative of Skenes as a pitcher?

It's cherry-picking to home in on Skenes, but his game score log translates to this: Skenes has pitched better than his starting opponent 76.9% of the time as a big leaguer, despite the treachery of the punchless offense behind him.

Now let's do one more list. Here are the three highest game score winning percentages, minimum 30 career starts, since 1901:

1. Paul Skenes, .769 (30-9)

2. Nick Maddox, .722 (52-20)

3. Smoky Joe Wood, .722 (114-44)

Wood is historically prominent, while Maddox, who pitched for the Pirates 115 years ago, is not. Still, since Maddox popped up, I have to share this late-in-life quote from him, because it so typifies the old-timer mindset, "These guys today aren't pitchers -- they're throwers. Why, in my day, I'd throw one so fast past that guy [Ralph] Kiner he'd get pneumonia from the wind."

Skenes is a pitcher and a thrower, a budding all-time great who is in conversation with pitchers who retired decades before he was born. If Skenes stays healthy (knock on wood) and his career builds, we can marvel at his accolades and statistical achievements. But will we ever say, "Skenes has a chance to be a 60 WAR guy" and expect that to resonate?

Maybe someday. But wouldn't it be more fun to track how many 20-win -- or even 30-win -- seasons he can rack up? Wouldn't it be more fun to count down his progress to 300 wins, which he is never going to sniff by traditional wins, unless the game itself changes dramatically?

Wouldn't it be more fun to align pitching's present with pitching's past? Wins have always been the currency of baseball in general, and of pitching in particular. It's just that up until now, pitching wins have been an unstable currency.

But it doesn't have to be that way.