1993-2007 was the Great Pitcher Era

Most baseball fans associate the 1990s and early 2000s with record-breaking, steroids-induced offensive production. After all, Roger Maris’ long standing home run record was broken six separate times from 1998-2001. The 15 year period I’m going to discuss here, 1993-2007, has the second highest run per game average (1994-2008 is the highest) of any 15 year period outside the “live-ball era.”

Therefore, you may be surprised to learn that 1993-2007 featured seven of the 30 greatest pitchers of all time as measured by career WAR.1 These seven pitchers didn’t just pitch at some point during that period, but each of their careers spanned the entire period.  The pitchers, career WAR, rank, and playing period follow:

Name Career Pitching WAR Rank among all Pitchers Seasons played
Roger Clemens 139.4 3 1984-2007
Greg Maddux 104.6 8 1986-2008
Randy Johnson 104.1 9 1988-2009
Pedro Martinez 86.0 17 1992-2009
Mike Mussina 82.7 24 1991-2008
Curt Schilling 80.7 26 1988-2007
Tom Glavine 74.0 28 1987-2008

A careful reader of this table may notice that all of these pitchers also pitched in 1992. However, this was Pedro Martinez’s first season and he only pitched eight innings so it seems a little cheap to include it.

Seven other of the top 30 pitchers pitched during the 17 season from 1967-1983: Tom Seaver, Phil Niekro, Gaylord Perry, Steve Carlton, Nolan Ryan, Fergie Jenkins, and Don Sutton.  Two others, Bert Blyleven (career started in 1970) and Bob Gibson (career ended in 1975) pitched for good chunks of this same period.  The only overlap with 1993-2007 in the top 30 that is not included in the table above is Nolan Ryan’s 13 games in 1993. No other 15 year period in baseball history (that doesn’t overlap with one of these two) has more than 2 pitchers on the top 30 list. Therefore, it is clear, that one of these two periods is THE GREAT PITCHER ERA™.

Since 1967-1983 is longer than 1993-2007 and it partially includes 9 great pitchers, at first blush it seems like it deserves the crown.  However, two factors favor 1993-2007: its great pitchers were greater overall, and they produced much more dominant seasons.

Simply adding up the WAR of the seven 1993-2007 pitchers and the seven 1967-1983 pitchers, yields 50 extra wins for the more recent group. In addition, the 1993 group was much better on a per game basis. The average of their waaWL%2 is .600 vs.  only .558 for the 1967 group. In fact, the average of the 1993 group is exactly equal to the best member of the 1967 group (Tom Seaver).

110409_pedromartinezThe 1993-1997 period was also full of many extremely dominant single-season pitching performances. 4 players in the 1993 group produced 17 seasons with waaWL% above .700, while only Tom Seaver (who had 2) produced any in the 1967 group:

Tom Seaver 1971, 1973, 1977 (excluded b/c 165 IP)
Roger Clemens 1986, 1990, 1992, 1997
Greg Maddux 1992, 1994, 1995
Randy Johnson 1995, 1997, 1998 (excluded b/c  84 IP), 1999, 2001, 2002
Pedro Martinez 1997, 1999, 2000 (my pick for greatest pitching season ever), 2001, 2003

The overall quality and single-season dominance of the 1993 group outweighs the length and the two (partially) extra pitchers in the 1967 group. I therefore, anoint 1993-2007 the THE GREAT PITCHER ERA™.

1 WAR — Wins Above Replacement for Pitchers. A single number that presents the number of wins the player added to the team above what a replacement player (think AAA or AAAA) would add. This value includes defensive support and includes additional value for high leverage situations. Scale: 8+ MVP Quality, 5+ All-Star Quality, 2+ Starter, 0-2 Reserve, < 0 Replacement Level. Developed by Sean Smith of BaseballProjection.com
2 waaWL% ▾ — Win-Loss% w/ Avg. Team. This is the win-loss of an otherwise average team in ONLY the games this player played in. For example, for a pitcher this would include only the games the pitcher threw in and ignoring games they did not play in.

Why the Pirates will (probably) win their 82nd game on the road

After defeating the Giants on the road last night, the Pirates are 75-52 and only 7 wins away from ending their 19-year “futility” streak. I’m sure there are some Pirates fans who still think that the Pirates will somehow continue their streak and fail to win just 20% of their remaining 35 games. That said, the more optimistic/realistic fans are wondering when the fateful event will finally happen.

To find out, I’ve again looked at the historical record of the 200 teams in the modern history of baseball with the most similar expected winning percentages at the same point of the season.  In particular, I looked at how many games it took for the “similar” teams to win 7 games.

The 2001 Boston Red Sox and the 1999 Philadelphia Phillies were the worst in the sample and took 31 games to win 7. If this years Pirates perform similarly I expect lots of extra business for Pittsburgh-area cardiologists. On the flip side, the 2007 Indians, 1982 Orioles, and 2002 Athletics each took only 8 games. The 25-percentile of the sample is 12 games, and the 75% is 17 games.

Unfortunately, the Pirates have a 9 game road trip from September 2 – 11, which are games 10-18. In the sample, 160 teams, or 80% won their 7th game during that period. That’s too bad for long-suffering Pirates fans who deserve to see their suffering end in person. On the other hand, a Pirates fan that road-trips to St. Louis for the September 6-8 series, has more than a 30% chance of watching the futility-end during that playoff-crucial series.

Pirates at the halfway mark: How many games will they win?

The Pirates swept their weekend series with the Brewers leaving them 51-30 at the season’s halfway point. I performed two analyses similar to those I posted on Friday. I looked at the 200 teams in baseball history with the most similar first-half-seasons. In the first analysis (herafter referred to as the actual winning percentage, or AWP, analysis), similarity was determined by the absolute value of the difference between current Pirates winning percentage (.630) and the other teams’ winning percentage. The second analysis (EWP) was the same except expected winning percentages were substituted for actual winning percentages. In both analyses I looked at the number of wins the Pirates would have if they finished the season with the same winning percentage as the “similar” team.

In the EWP analysis, two teams finished below .500, four finished at exactly .500, 102 finished with 95 or more wins (Pirates manager Clint Hurdle’s goal for the season), and 16 teams finished with 105 wins or more (105 losses to 105 wins in three seasons). This analysis suggests that it is 8x as likely that the Pirates finish with 105 wins than with another losing season.

The Pirates actual winning percentage (.630) is much better than their expected winning percentage (.569), so the AWP analysis unsurprisingly looks even better for the Pirates. In the AWP analysis, only one team finished below .500, two at exactly .500, 128 with at least 95 wins, 23 with at least 105 wins, and even one team with more than 110 wins. As I said on Friday, I strongly prefer the EWP analysis, so read the AWP results with caution.

For those who would like to explore the data a little further, I’ve made a bar chart with the results of both analyses. If you click on the chart on this site it will take you to an interactive version on my CMU website. The chart is here:

How many games will the Pirates Win?

Is the Pirates 20 season losing streak over?

The Pittsburgh Pirates have finished the last 20 seasons with more losses than wins. This is the longest such streak in any of the four major American professional sports leagues (NHL, NBA, MLB, NFL). The streak is likely to end this season, however, because the Pirates are tied for the best record in baseball. As it stands now, the Pirates are 48-30 which earns them a .615 winning percentage through just shy of half the season. I wondered whether any team with such a good first half finished with a losing record. Unfortunately some quick googling didn’t answer my question, so I resorted to querying the retrosheet gamelogs data I have stored on my computer1.

It turns out that there are four teams in the history of baseball2 with .600 or better records at the midpoint of the season3 who finished without a winning record. These four teams are:

Year Team Mid W Mid L Mid W% Full W Full L Full W%
1890 Philadelphia Athletics 41 22 .651 54 78 .409
1889 Cleveland Spiders 40 24 .625 61 72 .459
1977 Chicago Cubs 48 29 .623 81 81 .500
1905 Cleveland Indians 46 28 .622 76 78 .494

As you can see, two of the teams, the Philadelphia Athletics and Cleveland Spiders are truly ancient. They both experienced horrendous second halves, ending well south of .500. I doubt those teams resemble modern teams so I’m not sure there is much of a lesson to draw from them. One of the remaining teams, the 1977 Chicago Cubs, finished at exactly .500 which would technically end the Pirates losing streak, but would hardly feel satisfying. All four teams are drawn from the 366 teams with .600 records at the midpoint. Therefore, almost 99% of .600 teams finish with winning records. This analysis suggests that Pirates have an extremely strong chance of ending their streak.

Unfortunately, the Pirates are worse by many measures than the typical .600 team. In particular, the winning percentage one would predict from the runs they’ve scored and allowed is .557 not .6154. For math-phobic Pirates fans, let me give you one anecdote to try to convince you of the importance of expected winning percentage. The Pirates have suffered late-season collapses in each of the past two seasons and their pre-collapse expected records were much worse than their actual records in both seasons. This thought-process led me to a second analysis which looked at the second half records of teams that were similar to the Pirates at mid-season.

There are 262 teams with mid-season expected winning percentages within .015 of the Pirates’. Here’s a histogram of their second-half performances:
The distribution roughly matched my expectations. The vast majority of teams won between 40% and 70% of their second half games. The Pirates need to win at least 34 more games to finish with a winning record. This requires a .404 winning percentage for the rest of the season. 10 teams finished with records at least that poor, which suggests the Pirates still have an almost 4% chance of extending their losing streak. I suspect that percentage is too high for Pirates fans to feel comfortable yet. Sorry to disappoint you!

Let’s finish on a much happier note. According to the same analysis, the Pirates have a better chance of winning 105 games, virtually guaranteeing a playoff appearance and very likely a division win, than extending their losing streak. The 12 best teams in my sample6 completed their seasons with .675 or better records, which would yield 105 wins for the Pirates. In 2010, the Pirates lost exactly 105 games to extend the losing streak to 18 seasons. From 105 losses to 105 wins in 3 seasons — THAT would be some story. Certainly a happier one than a 21 season losing streak.

1. Yes I am that kind of dork.
2. I excluded any season with fewer than 100 games.
3. 48.1481481% of their season just like the Pirates so far.
4. See Pythagorean expectation on Wikipedia for more information. I used the 1.83 single number exponent for its mix of simplicity of implementation and accuracy.
5. I excluded seasons with fewer than 140 games this time, partly to exclude the ancient teams.
6. I would have chosen the top 10 teams again for symmetry with the losing-season bottom 10, but four teams were tied at exactly .675.