Skip to content

Someone Has to Finish, or Pitching to the Score

February 9, 2011

For me, one of the first signs of a new baseball season is the release of the preview magazines. If for no other reason than I like to track roster changes throughout the season in ink and on paper, The Sporting News always gets my  money, as their paper and roster formatting are more suitable to handwritten notes being added than some of the more glossy mags. This is not to say that I don’t get other guides to the season, like the Baseball Prospectus annual, or Baseball America’s Top Prospect handbook, because I do. But they always arrive at my door 1-3 weeks after the Sporting News rag hits the bookshelves up here in Winterland.

The Sporting News mags, as much as I use them as a resource in motion throughout the season, also has a few annoying quirks that bear mention. One thing I have bitched about to their producer in the past is the way they present the teams by division, in order of their predicted finish. Why does anyone think that is a good idea? Who cares what the TSN staff thought about roster power in January? How does that help me (or anyone else) quickly find any given team in July? If they aren’t going to list them purely alphabetically, or divisional-alphabetical, they should at least order it based on the previous year’s finish. Something to keep the presentation relevant over time.

Another previous peeve of mine was their selection of statistics to show in the team roster sections. In years past, they would list stats from every level a player on a 40-man roster as of press-time had appeared in the previous season. I liked that idea for the prospects on those rosters, as it did make sense to show how a player’s line changed between AA and AAA and the Majors. Of course, that approach sometimes backfired, with TSN including rehab appearances by well-established MLB players. Did we really need to have the stats from Bobby Jenks’ single rehab appearances at both the High-A and AA levels in 2009? What did that add to anyone’s understanding of how Jenks’ season went, or what kind of pitcher he was? So this year, they’ve thrown out that bath-water, along with the baby. In their 2011 rag, TSN has only listed MLB stats. Prospects are simply listed without any stats to their name. A lesser geek than I would not even know whether Alan Farina is a starting pitcher or a bullpen asset.

Before you come to the conclusion that this post is about baseball preview magazines, I will get to the point. One statistic included with the pitcher rosters that has (slightly) annoyed me for a few years is that of the complete game, the CG for those of you who require abbreviations whenever possible. The complete game, a dying art, is not even often in the control of the pitcher, but of his manager, who, earlier and earlier (it seems) goes to the bullpen for sweet and sour relief. As much as we can praise workhorses like Roy Halladay and CC Sabathia for finishing that which they start, even Halladay has only completed 34 of 129 starts over the last four years, just over one out of four starts. Sabathia, for all his sterling reputation as a finisher, has only completed 24 starts in the past five years, across 165 starts (14.5%). And so what? What does it all mean?

And that’s how I start asking questions. Anyone who has watched a baseball game over the past few years can talk about what a drag it is when the manager begins running laps between the dugout and the pitcher’s mound on a regular schedule by the 6th inning. So I decided to track the decline of the complete game over time. Then it hit me that, as fewer complete games means more work for the bullpen, to compare the number of complete games thrown with the occurrence of saves. Yes, I understand that not every complete game would have resulted in a save opportunity had the starter given up the mound (a complete game can even happen – and often does – in a losing cause!), but there is a relationship between the lasting power of the starting pitcher and the use of bullpens, so I also decided to track saves over that same period. I chose to start this study with the 1961 year, which is distinctive as the first year that the AL (preceding the National by one year) played a 162 game schedule, as well as the first of the modern day MLB expansion years. 1961 was also a year with great baseball cards, including my own prized Harmon Killebrew card.

One day I will learn how to create charts in Excel. For now, we’ll stick to the numbers. So here’s what I looked at: I started off with the total number of wins in each season from 1961 through 2010, so I could account for the differences in the number of teams playing, strike season, incomplete schedules and play-in games. I then took the MLB-wide totals for complete games and saves in each season, as well as the league-wide ERA (I’m curious). This allowed me to calculate how many starts were completed as a percentage, as well as how many wins required “saving”.  Then I was able to look at the combined save/complete game percentages, and the ratio of wins saved to starts completed.

Year Total MLB Wins Total MLB Complete Games % of starts completed Total MLB Saves % of games saved League ERA Combined % of games completed or wins saved Ratio of wins saved to games completed
2010 2430 165 3.395 1204 49.547 4.08 52.942 14.594
2009 2430 152 3.128 1202 49.465 4.32 52.593 15.816
2008 2428 136 2.801 1184 48.764 4.32 51.565 17.412
2007 2431 112 2.304 1198 49.280 4.47 51.584 21.393
2006 2429 144 2.964 1201 49.444 4.53 52.408 16.681
2005 2430 189 3.889 1254 51.605 4.29 55.494 13.270
2004 2428 150 3.089 1230 50.659 4.46 53.748 16.400
2003 2429 209 4.302 1198 49.321 4.40 53.623 11.464
2002 2425 214 4.412 1224 50.474 4.28 54.887 11.439
2001 2428 199 4.098 1210 49.835 4.42 53.933 12.161
2000 2428 234 4.819 1178 48.517 4.77 53.336 10.068
1999 2427 237 4.883 1217 50.144 4.71 55.027 10.270
1998 2430 302 6.214 1265 52.058 4.43 58.272 8.377
1997 2266 266 5.869 1139 50.265 4.39 56.134 8.564
1996 2266 290 6.399 1116 49.250 4.61 55.649 7.697
1995 2016 275 6.820 1006 49.901 4.45 56.721 7.316
1994 1599 255 7.974 777 48.593 4.51 56.567 6.094
1993 2268 371 8.179 1192 52.557 4.19 60.736 6.426
1992 2106 419 9.948 1109 52.659 3.75 62.607 5.294
1991 2104 366 8.698 1132 53.802 3.91 62.500 6.186
1990 2105 429 10.190 1113 52.874 3.86 63.064 5.189
1989 2103 483 11.484 1069 50.832 3.71 62.316 4.427
1988 2098 622 14.824 1049 50.000 3.73 64.824 3.373
1987 2105 561 13.325 971 46.128 4.29 59.454 3.462
1986 2102 579 13.773 1004 47.764 3.97 61.537 3.468
1985 2101 627 14.921 977 46.502 3.89 61.423 3.116
1984 2104 632 15.019 993 47.196 3.81 62.215 3.142
1983 2106 745 17.688 977 46.391 3.87 64.079 2.623
1982 2106 734 17.426 932 44.255 3.86 61.681 2.540
1981 1389 510 18.359 605 43.557 3.58 61.915 2.373
1980 2101 856 20.371 902 42.932 3.84 63.303 2.107
1979 2096 913 21.780 840 40.076 4.00 61.856 1.840
1978 2102 1034 24.596 804 38.249 3.69 62.845 1.555
1977 2103 907 21.564 845 40.181 4.00 61.745 1.863
1976 1939 1039 26.792 683 35.224 3.51 62.017 1.315
1975 1933 1052 27.212 669 34.609 3.71 61.821 1.272
1974 1940 1089 28.067 517 26.649 3.63 54.716 0.949
1973 1942 1061 27.317 819 42.173 3.75 69.490 1.544
1972 1858 1009 27.153 733 39.451 3.26 66.604 1.453
1971 1936 1083 27.970 689 35.589 3.47 63.559 1.272
1970 1943 852 21.925 878 45.188 3.89 67.113 2.061
1969 1943 982 25.270 745 38.343 3.61 63.613 1.517
1968 1619 897 27.702 597 36.875 2.98 64.577 1.331
1967 1617 782 24.181 647 40.012 3.30 64.193 1.655
1966 1613 736 22.815 667 41.352 3.52 64.166 1.813
1965 1619 739 22.823 678 41.878 3.50 64.700 1.835
1964 1620 797 24.599 668 41.235 3.58 65.833 1.676
1963 1618 865 26.731 589 36.403 3.46 63.133 1.362
1962 1618 844 26.082 618 38.195 3.96 64.277 1.464
1961 1423 745 26.177 501 35.207 4.03 61.384 1.345

And now for the analysis:

  • Slowly, we may be seeing a mild resurgence in complete games, as they are up by a full percentage point of games in the last four years. The number of complete games bottomed out in 2007 with a mere 112 out of 4862 games started – 2.304%. The percentage of games completed has steadily climbed back since then, with 2.801% in 2008, followed by 3.128% in 2009 and last year’s 3.395%. In the grand scheme of things, though, the last seven years (since 2004) have seen the lowest percentage of game’s completed in history – I don’t need to look up the years before 1961 to know that.
  • Looking back at my two earlier examples, Sabathia and Halladay. CC Sabathia has completed over 14% of his starts since 2006, more than 4 times the league-wide average in that span. Yet prior to 1985 (little Carsten was only 5 years old!), that figure would have been below league-average. Doc Halladay’s four-year run of completing 26.5% of his starts is amazing in the context of today’s game, with only 2.9% of all starts being finished in that time frame. Halladay alone has 6% of all complete games across the majors in the last four years, more than 9-times the MLB-wide average. And even that would have been below the historical MLB average as recently as 1976, when Leroy was still but a twinkle in his pappy’s eye. I miss watching him pitch regularly…
  • I will not be the first (nor last) person to equate the death of the complete game to the rise of the multi-headed Cerberus bullpen. But I did wonder how closely we can compare them. The save has been an official statistic since only 1969, but others have gone back in time to give us more historical statistics, much as we now know how many RBI’s Cap Anson had for the White Stockings in 1876 even though the statistic had not yet been invented. Without reprinting the rules for how saves are earned, my point can be made by stating that the guy who started the game cannot also save the game. In that light, I would have to expect that the percentage of saves would have been much lower before complete games began disappearing. If that was true, the combined percentage of complete games plus saved games would be close to equal over time. Before the save was made official in 1969, the percentage of wins requiring saving hovered between 35.207-41.878% since 1961. Other than an inexplicable dip to 26.649% in 1974, 39% of games in the 70’s were saved. 46.7% of games in the 80’s needed saving, as complete games also dipped below 20%. Things became much more interesting in the 90’s, as the rise in games needing saving tapered off to 51.3% while complete games continued to dwindle, down to less than 15% of all starts in the decade. So by 1999, the combined percentage of games either completed or saved (remember – they can’t be both) had dropped to just over 55%, down from a high of 69.5% in 1973. Since the run of the century, the percentage of games needing saving has actually dropped, last breaching 50% in 2005. And, as we have already discussed, complete games have only made a very mild comeback.
  • 1994. Yes, it was a season shortened by strike, but it also marks a fairly large reduction in the combined percentage of games either completed or saved, as well as a concurrent jump in overall MLB-wide ERA. In most years, ERA fluctuated by 0.2 or so. In 1994, MLB ERA was up to 4.51 (much higher than any previous year in the study), from 4.19 in ’93. The combined percentage of CG’s and Saves, which had tended to fluctuate within a 2% band between seasons, now dropped over 4%, from 60.736% to 56.567%. Most of this was in a huge drop in games needing a closer, while also seeing a slight drop in complete games. This tells me that, in this case, more middle relievers were pitching, and pitching poorly. I would need a much larger study of game logs to see if poor middle (middling) relievers were the cause of the games not needing saving, but it certainly passes the sniff test, doesn’t it?

I’m not yet sure what it all means, but it certainly seems like a sign that most teams are significantly overpaying relievers if fewer games are close enough in the scoreline to need a closer. I just read Moneyball again, so maybe I should blame Billy Beane – after all, he did try to build a roster that knew how to take more pitches and make the opposing starter work harder. Then again, league-wide on base percentage in 1980 was .326. In 1990, that dropped all the way to .325. In 2000, we were up to .345. But last year, MLB-wide OBP was back down to .325. Was 2000 a fluke? Maybe; 2005 was back down to .330, 1998 was .335. Maybe Beane was onto something but now the pitchers are fighting back with more strikes? I could continue to ask questions and look at broad swathes of numbers to find plausible answers, but at some point (now, maybe?), I should post something. Conditioning more pitchers to stretch it out, and not being so quick on the bullpen trigger may be warranted, if seeing fewer innings being eaten, as opposed to pitched, is a desirable goal. I think it is. If fewer games are within 3 runs going into the later innings, then it may not be such a bad an idea to give the starting pitcher that elusive 7th inning more often.

7 Comments leave one →
  1. tchwrtrextraordinaire permalink
    February 9, 2011 6:17 pm

    ugh, you billy beane fans. there are a slew of other factors that need to be considered. Not the least of which is the impact of the steriod era (which can be argued goes as far back as the late 80s) and the effects of expansion. More importantly, you have to consider the effects of our “win-now” sports culture, which I think, places enormous pressure on how games are managed. Patience is not a virtue anymore. This includes pressures placed by management to not over-extend a pitcher, as the insurance on some of these over-priced yahoos probably equals, or surpasses, a lesser player’s salary. You stat boys have to look beyond the numbers.

    • February 9, 2011 6:31 pm

      Thanks for commenting. Note that I never claimed to be a Beane fan. In fact, I went so far as to blame him, at least in part, for the rise in mediocre relief pitching. A lot of the use of relievers is to gain the platoon advantage, especially as used by managers like LaRussa. Of course, when the roster has 12 pitchers, you only have 13 position players. 8-9 are starting (depending on the league) so the bench is 4-5 deep, compared to 7 relief. The pitchers would then naturally have the upper hand, platoon-wise. The hitting team will run out of replacements before the pitching team runs out of relief pitchers. I would suggest limiting pitching staffs to 11 men, 5 starters and 6 relievers. The bench would then have 6 subs and in-game strategy would be less predictable. Furthermore, this would allow for the weeding out of mediocre middle relievers, for more robust, durable pitchers.

      • tchwrtrextraordinaire permalink
        February 9, 2011 6:48 pm

        Sorry for the beane remark. What I’m particularly referring to is managing to save or keep your job, which is something that can’t be quantified I think. I’m of the opinion that many managers don’t just manage to win. And I think managers, and team management, have become leary of over-extending pitchers ever since Dusty Baker’s handling of Mark Prior and Kerry Wood in Chicago. In the last 20 years, I think the role of managers in baseball has evolved beyond in-game strategy to include factors that specifically determine how long a pitcher remains in the game. It’s funny how the decline of complete games coincides with the growth of free agency since the late 70s.

      • February 9, 2011 6:58 pm

        It’s a very good point. I don’t know if I could quantify that aspect of managing either, although I certainly believe in its accuracy. There definitely is an aspect of regular season managing that requires the head honcho to plan out for the full 162, which necessitates an amount of arm coddling. I still thin it is possible to replace three situational relievers with very limited skill sets with 2 young starters in waiting who can pitch 60-75 games a year, 1-2 innings per appearance. In a given season, you get 1458 innings (9*162 – so really it’s less, once you take away the 9th from those road losses). If your starting pitchers average 185 IP per, that leaves 533 IP for the bullpen. Right now, that is split up 7 ways (76IP per). Going to a 6-man bullpen, would average 89IP per. When you consider roster turnover, injury replacements, and around 40 less road-9ths, I think getting the better relievers to pitch an additional 8-10 innings a year isn’t too much to ask.

  2. February 11, 2011 5:03 pm

    I miss Halladay, too.

  3. Shazbot permalink
    February 12, 2011 8:38 pm

    Err.. a start can be completed, and a save made in the same game. Just for opposing teams.

    It’s a combination of CG(L) and S, so yes, a game can have both.

    The effect noticed is a combination of fewer non-save situations being complete games, and fewer CG(L)’s being pitched, as you need a CG(L) to have two games pitched, and two (saves + CG) recorded, in the same game.

    • February 12, 2011 9:30 pm

      Thanks for reading and commenting. I know (and pointed out) that complete games can end occur in losing causes, but I don’t know how often that has happened over the years and am not sure how to find out when and where those occurred. I took the position that those cases are likely occurring as often as complete games as a whole, and so, over time, the ratio of complete game losses to complete games in total has remained roughly the same. If I’m wrong, I don’t think it changes the jist of my argument, however. If you think it does make a difference, I’d love to hear how.

Leave a Reply

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

You are commenting using your 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: