The Death of 20-Win Seasons?

Each February I visit a local grocery or book store, look over the baseball preview magazines, and pick out two or three to buy. Since 1980 (at the age of seven) I have always gotten the classic Street and Smith's Baseball Annual. The one or two others that I pick up, which vary from year to year, are usually entirely fantasy/roto baseball focused, to help prepare me for my upcoming drafts.

S&S still provides great team summaries (leaving aside the problem that they, like most magazines in February, are not up to date with the most recent off-season moves). Their advice for fantasy and roto leagues, now 14 pages in length, includes some strange opinions sometimes, with their rankings probably not being the most reliable (or even close to it).

But this year's issue had an odd two-page article called "Death of the 20-game Winner?" (sorry, not available online). Author Scot Gregor begins by noting that last year, for the first time in baseball history (excluding the strike-shortened seasons of 1981, 1994, and 1995), there were no pitchers who won 20 games or more. Long ago 20+ wins was not that impressive -- Cy Young did it a record 16 times, and both Christy Mathewson and Warren Spahn did it 13 times each. And as recent as 1971 the Orioles managed four starters with 20+ wins on the same team!

But last year the AL's best were Johan Santana and Chien-Ming Wang with 19 wins each, and in the NL six players tied for the lead... with only 16 wins each! So that is what prompted Mr. Gregor to ask his title question.

Many reasons for the decline in win totals are given: luck is needed, good health is needed, consistent run-support is needed, durability to go deep in games is needed. And since the 1970s two relevant, major changes have occurred. First, four-man rotations turned into five-man rotations, so starters are starting less games per season. And of course the nature and use of bullpens has changed in the past several decades, with not only closers being a specialist position, but more and more pitchers being groomed (and paid handsomely) to be lefty or righty setup men. Add to that the high salaries that premium starters get, and there is a natural desire to not wear them out with too many innings of work - get them out of there, and let someone fresh come in and finish the game (or a few such specialists). Winning the game is what matters, and winning a lot of games is what really matters -- so we need our best starters to be injury-free for the whole season. That is no doubt the thinking of many managers and pitching coaches these days, and if it means starting pitchers don't rack up as many W's, so be it.

The article mentions "offensive firepower" being dominant since the late 1990s, but I don't see that as being a factor in this question. But leaving that aside, the reasons listed above would lead one to assume that 20+ Win season should be becoming more and more rare. But vanish entirely? That was the article's headline, and I found that a bizarre speculation -- at least at this time. Yes, 2006 was a first. But to defend the thesis raised in the article's title question, one needs to consider trends -- not just list all the reasons as above, but actually look at number trends (since we are talking about a number afterall, namely 20). The article didn't even begin to do that, so I thought it would be a fun project for me to tackle.

It just so happens that this little 2-hour project afforded me the chance to do two other things I've been meaning to get around to. First, I download the latest version (5.4) of the Lahman baseball statistics database, available in various formats from baseball1.com. And to create nice charts of the data, I was able to use my newly purchased Excel 2007, from the Office 2007 Pro Suite I upgraded to recently. So what I did was create several SQL queries in Access to pull together the data that would show the count of various metrics for the years being considered. I decided to start with 1970, which is 37 years ago and I figure enough data to sensibly consider trends on this question.

So below are links for the four charts I've created. In each chart, you'll notice that I've added vertical blue lines for the three years where two teams were added to MLB: 1977, 1993, 1998. I did this because this meant there were more players, including more pitchers, and since these are raw numbers and counts, that could mean more pitchers reaching the totals tracked (in the first two charts). While expansion teams aren't likely to have great starting pitchers during their first few years, after a while, there are just more people in this role, and more total wins to be had in MLB, so I would think there could be more people getting the totals under consideration. (Think about it this way -- if there were 1,000 teams in MLB, rather than 30, you'd expect more 20+ winners each year, right?)

Also, the data omits 1981, 1994, and 1995, since as noted earlier these were strike-shortened seasons.

1. Count of 18+ and 20+ Winners since 1970 - Clearly there has been a decrease in the number of 20+ and 18+ game winners during this span. (I decided to track both because given the factors described earlier, including things completely outside the control of a pitcher, I figure there are a few games a year where the pitcher could have easily won if only some oddball factor had gone the other way.)
2. Count of 30+ and 35+ GS; 5+ CG since 1970 - Here we see quite clearly the effect of some of the factors described earlier. Complete Game totals have declined dramatically since the early 70s. Pitchers are being pulled for pitchcounts and in order to use specialists in the bullpen. The red line shows the impact that the shift from 4-man to 5-man rotations has had. Very few starters get 35 games started each year now. And Then the blue line show that many are still getting 30 games started, and I assume the slight increase here is due both to the 5-man rotation shift (spreading the games started amongst more guys) and also the influx of the six expansion teams during this era (more starting pitchers pitching).
3. Leader in IPouts (IP) since 1970 - This one shows the MLB leader in "IPOuts" -- if you divide by three, you'll get the more common statistic, Innings Pitched (IP). This number shows a dramatic drop off from the early 1970s through the mid-1980s, and since then has declined further, but only slightly. This reflects both the shift from 4-man to 5-man rotations (I think that makes up the dramatic early shift) and also the increasingly specialized nature of bullpens and the concern with pitchcounts (I think that makes up the continuing gradual decline).
4. Leader in W, GS, CG since 1970 - This one show the MLB leader in Wins, Games Started, adn Complete Games. Games Started has definitely declined during this span, again mostly because of the 4-man to 5-man rotation change I assume. Since 1988 the leader in GS each year has been pretty stable. The annual leader in Complete Games, on the other hand, keeps getting lower and lower. In the early 1970s the leader would usually get near 30 CG (!), today the best gets 6-10. This is the pitchcount and bullpen specialization effect in all its glory. And then we see Win leaders, which have also declined overall since 1970, though not all that dramatically. For instance, Bob Welch nabbed 27 victories in 1990, which equals the 27 Steve Carlton got in 1972. But overall, there has been a slight decline here -- but not so much, in my opinion, to warrant the alarmist notion the S&S article's headline raises.

In writing this blog entry I fought the urge to invoke Mark Twain as an opening. So instead I'll do so in closing. Death of the 20-game winner? Such reports, the numbers show, are exaggerated.

Lastly, on a related issue, see my blog posting last June: Tom Glavine: The Last to Win 300? As an update to that piece, Glavine is at 290 and Randy Johnson is at 280, so I'd now say Glavine is a near-lock, and Johnson has a good shot too. But beyond that, I still have to wonder if anyone will ever reach that milestone again (see that blog posting for others I considered).