The Greatest Pitcher Alive

Monday, ESPN published a much-publicized feature ranking the greatest living pitchers. The list was based on a poll of 32 ESPN baseball “experts,” who all agreed that Roger Clemens was the greatest living pitcher, followed by Tom Seaver, Sandy Koufax, Bob Gibson, Greg Maddux, Bob Feller, Randy Johnson, Pedro Martinez, Steve Carlton, and Juan Marichal to round out the top-ten.

First off, the whole idea of the greatest living anything is silly. What does it matter? If there were a nuclear war, and I was the only person remaining, would I suddenly qualify as the greatest living pitcher/hitter/physicist? It’s ridiculous. While I enjoy greatest of all-time debates, and understand great active player debates, the “greatest living” label seems pretty meaningless.

According to great posters on Baseball Primer, the whole obsession with the greatest living player started in 1969, when Joe DiMaggio was voted the greatest living baseball player by a group of sportswriters. DiMaggio became obsessed with being called the greatest living player, and demanded to be introduced as such from then on. Of course, he knew that he couldn’t be the greatest ever—that honor was reserved for Ruth.

Having established that, I decided that it would be fun to look into this question anyways. I’ve always had an obsession with pitching, and for years, I’ve worked on a way to rank pitchers objectively. Frankly, neither Bill James’ Win Shares nor Clay Davenport’s WARP do it for me, though I have borrowed ideas from both. The key to me, with both systems, is that they attempt to rate players based on the statistic that governs all of baseball, and on whose importance both stat-heads and normal baseball fans can agree on: wins.

Both try to take away context, at least in large part, from a pitcher’s statistics, and spit out his theoretical win value. About a year ago, I realized that I could approach the question from a different angle, in my opinion, a better angle. In June, I published an article on my blog detailing my Pitching Runs Created system. This January, I ran an article about PRC on THT. A week later, I wrote a follow-up, at the end of which, I threw in a paragraph about a system that could be derived from my PRC method: Pitching Wins (and Losses).

Here’s the basic idea, both with PRC and Pitching Wins. I take a player’s RA (it’s the same thing as ERA, but counting unearned runs), and find his expected winning percentage based on the run environment of the league. So a pitcher who allows 3 runs a game in a league where the average team scores 5 is expected to win 71.7% of the time. That’s his expected winning percentage.

Now this is where the two systems diverge. With PRC, I first try to find how many runs an offense would have to score per game with an average pitching staff to match that winning percentage. With pW, the process is more straightforward. First, I multiply the players expected winning percentage by his innings pitched and divide by 9. That tells us how many wins he would be expected to have. Then, I multiply his that number by the percentage of credit I believe he should be assigned, based on his strikeout totals. Basically, the more strike outs, the less dependent the pitcher is on his fielders, and the more credit he gets. Finally, I make an era adjustment to correct for changes in the quality of competition across the years. I derive Pitching Losses in the same way.

I think the system makes intuitive sense. There is minimum manipulation of statistics: it looks at what pitchers actually did on the field. While the lack of adjustments might make the yearly ratings more “fluky,” it also means that career numbers aren’t “over-corrected.” The pitcher/defense split is simple, and thus I’m more confident in using it then a more complicated system with all kinds of conflicting variables. And the era adjustment makes sense, I think.

That’s all I want to say. Obviously, this system, like any other, has its weaknesses. Luckily, those weaknesses appear mostly in seasonal ratings, but wash out over long careers. Thus, I feel pretty confident to rank pitchers using pW/L. I’ve done the numbers for every year in baseball, and will be writing stuff in the coming months using those numbers, which includes an article in the 2007 Hardball Times Annual (yes, we’ve started planning already).

So I took ESPN’s top-ten living pitchers and examined their career pW/L records. While ESPN’s top choice—Clemens—was right by any measure, the other players fall into different orders. If we order them by wins, the list looks as follows:

Last	   First       Win%	pWins	pLoss				
Clemens	   Roger       0.654	270	143				
Carlton	   Steve       0.568	249	189				
Seaver	   Tom         0.621	246	151				
Maddux	   Greg        0.634	227	131				
Johnson	   Randy       0.645	214	117				
Gibson	   Bob         0.606	211	138				
Feller	   Bob         0.599	185	124				
Marichal    Juan        0.579	163	118				
Martinez    Pedro       0.708	161	66				
Koufax	   Sandy       0.632	134	79	

Clemens is far-and-away ahead of the rest, with 270 pWins. That may seem like a low total, and the reason for that is the division of credit between the pitcher and his defense. The historical average is that about 2/3 of the credit goes to a pitcher, so if you want pWins to look like real win totals, just multiply by 1.5. That would make Clemens’ career total 405, and the top-six on this list the equivalent of 300-game winners. Essentially, in the pWins system, 200 wins is the equivalent of 300 wins in baseball.

Looking at the above-leader board, there is a clear problem with just looking at wins: the absence of losses. Sure, Carlton’s 249 pWins are impressive, but how much are they negated by his pLosses? The easiest way to find out is find a pitcher’s Wins Above Average:

Last	   First       Win%	pWins	pLoss	PRC	pWAA		
Clemens	   Roger       0.654	270	143	2665	64		
Johnson	   Randy       0.645	214	117	2128	48		
Maddux	   Greg        0.634	227	131	2271	48		
Seaver	   Tom         0.621	246	151	2406	48		
Martinez    Pedro       0.708	161	66	1671	47		
Gibson	   Bob         0.606	211	138	2058	37		
Feller	   Bob         0.599	185	124	1776	30		
Carlton	   Steve       0.568	249	189	2381	30		
Koufax	   Sandy       0.632	134	79	1332	27		
Marichal    Juan        0.579	163	118	1559	22	

As expected, Carlton falls back quite a bit, from second to seventh among ESPN’s top-ten pitchers, which is more in-line with where the “experts” ranked him. There’s also almost no difference between Johnson, Seaver, Maddux, and Martinez. Seaver is likely to fall back on this list in the coming years, assuming those pitchers can stay above average. But some might argue that a list ranking pitchers versus average isn’t the best way to do it either.

Versus average metrics underrate guys that eat a lot of innings, but not necessarily with stellar performances. Robin Roberts is a perfect example of that. So what if we instead looked at these guys’ pWins Above Replacement? I did just that, using .450 as a replacement level pitcher. Essentially, if you have a .450 pitcher and .450 offense, you’ll have a 100 loss team, which is as reasonable replacement level as any. I don’t know if this is the best way to do it, but I’m relatively happy with the results.

Last	   First       Win%	pWins	pLoss	pPRC	pWAA	pWAR
Clemens	   Roger       0.654	270	143	2665	64	83
Seaver	   Tom         0.621	246	151	2406	48	66
Maddux	   Greg        0.634	227	131	2271	48	65
Johnson	   Randy       0.645	214	117	2128	48	64
Martinez    Pedro       0.708	161	66	1671	47	58
Gibson	   Bob         0.606	211	138	2058	37	53
Carlton	   Steve       0.568	249	189	2381	30	51
Feller	   Bob         0.599	185	124	1776	30	45
Koufax	   Sandy       0.632	134	79	1332	27	37
Marichal    Juan        0.579	163	118	1559	22	36

Here, Pedro drops far back of the Johnson/Maddux/Seaver group, while Clemens continues to show that ESPN was right—he really does stand head-and-shoulders above his contemporaries. And if you buy the 2007 Annual, you’ll get to see how he compares to all pitchers throughout history as well.

One thing I noticed is how much we underrate Johnson’s greatness. For whatever reason, he never gets mentioned among the greatest of all-time, even though he’s been as valuable as any modern pitcher except for Clemens. It’s probably due to how late he finally attained that greatness, but if you think about it, Johnson really would be a prime candidate to be overrated, not underrated. He’s dominating, strikes out a lot guys, and scares the bejeezus out of you on the mound. The other modern pitcher like that, Nolan Ryan, is among the most overrated players ever. Yet Johnson doesn’t get enough credit. Weird.

On the other hand, you have Koufax, who was third in ESPN’s rankings, and just barely behind Seaver. Yet no objective measure would rate him as better than anyone on this list except for Marichal. He wasn’t as dominant as Pedro, nor did he last as long, so why does Koufax rate ahead of him? It had to be the early retirement. It created an aura, a legacy that makes people overrate Koufax.

But what’s clear is that Clemens, despite being a despicable, backstabbing, lying bastard, is indeed the greatest living pitcher. Whatever that means.